# Toolcloak article: full text of every article

> Plain-English guides to free browser-based tools, with real numbers. Each article is also available on its own as Markdown at /articles/<slug>.md. Site summary: https://toolcloak.pages.dev/llms.txt

# How to calculate your exact age in years, months and days, and why leap years and cut-off dates matter

> Age sounds simple until a form asks for it as on a particular date. This guide shows how exact age is worked out, how the days are borrowed from the previous month, how 29 February birthdays behave, and how to check eligibility on a cut-off date.

Published 2026-10-09 by Toolcloak. 4 minute read. Original: https://toolcloak.pages.dev/articles/how-to-calculate-your-exact-age-in-years-months-and-days-and-why-leap-years-and-cut-off-dates-matter
Related free tool: [Age Calculator](https://toolcloak.pages.dev/tools/age-calculator)

Your **age** is the time that has passed since your date of birth, normally written in completed years. Forms for exams, jobs and insurance often want more: years, months and days **as on a given date**. Calendar months have different lengths, so the result is not just a subtraction of numbers, and a small mistake can decide whether you meet an age limit.

> **In short**
> - Someone born on **1 January 2000** is **26 years, 9 months and 8 days** old on **9 October 2026**.
> - That is **9,778 days**, **1,396 full weeks** and **234,672 hours**, and the next birthday is **84 days** away.
> - For an age limit, calculate age **as on the cut-off date** in the notice, not as on today's date.
> - Someone born on **29 February** has a birthday on 1 March in this tool in non-leap years.

## How exact age is worked out

The method is the same one you would use on paper. Count the full years, then the full months left over, then the remaining days. When the day of the month in the target date is smaller than the birth day, you cannot count the last month in full, so you count one month fewer and take the days from the last complete month.

**Example: born 1 January 2000, age on 9 October 2026**

1. **Count the full years.** 1 Jan 2000 to 1 Jan 2026 is **26 years**.
2. **Count the full months.** 1 Jan 2026 to 1 Oct 2026 is **9 months**.
3. **Count the leftover days.** 1 Oct to 9 Oct is **8 days**.
4. **Write the result.** **26 years, 9 months, 8 days**.

*The tool shows more than that: total days, total weeks, total months, total hours, days to the next birthday and the weekday you were born on.*

Try it: [Age Calculator](https://toolcloak.pages.dev/tools/age-calculator). Choose a date of birth and, if you need it, a comparison date. See your age in years, months and days.

## When the days do not line up

The tricky case is a target date with a smaller day number than the birth date. Take a birth date of 25 March 1995 and the date 9 October 2026. From 25 March 1995 to 25 September 2026 is 31 years and 6 months. Then 25 September to 9 October is 14 days, which is the month's remaining days and not 9 − 25.

| Born on | Age on 9 October 2026 | Total days |
| --- | --- | --- |
| 1 January 2000 | 26 years, 9 months, 8 days | 9,778 |
| 25 March 1995 | 31 years, 6 months, 14 days | 11,521 |
| 9 October 2000 | 26 years, 0 months, 0 days (birthday today) | 9,496 |

*A person whose birthday is on the comparison date counts the full year, and the next-birthday figure shows 'Today'.*

> **Birthdays at the end of a month:** If you were born on the 31st and the previous month is shorter, the tool counts the days from the last day of that shorter month. A person born on 31 December 1992 is 33 years, 2 months and 1 day old on 1 March 2026.

## How many days have you lived?

Total days is the difference between the two dates. A year has 365 days, or 366 in a leap year, so the average is about 365.2425. That is why a person of 30 has lived about 10,957 days and not exactly 30 × 365 = 10,950.

**Days lived at different ages (about)**

|  | Value |
| --- | --- |
| 18 | 6,574 |
| 21 | 7,670 |
| 25 | 9,131 |
| 30 | 10,957 |
| 40 | 14,610 |
| 50 | 18,262 |
| 60 | 21,915 |

*Values are rounded and can differ by a day or two depending on how many leap days fall in your own life. The tool counts the exact days between the two dates.*

**Born 1 January 2000, on 9 October 2026**

|  | Value | Note |
| --- | --- | --- |
| Total days | 9,778 | Exact count |
| Total weeks | 1,396 | Whole weeks only |
| Total months | 321 | 26 × 12 + 9 |
| Total hours | 234,672 | Days × 24 |

*Total hours is days times 24, so it does not count the part of today that has passed.*

## Age as on a cut-off date

Exam and job notices usually say 'age as on 1 January 2027' or 'as on the last date of application'. Use that date, not today's. The calculator has a second field for the comparison date. Leave it empty for today, or enter the notice date.

**A minimum age of 18 as on 1 January 2027**

| Born 2 January 2009 | Born 1 January 2009 |
| --- | --- |
| 17 years 11 months 30 days | 18 years 0 months 0 days |
| One day short on the cut-off date | Turns 18 on the cut-off date |
| Does not meet the minimum age | Meets the minimum age |
| Birthday falls the day after the cut-off | Check how the notice counts the exact day |

*A single day can decide eligibility. Always check the exact rule: some notices say 'completed 18 years' and others say 'not below 18 years'.*

> **Read the age rule carefully:** Notices differ on whether the cut-off date itself counts, whether age relaxations apply to your category and which proof is accepted. Use the calculator to check your arithmetic, then follow the official notice and your certificate.

## Born on 29 February

A leap-day birthday exists only in leap years. In other years people usually celebrate on 28 February or 1 March, and the law in different countries treats the legal birthday differently. This tool treats the birthday in a non-leap year as 1 March, so a person born on 29 February 2004 has 143 days to the next birthday on 9 October 2026, because that birthday falls on 1 March 2027.

**Leap-day facts**

|  | Value | Note |
| --- | --- | --- |
| Chance of a leap-day birthday | 1 in 1,461 | One leap day in a 4-year cycle of 1,461 days |
| Years without 29 February | 3 of 4 | A birthday on 1 March in this tool |
| Birthdays at 84 years | 21 | Only one in four years |

*A person born on 29 February 2004 has had only five actual 29 February birthdays by 2026, but is 22 years old in calendar terms.*

## Other ways of counting age

- **Completed years** is the standard in most of the world: you are 26 until you reach your next birthday.
- **East Asian age reckoning** traditionally counted a person as one year old at birth. South Korea moved to the international system for most legal and official purposes in June 2023.
- **Age last birthday** is used by many insurers and is the same as completed years.
- **Age nearest birthday** is used by some life insurers and rounds to the closest year.

## A quick checklist for age on a form

1. **Find the date the form asks about.** It may be today, the last date to apply or a fixed cut-off.
2. **Enter your date of birth exactly** as on your certificate.
3. **Use the same date in the second field** if the age is as on a cut-off.
4. **Write the age the way the form asks:** years only, or years, months and days.
5. **Keep the supporting document** ready. The calculator is a check, and your certificate is the proof.

## Frequently asked questions

### How do I calculate my exact age in years, months and days?

Count the full years from your date of birth, then the full months, then the remaining days. If the day of the month in the target date is smaller than your birth day, count one fewer month and take the leftover days from the last full month.

### How many days old am I?

Enter your date of birth in the Age Calculator. It subtracts the two dates to give the exact total days, and also shows weeks, months and hours. Someone born on 1 January 2000 is 9,778 days old on 9 October 2026.

### How do I calculate age as on a particular date?

Enter your date of birth and also the cut-off date in the second field. Leave the second field empty if you want your age as on today.

### When is my next birthday?

The tool shows the number of days to your next birthday. If your birthday is today it says Today. For 1 January 2000 on 9 October 2026, the next birthday is 84 days away.

### How are leap years treated?

The calculation uses real calendar dates, so every leap day is counted in the total days. For a 29 February birthday, the next birthday in a non-leap year is counted as 1 March.

### Is the result accepted as proof of age?

No. The [Age Calculator](https://toolcloak.pages.dev/tools/age-calculator) is a quick check of arithmetic. Official proof of age comes from your birth certificate, school certificate or other document that the form accepts.



---

# The “average” salary says ₹70,000, but most people there earn ₹32,000. Mean, median and mode explained.

> One big number can drag the mean far from what a typical person gets. A seven-person example, when to use mean, median or mode, and the average-speed trap that gives 48 and not 50.

Published 2026-10-09 by Toolcloak. 3 minute read. Original: https://toolcloak.pages.dev/articles/average-salary-says-70000-but-most-people-earn-32000
Related free tool: [Average Calculator](https://toolcloak.pages.dev/tools/average-calculator)

A small company advertises an average monthly salary of ₹70,000. A job seeker hears that, imagines a comfortable middle, and applies. Then the offer comes in at ₹30,000 and everyone on the team earns about the same. Nobody lied. The word “average” just hid which average was used.

There are three common averages, and they answer three different questions. Picking the right one is the difference between a number that describes the group and a number that only sounds good.

## The three averages in one example

Take seven people at that company. Six are employees. One is the owner, who also pays himself through the same payroll. Monthly pay in thousands of rupees:

| Person | Monthly pay |
| --- | --- |
| Employee 1 | ₹25,000 |
| Employee 2 | ₹28,000 |
| Employee 3 | ₹30,000 |
| Employee 4 | ₹32,000 |
| Employee 5 | ₹35,000 |
| Employee 6 | ₹38,000 |
| Owner | ₹3,00,000 |

*Total payroll is ₹4,88,000 across seven people.*

**Same seven salaries, three answers**

|  | Value | Note |
| --- | --- | --- |
| Mean | ₹69,714 | ₹4,88,000 ÷ 7 |
| Median | ₹32,000 | The middle of the sorted list |
| Mode | None | No pay appears twice |

*The mean is more than double what six of the seven people earn. The median stays with the typical person.*

- **Mean**: add everything up and divide by how many there are. Every value counts, so one huge value pulls it up.
- **Median**: sort the list and take the middle one (for an even count, the mean of the two middle ones). It ignores how far the extremes are.
- **Mode**: the value that appears most often. It is the only one that works for categories, like the most common shoe size.

## Why one number can move the mean so much

Remove the owner and the mean drops to ₹31,333. One person moved the “average” by more than ₹38,000. Now add an eighth employee earning ₹40,000: the mean becomes ₹66,000, the median becomes ₹33,500.

**What each average says for the same group**

|  | ₹ per month |
| --- | --- |
| Mean, all 7 | ₹69,714 |
| Median, all 7 | ₹32,000 |
| Mean, without owner | ₹31,333 |
| Median, 8 people | ₹33,500 |

*The median barely moved when the group changed. The mean swung by tens of thousands.*

> **A quick test:** If the mean and the median are far apart, the data has a few extreme values pulling one way. Report the median, or report both and say why they differ.

## Which one to use

| Question | Best average | Why |
| --- | --- | --- |
| What does a typical person earn? | Median | Not distorted by a few very high or low values |
| What will this cost in total, per head? | Mean | You need the total divided fairly, extremes included |
| Which size should the shop stock most? | Mode | It is the most common value |
| Marks in a class with no extreme scores | Mean | Values are balanced and every score matters |

*Mean for totals, median for typical, mode for most common.*

Shoe sizes make the mode concrete. A shop sold sizes 6, 7, 7, 7, 8, 8, 9, 10 and 11 in a day. The mean is 8.1, the median is 8 and the mode is 7. A shop that stocks 8.1 can’t exist. A shop that stocks more 7s sells more shoes.

## The average-speed trap

You drive 120 km to a city at 60 km/h and come back at 40 km/h. The average of 60 and 40 is 50, but your average speed for the trip is not 50. The way out is to ask for total distance over total time.

**Average speed for a round trip**

1. **Time going.** 120 km ÷ 60 km/h = 2 hours.
2. **Time returning.** 120 km ÷ 40 km/h = 3 hours.
3. **Total distance over total time.** 240 km ÷ 5 hours = 48 km/h.

You spent more time at the slower speed, so it counts for more. The same logic shows up with percentages: averaging growth rates directly gives the wrong answer, which [the story of a 50% rise followed by a 50% fall](https://toolcloak.pages.dev/articles/up-50-percent-down-50-percent-and-you-are-not-back-to-zero) shows.

Try it: [Average Calculator](https://toolcloak.pages.dev/tools/average-calculator). Paste a list of numbers and see the mean, median, mode, range and standard deviation side by side.

## Frequently asked questions

### What is the difference between mean, median and mode?

The mean is the sum divided by the count. The median is the middle value when the numbers are sorted. The mode is the value that appears most often.

### Which average is best for salaries?

The median, because a few very high salaries inflate the mean. In the example above the mean is ₹69,714 while six of seven people earn ₹38,000 or less.

### Can there be more than one mode, or none?

Yes. If two values tie for most common there are two modes. If every value appears once there is no mode.

### How do I find the median of an even number of values?

Sort the list and take the mean of the two middle values. For 25, 28, 30, 32, 35, 38, 40 and 300, the middle two are 32 and 35, so the median is 33.5.

### Why is the average speed of 60 and 40 km/h not 50?

Because you spend more time at the slower speed. Use total distance divided by total time: 240 km over 5 hours is 48 km/h.



---

# CGPA to percentage: how to convert with the 9.5 rule and why universities differ

> Job forms and college applications often ask for a percentage, but your marksheet shows a CGPA. This guide covers the CBSE 9.5 multiplier, the reverse conversion, a full lookup table, why other universities use other formulas, and how to avoid a mismatch on your application.

Published 2026-10-09 by Toolcloak. 3 minute read. Original: https://toolcloak.pages.dev/articles/cgpa-to-percentage-how-to-convert-with-the-9-5-rule-and-why-universities-differ
Related free tool: [CGPA to Percentage Converter](https://toolcloak.pages.dev/tools/cgpa-percentage-converter)

A **CGPA** (cumulative grade point average) is an average of grade points on a scale, usually 10 in India. A **percentage** is the share of marks you scored out of the total. The two are not the same measurement, so the conversion needs a formula, and the right formula depends on who issued your result.

> **In short**
> - For CBSE, **percentage = CGPA × 9.5**. A CGPA of 8.4 is 79.8%.
> - The reverse is **CGPA = percentage ÷ 9.5**. A 78.5% is about 8.26.
> - Other boards and universities use their own formulas, such as **(CGPA − 0.75) × 10** or a plain **× 10**. Always check the one printed by your institution.
> - If an application asks for a percentage, use your institution's official rule or certificate, not a random online formula.

## The CBSE rule: multiply by 9.5

CBSE's guidance for its 10-point scale is to multiply the CGPA by 9.5 to get an approximate percentage. The 9.5 comes from studying how grade points relate to marks across the grading bands, and it is a deliberate round figure rather than an exact one.

**Example: CGPA 8.4**

1. **Take the CGPA.** 8.4 on a 10-point scale.
2. **Multiply by 9.5.** 8.4 × 9.5 = **79.8**.
3. **Write the result.** Approximately **79.8%**.

*The reverse: 79.8 ÷ 9.5 = 8.4. This is an approximation, not the actual marks scored.*

Try it: [CGPA to Percentage Converter](https://toolcloak.pages.dev/tools/cgpa-percentage-converter). Enter your CGPA to get the percentage, or the other way round, with the formula shown.

## Lookup table

| CGPA | × 9.5 (CBSE style) | (CGPA − 0.75) × 10 |
| --- | --- | --- |
| 6.0 | 57.0% | 52.5% |
| 6.5 | 61.75% | 57.5% |
| 7.0 | 66.5% | 62.5% |
| 7.5 | 71.25% | 67.5% |
| 8.0 | 76.0% | 72.5% |
| 8.4 | 79.8% | 76.5% |
| 8.5 | 80.75% | 77.5% |
| 9.0 | 85.5% | 82.5% |
| 9.5 | 90.25% | 87.5% |
| 10.0 | 95.0% | 92.5% |

*Two common rules give answers 3 to 4 points apart. The second formula is used by some universities, so ask yours.*

**Percentage for CGPA 6 to 10 under the 9.5 rule**

|  | Percentage |
| --- | --- |
| CGPA 6.0 | 57% |
| CGPA 7.0 | 66.5% |
| CGPA 8.0 | 76% |
| CGPA 9.0 | 85.5% |
| CGPA 10.0 | 95% |

*Under the 9.5 rule even a perfect 10 converts to 95%, not 100%. That is by design.*

## Percentage back to CGPA

Divide by 9.5. A student with 78.5% has a CGPA of about 8.26. Round according to the form's instructions, usually to one or two decimals.

| Percentage | CGPA (÷ 9.5) |
| --- | --- |
| 60% | 6.32 |
| 70% | 7.37 |
| 80% | 8.42 |
| 90% | 9.47 |
| 95% | 10.00 |

*Above 95% the formula would give more than 10, which cannot exist on the scale, so such results are capped.*

## Why different institutions use different formulas

**Four common approaches you may meet**

| Box | What it means |
| --- | --- |
| CGPA × 9.5 | Used by CBSE for its 10-point scale. Widely copied by other schools. |
| (CGPA − 0.75) × 10 | Used by some universities. It treats 0.75 as a baseline offset. |
| CGPA × 10 | A rough shortcut. Simple, but overstates the result compared with the 9.5 rule. |
| Custom university tables | Some universities publish a table per grade band or a formula with a constant. Follow it exactly. |

*A grade point is an average over grading bands, not a score. Universities set their own bands, which is why no formula fits everyone.*

> **Use the official rule:** Employers, universities and scholarship panels may check against your institution's own conversion. Ask your registrar for a conversion certificate if a form needs one, and use the same figure everywhere.

## The same CGPA under three rules

**CGPA 8.4 converted three ways (%)**

|  | Value |
| --- | --- |
| × 9.5 | 79.8% |
| × 10 | 84% |
| (−0.75) × 10 | 76.5% |

*A 7.5-point spread for one CGPA. This is why you should not mix formulas between applications.*

## CGPA, SGPA and GPA: what is the difference?

- **SGPA** is for one semester. It is the grade-point average of that semester's subjects, weighted by credits.
- **CGPA** is cumulative. It averages every semester so far, again weighted by credits.
- **GPA** is the general term. Many countries use a 4.0 scale, so an Indian 10-point CGPA cannot be compared to a US GPA without a conversion service.
- **Percentage** needs the total marks. CGPA hides the exact marks inside each grade band.

Weighted by credits means a 4-credit subject counts twice as much as a 2-credit one. For example, grade points 9 in a 4-credit subject and 7 in a 2-credit subject give (9 × 4 + 7 × 2) ÷ 6 = **8.33**, not the plain average of 8.

## Filling in a form correctly

1. **Read the instruction** next to the field. It may say 'percentage as per university formula'.
2. **Use your institution's rule.** If none is stated, say which one you used.
3. **Round the way you are told.** If no instruction, two decimals is safe.
4. **Keep the CGPA too** if the form has a field for it.
5. **Carry your marksheet.** Verification will use the official document.

## Frequently asked questions

### How do I convert CGPA to percentage?

For CBSE's 10-point scale, multiply the CGPA by 9.5. For example, a CGPA of 8.4 is 79.8 percent. Other universities may use their own formula.

### How do I convert percentage to CGPA?

Divide the percentage by 9.5. A 78.5 percent score is about 8.26 CGPA.

### Why is 9.5 used and not 10?

The multiplier reflects the way grade points map to marks across grading bands. It gives a conservative approximation, so a CGPA of 10 converts to 95 percent.

### Is the converted percentage my actual marks?

No. It is an approximation. The actual marks scored may be higher or lower than the converted figure.

### Do all universities use the same formula?

No. Some use CGPA × 9.5, others use (CGPA − 0.75) × 10, CGPA × 10 or a custom table. Use the rule published by your own institution.

### What is the difference between SGPA and CGPA?

SGPA is the grade-point average of one semester, while CGPA is the cumulative average across all semesters, weighted by credits.



---

# chmod permissions explained: what 755, 644 and 777 actually mean

> Every file on Linux and macOS carries nine permission bits. chmod 755 sets them in one go, but only if you can read the number. This guide decodes the digits, lists the values you will use most, explains umask and the special bits, and shows why 777 is almost never the answer.

Published 2026-10-09 by Toolcloak. 5 minute read. Original: https://toolcloak.pages.dev/articles/chmod-permissions-explained-what-755-644-and-777-actually-mean
Related free tool: [Chmod Calculator](https://toolcloak.pages.dev/tools/chmod-calculator)

When you run `ls -l` you see something like `-rwxr-xr-x`. That string of ten characters is the whole access policy for the file: who may **read** it, who may **write** it and who may **run** it. `chmod` changes it. The three-digit number you type after it, such as 755, is just a compact way of writing those nine switches.

> **In short**
> - Permissions come in **three groups** (owner, group, others) and **three kinds** (read = 4, write = 2, execute = 1).
> - Add the numbers for each group. **7 = rwx**, **6 = rw-**, **5 = r-x**, **4 = r--**, **0 = ---**.
> - **755** is the usual value for programs and folders. **644** is the usual value for ordinary files. **600** is for private keys.
> - **777 lets everyone change the file.** It is rarely needed and often a security hole.

## Reading the permission string

Take `-rwxr-xr-x`. The first character is the file type (`-` for a normal file, `d` for a directory, `l` for a link). The next nine are three triplets, in this order: owner, group, others.

**-rwxr-xr-x, piece by piece**

1. **-.** A regular file. A `d` here would mean a directory.
2. **rwx.** Owner: can read, write and execute.
3. **r-x.** Group: can read and execute, but not write.
4. **r-x.** Others (everyone else): can read and execute, but not write.

*Written as numbers this is 7, 5, 5, so the command is `chmod 755 filename`.*

Try it: [Chmod Calculator](https://toolcloak.pages.dev/tools/chmod-calculator). Tick the boxes to get the octal number and the chmod command, or type a number to see the boxes.

## Why the numbers are 4, 2 and 1

Each triplet is three on-or-off switches, which is a three-digit binary number. Read is the left bit (4), write the middle (2) and execute the right (1). Adding the values you want gives a single digit from 0 to 7.

| Digit | Binary | Symbolic | Meaning |
| --- | --- | --- | --- |
| 0 | 000 | --- | No access |
| 1 | 001 | --x | Execute only |
| 2 | 010 | -w- | Write only |
| 3 | 011 | -wx | Write and execute |
| 4 | 100 | r-- | Read only |
| 5 | 101 | r-x | Read and execute |
| 6 | 110 | rw- | Read and write |
| 7 | 111 | rwx | Full access |

*Three digits, eight options each: 8 × 8 × 8 = 512 possible permission settings before the special bits.*

## The values you will actually use

| Mode | Symbolic | Typical use |
| --- | --- | --- |
| `644` | rw-r--r-- | Web pages, images, config files. Owner edits, everyone else reads. |
| `755` | rwxr-xr-x | Scripts, programs and folders. Everyone can run or enter, only the owner changes. |
| `600` | rw------- | SSH private keys, secrets. Only the owner can touch it. |
| `700` | rwx------ | Private folders and personal scripts. |
| `664` | rw-rw-r-- | Shared project files, where the group may edit. |
| `775` | rwxrwxr-x | Shared project folders for a team group. |
| `640` | rw-r----- | Files a service group may read but others may not, such as some logs. |
| `444` | r--r--r-- | Read-only for all, even the owner. |

*If you are unsure, start with 644 for files and 755 for folders and tighten from there.*

**How much access each common mode gives (read + write + execute bits set, out of 9)**

|  | Bits set |
| --- | --- |
| 600 | 2 |
| 644 | 4 |
| 664 | 5 |
| 755 | 7 |
| 775 | 8 |
| 777 | 9 |

*The more bits that are set, the more people can do. 777 sets all nine.*

## What the permissions mean on folders

On a folder the bits do slightly different jobs. This catches many people out.

**Read, write and execute on a directory**

| Box | What it means |
| --- | --- |
| Read (r) | You can list the names of the files inside, for example with `ls`. |
| Write (w) | You can create, rename and delete files inside. This works only together with execute. |
| Execute (x) | You can enter the folder and reach files whose names you know. Without it, the folder is closed. |
| Why 755 and not 644 | A folder at 644 cannot be entered, because the x bit is missing. That is why folders usually get 755. |

*A folder with read but no execute shows its file names but denies access to the files themselves.*

## Why 777 is a bad habit

A stubborn 'permission denied' error tempts people to run `chmod 777`. It works because it removes every barrier, including the ones that were protecting you. On a shared server or a website folder, any other user or any compromised process can now change or replace the file.

**Fixing a permission error**

| Quick but risky | Better |
| --- | --- |
| chmod 777 | Right owner, 755 or 644 |
| Anyone can edit or replace the file | Use `chown` so the right user owns the file |
| Uploaded malware can overwrite your scripts | Give only the access that is needed |
| Hides the real cause of the error | Fixes the cause, not the symptom |

*Most 'permission denied' errors are really about the wrong owner or group. Check with `ls -l` before changing the mode.*

## Symbolic mode: change one bit at a time

You do not have to type numbers. Letters let you add or remove a single permission without touching the rest.

| Command | What it does |
| --- | --- |
| `chmod u+x script.sh` | Give the owner (user) execute permission |
| `chmod go-w file` | Remove write from group and others |
| `chmod a+r file` | Give everyone read permission (a = all) |
| `chmod u=rw,go=r file` | Set exactly rw- for owner and r-- for group and others (same as 644) |
| `chmod -R 755 folder` | Apply 755 to the folder and everything inside it |

*Careful with -R. Setting 755 on every file makes data files executable, which is rarely what you want.*

## umask: the default for new files

New files do not start at 777. The system removes the bits named in the **umask**. A umask of 022 removes write from group and others, so new folders start at 755 and new files at 644. Files never start with execute, which is why you must chmod a new script yourself.

| umask | New folders | New files | Meaning |
| --- | --- | --- | --- |
| `022` | 755 | 644 | Common default. Others may read. |
| `002` | 775 | 664 | Group may write, used for shared teams. |
| `027` | 750 | 640 | Others get nothing. |
| `077` | 700 | 600 | Private. Only the owner has access. |

*Note the numbers are not a plain subtraction in every case. The umask switches bits off, so check the result with `ls -l`.*

## The special fourth digit

You may see four-digit modes like `4755` or `2775`. The extra leading digit sets three special bits.

**Special permission bits**

| When | What |
| --- | --- |
| 4 | setuid. A program runs with the rights of its owner, not the person who started it. `passwd` uses this. Dangerous on scripts. |
| 2 | setgid. On a program, it runs with the group's rights. On a folder, new files inherit the folder's group, useful for team folders. |
| 1 | sticky bit. On a folder, only a file's owner can delete it. `/tmp` uses this (mode 1777). |

*In the symbolic string these show as `s` in place of x for setuid and setgid, and `t` for the sticky bit.*

> **Keys need strict permissions:** SSH refuses to use a private key that other users can read, with a warning that the permissions are too open. Set it to 600, and the `.ssh` folder to 700.

## A quick checklist

1. **Look first.** Run `ls -l` to see the current owner, group and mode.
2. **Fix ownership before mode** when a service cannot read a file.
3. **Use 644 for files, 755 for folders and programs.**
4. **Use 600 or 700 for anything secret.**
5. **Avoid -R with 777.** Use `find` with `-type f` and `-type d` to set different modes for files and folders.

## Frequently asked questions

### What does chmod 755 mean?

It gives the owner read, write and execute (7), and the group and everyone else read and execute (5 each). It is the usual setting for programs and folders.

### What does chmod 644 mean?

The owner can read and write (6), while the group and others can only read (4). It is the usual setting for normal files such as web pages.

### Why is chmod 777 dangerous?

It lets every user on the system read, change and run the file. On web servers this can let attackers replace your files.

### How do I calculate a chmod number?

Add read = 4, write = 2 and execute = 1 for each of the owner, group and others. For example rw- is 4 + 2 = 6 and r-x is 4 + 1 = 5.

### What is umask?

It is a setting that removes permission bits from newly created files and folders. With a umask of 022, new folders are 755 and new files are 644.

### What do the first characters in ls -l mean?

The first character is the type: a dash for a file, d for a directory, l for a link. The next nine characters are the owner, group and others permissions.



---

# Commission calculator: how to work out sales commission, a bonus rate above target and total pay

> Commission looks simple until a bonus rate starts above a sales target. This guide shows flat and tiered commission, which sales the higher rate applies to, the effective rate you really earn, and how a base salary changes total pay.

Published 2026-10-09 by Toolcloak. 3 minute read. Original: https://toolcloak.pages.dev/articles/commission-calculator-how-to-work-out-sales-commission-a-bonus-rate-above-target-and-total-pay
Related free tool: [Sales Commission Calculator](https://toolcloak.pages.dev/tools/commission-calculator)

**Commission** is pay that rises with what you sell. The simplest plan pays one percentage of all sales. Many plans pay a higher percentage once sales pass a target. The key question for the second kind is whether the higher rate applies to **all** the sales or only to the sales **above** the target. The answer changes the pay a lot.

> **In short**
> - **Flat commission = sales × rate.** On ₹5,00,000 of sales at 5% that is **₹25,000**.
> - With a 5% rate up to a ₹4,00,000 target and 8% above it, the same sales earn ₹4,00,000 × 5% + ₹1,00,000 × 8% = **₹28,000**.
> - The **effective rate** is commission ÷ sales, here **5.6%**, which sits between the two rates.
> - In this tool the bonus rate applies **only to the sales above the target**. Check how your own plan works.

## The formula, step by step

**Example: ₹5,00,000 of sales, 5% up to ₹4,00,000, 8% above**

1. **Commission on sales up to the target.** ₹4,00,000 × 5% = **₹20,000**.
2. **Find the sales above the target.** ₹5,00,000 − ₹4,00,000 = **₹1,00,000**.
3. **Commission at the bonus rate.** ₹1,00,000 × 8% = **₹8,000**.
4. **Add them.** ₹20,000 + ₹8,000 = **₹28,000**.

*If sales are at or below the target, or no target is set, the tool simply uses sales × the standard rate.*

Try it: [Sales Commission Calculator](https://toolcloak.pages.dev/tools/commission-calculator). Enter sales, the rate, an optional target and bonus rate, and a base salary to see commission and total earnings.

## What different sales levels earn

| Total sales | Commission | Effective rate | Sales above target |
| --- | --- | --- | --- |
| ₹3,00,000 | ₹15,000 | 5.00% | ₹0 |
| ₹4,00,000 | ₹20,000 | 5.00% | ₹0 |
| ₹5,00,000 | ₹28,000 | 5.60% | ₹1,00,000 |
| ₹6,00,000 | ₹36,000 | 6.00% | ₹2,00,000 |
| ₹8,00,000 | ₹52,000 | 6.50% | ₹4,00,000 |

*Same plan in every row: 5% up to ₹4,00,000, 8% above. The effective rate rises with sales but never reaches the bonus rate.*

**Commission earned at each sales level (₹ thousand)**

|  | Value |
| --- | --- |
| ₹300k | ₹15 k |
| ₹400k | ₹20 k |
| ₹500k | ₹28 k |
| ₹600k | ₹36 k |
| ₹800k | ₹52 k |

*The slope gets steeper after the target, which is why the last ₹1 lakh earns more than the first.*

## Higher rate on everything, or only on the extra?

Plans differ. In an **incremental** plan, the higher rate applies only to sales above the target, as in this calculator. In a **retroactive** plan, once you pass the target the higher rate applies to all sales. With a retroactive plan, ₹5,00,000 would earn ₹5,00,000 × 8% = ₹40,000, which is ₹12,000 more than the incremental result.

**The same sales under two plan styles**

| Incremental | Retroactive |
| --- | --- |
| ₹28,000 | ₹40,000 |
| 5% on the first ₹4,00,000 | 8% on all ₹5,00,000 |
| 8% only on the ₹1,00,000 above | Higher rate backdated to the first rupee |
| What this calculator computes | Compute it as sales × bonus rate |

*Ask your employer which plan you are on, and get the wording in writing. The calculator covers only the first style.*

> **Plans often have extra rules:** Some employers cap commission, pay it only on collected payments, deduct returns and refunds, or split a sale between two people. The calculator uses the simple version. Compare its answer with your plan document, not just the formula.

## Adding a base salary

If you also earn a base salary, total earnings are base plus commission. With a base of ₹15,000 the example month pays ₹15,000 + ₹28,000 = **₹43,000**. Commission is usually taxable income like salary, so the figure is before tax.

**The example month in four numbers**

|  | Value | Note |
| --- | --- | --- |
| Total sales | ₹5,00,000 | The month's sales |
| Commission | ₹28,000 | Incremental plan |
| Effective rate | 5.6% | Commission ÷ sales |
| Total with a ₹15,000 base | ₹43,000 | Before tax |

*Use the effective rate to compare different plans on the same sales.*

## Checking a plan before you accept it

1. **Find the rate and the target** in writing.
2. **Ask whether the higher rate is incremental or retroactive.**
3. **Check what counts as a sale:** booked, invoiced or paid.
4. **Check the caps, clawbacks and returns rules.**
5. **Run your realistic sales** through the calculator, and a poor month too.

## Frequently asked questions

### How do I calculate sales commission?

Multiply the sales by the commission rate. For 5,00,000 of sales at 5%, the commission is 25,000.

### How does a bonus rate above target work in this calculator?

The standard rate applies up to the target and the bonus rate applies only to the sales above it. With a 4,00,000 target, 5% and 8%, sales of 5,00,000 earn 20,000 plus 8,000, which is 28,000.

### What is the effective commission rate?

It is the total commission divided by the total sales. In the example it is 5.6%, between the 5% base rate and the 8% bonus rate.

### What is the difference between incremental and retroactive commission?

Incremental pays the higher rate only on sales above the target. Retroactive pays the higher rate on all sales once the target is passed. The retroactive plan pays more.

### Does the calculator include base salary?

Yes, there is an optional base salary field. The total earnings line adds it to the commission. It does not include tax.

### Can I leave the target empty?

Yes. With no target, or a target of zero, the tool uses the standard rate on all sales. Use the [Percentage Calculator](https://toolcloak.pages.dev/tools/percentage-calculator) for a quick one-off percentage.



---

# Confidence intervals explained: what 95% really means and how sample size shrinks it

> A poll says 52% support a plan, give or take 3 points. A confidence interval is that 'give or take', done properly. This guide shows the formula, a worked example, how level and sample size change the width, and the common misreading of '95% confident'.

Published 2026-10-09 by Toolcloak. 3 minute read. Original: https://toolcloak.pages.dev/articles/confidence-intervals-explained-what-95-percent-really-means-and-how-sample-size-shrinks-it
Related free tool: [Confidence Interval Calculator (Mean)](https://toolcloak.pages.dev/tools/confidence-interval-calculator)

You measured a sample and got one number: an average, or a percentage. The true value for the whole population is almost certainly a little different. A **confidence interval** is a range built around your number that says how far off it could reasonably be. It is the honest way to report an estimate.

> **In short**
> - For a mean: **interval = mean ± (critical value × standard error)**, where standard error = standard deviation ÷ √n.
> - At **95%** the critical value is about **1.96** (z) for large samples. At 90% it is 1.645, at 99% it is 2.576.
> - To **halve** the width you need **four times** the sample. Width shrinks with the square root of n.
> - '95% confident' describes the **method**: 95% of intervals built this way would capture the true value. It is not a 95% chance for one interval.

## A worked example

A company samples 36 delivery times. The average is 72 minutes and the standard deviation is 12 minutes. What range is the true average likely to sit in?

**95% interval for the average delivery time**

1. **Standard error.** 12 ÷ √36 = 12 ÷ 6 = **2 minutes**.
2. **Margin of error.** 1.96 × 2 = **3.92 minutes**.
3. **Interval.** 72 − 3.92 to 72 + 3.92 = **68.08 to 75.92 minutes**.
4. **Say it.** We are 95% confident the true average is between about 68 and 76 minutes.

*The sample mean sits in the middle. The margin of error is how far the interval reaches on each side.*

Try it: [Confidence Interval Calculator (Mean)](https://toolcloak.pages.dev/tools/confidence-interval-calculator). Enter mean, standard deviation and sample size to get the interval at your chosen level.

## Using the t value for small samples

The 1.96 above assumes you know the population standard deviation, or have a big sample. With a sample of 36 and an estimated spread, textbooks use the **t distribution**, which is a little wider to allow for that extra uncertainty. For 35 degrees of freedom the 95% critical value is about 2.03.

| Method | Critical value | Margin | Interval |
| --- | --- | --- | --- |
| z (1.96) | 1.960 | 3.92 | 68.08 to 75.92 |
| t, 35 df (2.030) | 2.030 | 4.06 | 67.94 to 76.06 |

*The difference is small at n = 36 and tiny by n = 100. For n under about 30 it matters more.*

**95% critical value by sample size**

|  | Critical value |
| --- | --- |
| n = 5 (t) | 2.776 |
| n = 10 (t) | 2.262 |
| n = 30 (t) | 2.045 |
| n = 100 (t) | 1.984 |
| Large n (z) | 1.960 |

*The t value falls toward 1.96 as the sample grows, because a bigger sample estimates the spread better.*

## What changes the width

Three things control how wide the interval is: how variable the data is, how many values you have, and how sure you want to be.

**Margin of error at 95% for different sample sizes (SD = 12, minutes)**

|  | Value |
| --- | --- |
| n = 9 | 7.84 |
| n = 36 | 3.92 |
| n = 100 | 2.35 |
| n = 400 | 1.18 |

*Going from 9 to 36 values (four times as many) cuts the margin in half. From 100 to 400 does the same again.*

| Confidence level | Critical value (z) | Margin at n = 36 | Interval around 72 |
| --- | --- | --- | --- |
| 90% | 1.645 | 3.29 | 68.71 to 75.29 |
| 95% | 1.960 | 3.92 | 68.08 to 75.92 |
| 99% | 2.576 | 5.15 | 66.85 to 77.15 |

*More confidence means a wider net. You cannot be more certain without being less precise.*

## Polls and percentages

For a proportion the formula changes slightly. The standard error is √(p × (1 − p) ÷ n). A poll of 1,000 people finds 52% in favour.

**The poll of 1,000 people**

|  | Value | Note |
| --- | --- | --- |
| Sample result | 52% | 520 of 1,000 |
| Margin of error | ±3.1 | 1.96 × √(0.52 × 0.48 ÷ 1000) |
| Lower end | 48.9% | Plausible low |
| Upper end | 55.1% | Plausible high |

*The interval includes 50%, so this poll cannot tell you whether most people are for or against. A 52% 'lead' is within the noise.*

Headline margins use the worst case, p = 50%, which gives the widest interval. That is where the familiar 'plus or minus 3 points for 1,000 people' comes from.

**Poll margin of error as the sample grows (percentage points)**

| People surveyed | Margin of error at 95% |
| --- | --- |
| 100 | 9.8 |
| 400 | 4.9 |
| 1,000 | 3.1 |
| 2,500 | 2 |
| 10,000 | 1 |

*Surveying ten times as many people shrinks the margin to only about a third of its size, and each extra person helps less. That is why most polls stop around 1,000.*

## What '95% confident' really means

**Two readings of the same interval**

| The correct reading | The common misreading |
| --- | --- |
| About the method | A 95% chance for this one |
| If we repeated the study many times | Treats the true value as a moving target |
| About 95% of the intervals would contain the true value | The true value is fixed, the interval varies |
| Any single interval either does or does not | A subtle but real difference |

*In everyday writing, 'we are 95% confident the true value lies between 68 and 76' is fine. Just do not read it as a probability about the true value itself.*

**100 intervals from 100 repeated samples at 95% confidence**

| Group | Count |
| --- | --- |
| Contain the true value | 95 |
| Miss it | 5 |

*Five in a hundred intervals will miss, and you cannot tell which ones from a single sample.*

## Assumptions behind the formula

- **Random sampling.** If the sample is biased, the interval is centred on the wrong value no matter how narrow it is.
- **Independent values.** Ten measurements on the same person are not ten independent people.
- **Roughly bell-shaped means.** With a large sample this holds even for skewed data. For tiny, skewed samples, be careful.
- **Spread is estimated.** The standard deviation comes from your sample, so it carries its own uncertainty.

> **Narrow does not mean right:** A very large biased sample gives a very narrow interval around the wrong answer. The interval measures random error only, not mistakes in how the data was collected.

## Choosing a sample size

You can work backwards. For a proportion, the sample needed for margin E at 95% is about 0.9604 ÷ E² (using p = 50%). For a margin of 3 points, that is 0.9604 ÷ 0.0009 ≈ **1,067 people**. For 5 points it is about 385.

1. **Decide the margin you can live with.**
2. **Pick a confidence level**, usually 95%.
3. **Estimate the spread**, from a pilot study or by using the worst case for proportions.
4. **Calculate n**, then add a margin for non-responses.
5. **Report the interval, not just the mean.**

## Frequently asked questions

### What is a confidence interval?

It is a range around a sample estimate, such as an average or percentage, that is likely to contain the true population value. It is calculated from the sample's size and variability.

### How do I calculate a 95% confidence interval for a mean?

Compute the standard error as the standard deviation divided by the square root of n, multiply by 1.96 (or the t value for small samples) and add and subtract the result from the mean.

### What does 95% confidence mean?

It means that if the study were repeated many times, about 95% of the intervals built this way would contain the true value. It does not mean there is a 95% chance this one does.

### How does sample size affect the interval?

A larger sample gives a narrower interval. The width shrinks with the square root of n, so you need four times the sample to halve the margin.

### When should I use t instead of z?

Use t when the population standard deviation is unknown and estimated from the sample, especially for samples under about 30. For large samples the two give nearly the same result.

### What is margin of error?

It is the amount added to and subtracted from the estimate to form the interval. In a poll of 1,000 people the margin is about 3 percentage points.



---

# Correlation: what r tells you, and what it never can

> Ice cream sales and drownings rise together. Does ice cream cause drowning? One number, r, measures how closely two lists move together. It is easy to compute and even easier to misread. Here is how to read it properly.

Published 2026-10-09 by Toolcloak. 3 minute read. Original: https://toolcloak.pages.dev/articles/correlation-what-r-tells-you-and-what-it-never-can
Related free tool: [Correlation Coefficient Calculator (Pearson r)](https://toolcloak.pages.dev/tools/correlation-coefficient-calculator)

You have two lists of numbers for the same ten people: hours studied and exam score. Do they move together? The **Pearson correlation coefficient**, written r, answers that in one number between −1 and +1. It is one of the most used statistics there is, and one of the most misunderstood.

> **In short**
> - **r = +1** is a perfect upward line, **−1** a perfect downward line, **0** no straight-line link at all.
> - **r²** shows how much of the ups and downs of one list lines up with the other. r = 0.98 gives r² = 0.97.
> - A single outlier can crash r from 0.98 to 0.16. Always look at the points, not just the number.
> - Correlation is **not** causation. r never tells you why two things move together.

## What r actually measures

Pearson's r measures how close the points would sit to a **straight line**. Positive means both rise together. Negative means one rises as the other falls. The closer to 1 or −1, the tighter the points hug the line. Near zero, there is no straight-line pattern to speak of.

**A rough way to read r**

| Zone | From | To |
| --- | --- | --- |
| Strong − | -1 | -0.7 |
| Medium − | -0.7 | -0.3 |
| Weak | -0.3 | 0.3 |
| Medium + | 0.3 | 0.7 |
| Strong + | 0.7 | 1 |

*These cut-offs are a convention, not a law. In physics, 0.9 can be disappointing. In social science, 0.4 can be notable. What counts as strong depends on the field.*

## A worked example

Ten students record the hours they studied the week before an exam and their score out of 100.

| Student | Hours | Score |
| --- | --- | --- |
| 1 | 1 | 48 |
| 2 | 2 | 52 |
| 3 | 2 | 55 |
| 4 | 3 | 60 |
| 5 | 4 | 62 |
| 6 | 4 | 70 |
| 7 | 5 | 68 |
| 8 | 6 | 78 |
| 9 | 7 | 82 |
| 10 | 8 | 88 |

*Paste the two columns as pairs, for example 1, 48, one pair per line, into the calculator.*

The result is **r = 0.985**. The points sit very close to a rising line. In rough terms each extra hour goes with about 5.7 extra marks in this small group.

Try it: [Correlation Coefficient Calculator (Pearson r)](https://toolcloak.pages.dev/tools/correlation-coefficient-calculator). Enter x, y pairs one per line to get r, the strength and the direction.

## r² : the share that lines up

Square r and you get r², sometimes called the coefficient of determination. It is the share of the variation in one list that lines up with the other through a straight line. Here r² = 0.985² ≈ **0.97**, so about 97 percent of the spread in scores matches the spread in hours.

**How the variation in scores splits up (r² = 0.97)**

|  | Share |
| --- | --- |
| Matches the hours studied | 97% |
| Everything else (sleep, luck, topic, nerves) | 3% |

*'Lines up with' is careful wording on purpose. It describes the pattern in this group. It does not say the hours caused 97 percent of the result.*

## One outlier can break the number

Suppose student 10 studied for 8 hours but scored 30, perhaps because they fell ill on the day. Nothing else changes. Watch r:

**r for the same group under three small changes**

|  | r |
| --- | --- |
| All ten students | r = 0.98 |
| Drop student 10 | r = 0.98 |
| Student 10 scores 30, not 88 | r = 0.16 |

*One unusual point dragged r from strong to weak. This is why you should always plot your data before trusting r. A scatter plot would show that nine students follow a clear line and one does not.*

## r = 0 does not mean 'no relationship'

Pearson's r only detects **straight-line** patterns. Take x from −3 to 3 and let y = x². Every y is fully determined by x, yet r is exactly 0, because the left half of the curve cancels the right half.

**Two perfect patterns, two very different r**

| y rises by 5 for every x | y = x² (a U shape) |
| --- | --- |
| r = 1.00 | r = 0.00 |
| Straight line | Perfect relationship, but curved |
| r captures it fully | Up on one side, down on the other |
| Easy to summarise | r sees nothing |

*Moral: r = 0 says there is no straight-line link. Curves, clusters and thresholds can all hide behind a small r.*

## Correlation is not causation

If two lists are correlated there are at least four stories that could explain it. The number cannot tell you which one is true.

**Four reasons A and B can move together**

| Box | What it means |
| --- | --- |
| A causes B | More study hours really do raise scores. Plausible, but still needs testing. |
| B causes A | Students who score well early feel motivated and study more. |
| A third thing causes both | Hot weather raises both ice cream sales and swimming, and so drownings. |
| Pure chance | With few points, random patterns look convincing. |

*The ice cream and drowning pair is the classic case of a third factor. Summer drives both. Neither causes the other.*

## Small samples fool you easily

With only a handful of pairs, a high r can appear by luck. The table shows roughly how large |r| must be before it would be unusual if the two lists were really unrelated, using the usual 5 percent test.

**Smallest |r| that would be unlikely by chance (approx., two-sided 5% level)**

|  | |r| |
| --- | --- |
| 5 pairs | 0.88 |
| 10 pairs | 0.63 |
| 20 pairs | 0.44 |
| 30 pairs | 0.36 |
| 100 pairs | 0.20 |

*Standard critical values for Pearson's r. With 5 pairs even r = 0.8 is not convincing. With 100 pairs, r = 0.25 is statistically clear, though it may still be a weak, small effect.*

## A short checklist

1. **Plot first.** Draw the scatter. Look for outliers, curves and separate clusters.
2. **Compute r.** Use it as a summary of the straight-line part only.
3. **Check the sample size.** Few pairs mean wide uncertainty.
4. **Think about a third factor.** Ask what else could drive both lists.
5. **Claim carefully.** Say 'is associated with', not 'causes', unless an experiment supports it.

> **Try it:** Paste the ten pairs from the table into the calculator, then change the last score to 30. You should see r fall from about 0.98 to about 0.16.

## Frequently asked questions

### What does the correlation coefficient r tell you?

Pearson's r measures how closely two sets of numbers follow a straight line. It runs from −1 (perfect downward line) through 0 (no straight-line link) to +1 (perfect upward line).

### What is a good correlation coefficient?

It depends on the field. A common rough guide calls 0.7 or above strong, 0.3 to 0.7 moderate and below 0.3 weak, but these cut-offs are conventions, not rules.

### What is the difference between r and r squared?

r² is r multiplied by itself. It shows the share of the variation in one variable that lines up with the other through a straight line. An r of 0.98 gives an r² of about 0.97.

### Does correlation mean causation?

No. Two things can be correlated because one causes the other, because the effect runs the other way, because a third factor drives both, or by chance.

### Can r be zero even when two variables are related?

Yes. Pearson's r only detects straight-line patterns. A perfect U-shaped relationship such as y = x² can give r = 0.



---

# Days until a date: how to count days, weeks and months to an event, and plan backwards from it

> A deadline feels different when it is 84 days away than when it is 'three months' away. This guide shows how a countdown is worked out, whether today is counted, how to turn days into weeks and months, and how to plan a schedule backwards from a date.

Published 2026-10-09 by Toolcloak. 3 minute read. Original: https://toolcloak.pages.dev/articles/days-until-a-date-calculator-how-to-count-days-weeks-and-months-to-an-event-and-plan-backwards
Related free tool: [Days Until Calculator](https://toolcloak.pages.dev/tools/days-until-date-calculator)

A **countdown** turns a date on the calendar into a number you can plan with. 'In three months' is vague. '84 days, 12 weeks, 60 working days' is something you can divide into tasks. The calculation is a date subtraction, but the details, such as whether today counts, change the answer by one day.

> **In short**
> - From **9 October 2026** to **1 January 2027** is **84 days**, exactly **12 weeks**, about **2.8 months** and **2,016 hours**. That date falls on a **Friday**.
> - The tool counts from the start of today, so today itself is **not** included. If the target is today it says 'Today!'.
> - Weeks = days ÷ 7. Months are approximate, using an average month of 30.44 days.
> - Calendar days include weekends. Between those two dates there are about **60 working days** (Monday to Friday).

## How the count works

**Example: counting to 1 January 2027**

1. **Take today's date.** 9 October 2026, at the start of the day.
2. **Take the target date.** 1 January 2027.
3. **Subtract and count whole days.** 22 left in October + 30 in November + 31 in December + 1 = **84 days**.
4. **Convert.** 84 ÷ 7 = **12.0 weeks**, 84 ÷ 30.44 = **2.8 months**, 84 × 24 = **2,016 hours**.

*The tool shows days left (or days passed, if the date is behind you), weeks, approximate months, hours and the weekday of the target.*

Try it: [Days Until Calculator](https://toolcloak.pages.dev/tools/days-until-date-calculator). Pick a target date to see the days, weeks, months and hours left, and the weekday it falls on.

## Counting to well-known dates from 9 October 2026

| Date | Day | Weekday | Days left | Weeks left |
| --- | --- | --- | --- | --- |
| Christmas Day | 25 December | Friday | 77 | 11.0 |
| New Year's Day | 1 January | Friday | 84 | 12.0 |
| Republic Day | 26 January | Tuesday | 109 | 15.6 |
| End of the financial year | 31 March | Wednesday | 173 | 24.7 |
| Independence Day | 15 August | Sunday | 310 | 44.3 |

*Counted on 9 October 2026, the date this guide was written. The tool counts from the date on your own device, so your numbers change every day.*

**Days from 9 October 2026 to each date**

|  | Value |
| --- | --- |
| 25 Dec | 77 |
| 1 Jan | 84 |
| 26 Jan | 109 |
| 31 Mar | 173 |
| 15 Aug | 310 |

*Counted in whole calendar days. Add or remove a day if your rule counts the last day too.*

## Is today counted?

This is the main source of off-by-one disagreements. The tool counts the number of midnights between today and the target, so tomorrow is 1 day away and today is 0. In a legal or contractual notice period, the rule may count differently, for example 'within 30 days of receipt', where the first day may or may not be counted. Check the wording of your own document.

> **For legal deadlines, read the rule:** Court deadlines, tax filing dates and notice periods often have specific counting rules, including what happens when the last day is a holiday. Use the calculator for planning, and confirm the official date with the authority or your lawyer.

## Weeks and months are not the same size

- A **week** is always 7 days, so weeks are exact: 84 days is 12 weeks.
- A **month** is 28 to 31 days. The tool divides by an average of 30.44 days, so '2.8 months' is an estimate.
- Counting by **calendar months** from 9 October to 1 January gives 2 months and 23 days, which is a different way to say the same 84 days.

## Calendar days against working days

For planning work, calendar days overstate the time you have. Of the 84 days to 1 January, only about 60 are weekdays, and holidays reduce that further. Use the [Working Days Calculator](https://toolcloak.pages.dev/tools/working-days-calculator) when you need a deadline in business days.

**The same 84 days, counted three ways**

|  | Value | Note |
| --- | --- | --- |
| Calendar days | 84 | Every day |
| Full weeks | 12 | 84 ÷ 7 |
| Weekdays | 60 | Monday to Friday, no holidays |
| Hours | 2,016 | Days × 24 |

*Weekday counts vary by the day you start. Subtract public holidays and leave days for your own total.*

## Planning backwards from the date

1. **Count the days left** to the date with the calculator.
2. **Convert to weeks** and subtract the weeks you cannot work, such as travel or festivals.
3. **Split the remaining weeks** into phases, for example learn, practise and revise for an exam.
4. **Set a milestone date** at about the halfway point to check progress.
5. **Add a buffer** of at least 10% of the time, because things always take longer than planned.

**84 days split into phases for an exam or project**

| When | What |
| --- | --- |
|  | Weeks 1 to 6. Learn the material or build the main part of the work. |
|  | Weeks 7 to 10. Practise or test, and fix the weak spots. |
|  | Weeks 11 to 12. Revise, finish and keep the last days free as a buffer. |

*A rough split like this works for many projects. Adjust the weeks to the size of the job.*

## Frequently asked questions

### How do I calculate the number of days until a date?

Subtract today's date from the target date and count the whole days between them. For example, from 9 October 2026 to 1 January 2027 is 84 days.

### Does the calculator count today?

No. It counts the days from the start of today, so tomorrow is 1 day away and today is 0. If the target is today, it shows Today.

### How many weeks are there until a date?

Divide the number of days by 7. Eighty-four days is exactly 12 weeks.

### Why are the months approximate?

Calendar months have 28 to 31 days. The tool divides the days by an average month length of 30.44, so the number of months is an estimate.

### What if the date has already passed?

The tool shows the number of days passed instead of days left, with the same weeks, months and hours for the gap.

### Does it count working days?

No. It counts calendar days. For weekdays only, use the [Working Days Calculator](https://toolcloak.pages.dev/tools/working-days-calculator), and subtract any holidays that apply to you.



---

# Snowball or avalanche: which way to clear several debts?

> Pay the smallest balance first, or the highest interest first? People argue about it online as if one answer is right. We ran the same four debts through both methods. The gap was ₹1,289. The real lesson is somewhere else.

Published 2026-10-09 by Toolcloak. 3 minute read. Original: https://toolcloak.pages.dev/articles/snowball-or-avalanche-which-way-to-clear-several-debts
Related free tool: [Debt Payoff Calculator (Snowball vs Avalanche)](https://toolcloak.pages.dev/tools/debt-payoff-calculator)

You have four debts and a little extra money each month. Every rupee you add can only go to one of them. Which one should get it? There are two famous answers, and they have two famous nicknames: **snowball** and **avalanche**.

> **In short**
> - **Snowball**: pay the smallest balance first. **Avalanche**: pay the highest interest rate first.
> - Avalanche never costs more interest. Snowball often costs only a little more and gives quicker wins.
> - In our test both finished in 27 months and the interest differed by ₹1,289 on ₹3.2 lakh of debt.
> - Whatever you pick, keep paying the **full total** each month, even after a debt is cleared.

## The two methods in one picture

**Same money, different order**

| Snowball | Avalanche |
| --- | --- |
| Smallest balance first | Highest rate first |
| Clears a whole debt quickly | Cuts interest fastest |
| Fewer bills to track sooner | Mathematically the cheapest |
| Builds momentum and motivation | First win can take longer |
| May cost a bit more interest | Needs patience if the top-rate debt is large |

*Both methods pay the **minimum** on every debt. They differ only in where the extra money goes.*

## Our four debts

To compare fairly we need one set of debts. Here is the example used for everything below. Interest is charged monthly at one twelfth of the yearly rate, and every month we pay the minimums plus an extra ₹5,000, so ₹14,100 goes out in total.

| Debt | Balance | Interest a year | Minimum payment |
| --- | --- | --- | --- |
| Card A | ₹20,000 | 36% | ₹600 |
| Card B | ₹60,000 | 24% | ₹1,500 |
| Personal loan | ₹1,50,000 | 14% | ₹4,000 |
| Bike loan | ₹90,000 | 10% | ₹3,000 |

*Total owed: ₹3,20,000. Total minimums: ₹9,100. Add ₹5,000 extra and the monthly budget is ₹14,100.*

> **A warning inside the table:** Card A's minimum is ₹600. Its monthly interest is 3% of ₹20,000, which is also **₹600**. Pay only the minimum and the balance never moves. This is how a card can stay on your books for years while you pay on time every month.

**Where your first ₹14,100 goes**

|  | Share |
| --- | --- |
| Interest charged this month | ₹4,300 |
| Actually reduces what you owe | ₹9,800 |

*Almost a third of month one goes to interest. Most of it is from the card at 36 percent and the personal loan.*

Try it: [Debt Payoff Calculator (Snowball vs Avalanche)](https://toolcloak.pages.dev/tools/debt-payoff-calculator). Paste your debts, one per line, add your extra amount and compare both methods.

## How each method plays out

**Snowball** orders by balance: Card A (₹20,000), Card B (₹60,000), Bike loan (₹90,000), Personal loan (₹1,50,000). **Avalanche** orders by rate: Card A (36%), Card B (24%), Personal loan (14%), Bike loan (10%). In our example the first two are the same in both lists, which is why the early months match exactly.

**When each debt disappears**

| When | What |
| --- | --- |
| Month 4 | Card A is gone. Both methods. It is both the smallest and the most expensive. |
| Month 13 | Card B is gone. Both methods. Its old minimum now joins the pile. |
| Month 20 | Snowball clears the bike loan. Avalanche is still working on the personal loan. |
| Month 25 | Avalanche clears the personal loan. Snowball is still paying it, with the bike loan already finished. |
| Month 27 | Debt-free either way. Snowball finishes the personal loan. Avalanche finishes the bike loan. |

*Snowball has three debts cleared by month 20. Avalanche has two. That is the motivation difference, in a calendar.*

**Total debt left, month by month (thousand rupees)**

| Months | Snowball | Avalanche |
| --- | --- | --- |
| 0 | 320000 | 320000 |
| 3 | 290000 | 290000 |
| 6 | 258200 | 258200 |
| 9 | 224800 | 224800 |
| 12 | 189800 | 189800 |
| 15 | 153200 | 153100 |
| 18 | 115500 | 115200 |
| 21 | 76800 | 76100 |
| 24 | 36700 | 35600 |
| 27 | 0 | 0 |

*The two lines are nearly on top of each other. By month 24 avalanche is ahead by about ₹1,100, and then both reach zero in the same month.*

**Total interest paid until debt-free**

|  | Interest |
| --- | --- |
| Snowball | ₹55,877 |
| Avalanche | ₹54,588 |

*Avalanche saves ₹1,289, about 2.3 percent of the interest. Real, but small next to the bigger choices below.*

## When the choice does matter

Our debts were close in size and the biggest rate belonged to the smallest balance, so the two methods barely differed. The gap grows when the **largest** balance carries the **highest** rate. Then snowball leaves the expensive debt for last and you pay its interest for much longer. If you have a big loan at 30 percent and several small ones at 10, avalanche can save noticeably more.

**Pick by your situation**

| Box | What it means |
| --- | --- |
| Rates are all similar | Snowball is nearly free. Take the quick wins. |
| One debt is far costlier | Avalanche. Attack it first, whatever its size. |
| You tend to give up | Snowball. A plan you finish beats a perfect plan you drop. |
| Small balance and top rate | Both methods agree. Start there. |

*If you cannot decide, begin with the debt that has the highest rate **and** a small balance. It is usually the best of both worlds.*

## Five rules that matter more than the method

1. **Never miss a minimum.** A late fee or penalty rate can wipe out a month of progress.
2. **Keep the total payment fixed.** When a debt is cleared, roll its old minimum into the next target. This is the actual snowball.
3. **Stop adding new debt.** Paying down a card while spending on it is a treadmill.
4. **Keep a small emergency cushion.** Without it, the next surprise bill goes back on the card.
5. **Ask about cheaper money.** A balance transfer or a consolidation loan at a lower rate can help, but check the fees and do the maths first.

> **Make it visible:** Write the debts on a sheet with the payoff month beside each. Cross them off as they go. It sounds childish and it works, because the payoff date is more motivating than the balance.

## Frequently asked questions

### What is the difference between the debt snowball and the debt avalanche?

The snowball pays extra money to the smallest balance first. The avalanche pays it to the debt with the highest interest rate first. Both pay the minimum on every other debt.

### Which saves more money, snowball or avalanche?

Avalanche never costs more interest and usually saves some. In a four-debt example of ₹3.2 lakh it saved ₹1,289, with both methods finishing in 27 months.

### Why does snowball work for many people?

Clearing a whole debt early gives a visible win and removes a bill. That motivation helps people keep going, which is often what decides whether the debt actually gets paid.

### Should I keep paying the same total after one debt is cleared?

Yes. Move the freed-up minimum to the next debt on the list. The total monthly payment stays the same while each debt falls faster.

### Can paying only the minimum on a credit card keep me in debt forever?

It can. If the minimum is about equal to the monthly interest, the balance barely falls. On a ₹20,000 balance at 36 percent the monthly interest is ₹600.



---

# Dividend yield explained: the yield trap, payout ratio and the power of reinvesting

> A high dividend yield looks like free money. Sometimes it is; sometimes it is a warning sign. This guide shows how yield is calculated, why it can jump without the company paying a rupee more, what payout ratio tells you and how reinvested dividends change a 20-year result.

Published 2026-10-09 by Toolcloak. 4 minute read. Original: https://toolcloak.pages.dev/articles/dividend-yield-explained-the-yield-trap-payout-ratio-and-reinvesting
Related free tool: [Dividend Yield Calculator](https://toolcloak.pages.dev/tools/dividend-yield-calculator)

When a company earns profit, it can keep the money to grow or pay part of it to shareholders as a **dividend**. The **dividend yield** says how big that payment is compared with the share price: the yearly dividend per share divided by the current price, shown as a percentage.

> **In short**
> - **Yield = yearly dividend per share ÷ share price × 100.** A ₹20 dividend on a ₹500 share is a 4% yield.
> - Yield moves when the price moves. If the price halves and the dividend stays, the yield **doubles**, which can be bad news, not a bargain.
> - The **payout ratio** (dividend ÷ profit per share) shows whether the dividend can last.
> - Reinvesting dividends is a big part of long-term returns, but dividends are taxed, and nothing here is advice to buy any share.

## The formula in one line

**Dividend yield for a ₹500 share that paid ₹5 each quarter**

1. **Add up a year.** 4 payments × ₹5 = ₹20 per share in the last 12 months.
2. **Divide by the price.** ₹20 ÷ ₹500 = 0.04.
3. **Turn it into a percentage.** 0.04 × 100 = 4%. That is the trailing yield.

*Use the dividends already paid for a trailing yield. A forward yield uses the dividend the company has announced or is expected to pay.*

Try it: [Dividend Yield Calculator](https://toolcloak.pages.dev/tools/dividend-yield-calculator). Enter the price and yearly dividend to get yield, and how much income your holding would pay.

## Same dividend, different prices

The dividend is a rupee amount; the price changes every day. So the same ₹20 gives a different yield depending on when you look. This is why quoting a yield without the price date can mislead.

| Share price | Yearly dividend | Yield |
| --- | --- | --- |
| ₹500 | ₹20 | 4.00% |
| ₹400 | ₹20 | 5.00% |
| ₹300 | ₹20 | 6.67% |
| ₹250 | ₹20 | 8.00% |
| ₹200 | ₹20 | 10.00% |
| ₹160 | ₹20 | 12.50% |

*The company did not become more generous as the price fell. The same cash simply became a bigger slice of a smaller price.*

**Yield on the same ₹20 dividend as the price falls (%)**

|  | Value |
| --- | --- |
| ₹500 | 4% |
| ₹400 | 5% |
| ₹300 | 6.67% |
| ₹250 | 8% |
| ₹200 | 10% |
| ₹160 | 12.5% |

*A very high yield can mean the market expects the dividend to be cut.*

## The yield trap

A **yield trap** is a stock whose yield looks great only because its price has dropped, often because the business is struggling. If profits fall below the dividend, the company may cut or stop the payout, and the price can fall again. Buyers who chased the yield then lose twice: less income and a lower price.

**Two stocks, both showing a 8% yield**

| Healthy payer | Yield trap |
| --- | --- |
| Yield 8% is a surprise | Yield 8% is a warning |
| Profits comfortably cover the dividend | Price has fallen a lot in a short time |
| Dividend has grown or held for years | Profits are below the dividend |
| Debt is manageable | Heavy debt or shrinking sales |

*The number alone cannot tell them apart. You need to look at profits, debt and the dividend record.*

## Payout ratio: can the dividend last?

The **payout ratio** is the dividend per share divided by earnings per share. A company that earns ₹44 per share and pays ₹20 has a payout ratio of about 45%. Under 100% means it pays out less than it earns; over 100% means it is paying from savings or borrowing.

**Payout ratio of 45% (dividend ₹20, earnings ₹44 per share)**

| Zone | Range |
| --- | --- |
| Retains most | under 30% |
| Balanced | 30–60% |
| High | 60–80% |
| Stretched | over 80% |

*These bands are a rough rule of thumb, not an official standard. Banks, utilities and growth businesses normally sit in different places.*

## Yield on cost: your yield versus the market's yield

**Current yield** uses today's price. **Yield on cost** uses the price you actually paid. Suppose you bought at ₹200 when the dividend was ₹8 (4%). If the dividend grows 8% a year, then after 10 years it is about ₹17.27 a year. On your original ₹200 that is an 8.6% yield on cost, even if the share price has moved to ₹600 and the current yield is only 2.9%.

**Yearly dividend from one share bought at ₹200 (₹8 growing 8% a year)**

| Years after buying | Dividend per share (₹) |
| --- | --- |
| 0 | 8 |
| 2 | 9.33 |
| 4 | 10.88 |
| 6 | 12.69 |
| 8 | 14.81 |
| 10 | 17.27 |

*Yield on cost = dividend ÷ price paid: 4.0% at the start, 5.9% after 5 years, 8.6% after 10. It is a way to see income growth, but it does not show whether the share was a good use of money.*

## Reinvest or spend? A 20-year example

Take ₹1,00,000 in a share with a 4% yield, where the price and dividend both grow 6% a year (so the yield stays about 4%). Compare keeping the dividends as cash with buying more shares each year. These are made-up round numbers to show the effect, not a forecast.

**₹1,00,000 over 20 years: dividends spent against reinvested**

| Years | Dividends kept as cash (share value + cash received) | Dividends reinvested |
| --- | --- | --- |
| 0 | 100000 | 100000 |
| 5 | 156371 | 161051 |
| 10 | 231808 | 259374 |
| 15 | 332760 | 417725 |
| 20 | 467856 | 672750 |

*Reinvesting makes the total about ₹6.73 lakh against ₹4.68 lakh. The difference grows because reinvested dividends earn their own dividends and price growth.*

**Where the ₹4.68 lakh came from when dividends were not reinvested**

|  | Share value after 20 years | Dividends received |
| --- | --- | --- |
| After 20 years | ₹3.21 lakh | ₹1.47 lakh |

*Without reinvesting, the shares are worth about ₹3.21 lakh and the cash received adds up to about ₹1.47 lakh. Taxes are ignored here.*

## Total return: dividends are only half the story

**Where a 10% yearly total return came from in the example**

|  | Share |
| --- | --- |
| Price growth | 6% |
| Dividend yield | 4% |

*Total return = price change + dividends. A stock with a 6% yield and a falling price can lose to a 1% yield stock whose price climbs.*

## What to check before trusting a yield

1. **Date of the price.** Yield changes daily; compare at the same time.
2. **Payout ratio.** Is the dividend comfortably below profit?
3. **Dividend history.** Steady or growing payments are a better sign than one big special dividend.
4. **Debt and cash flow.** A company with heavy debt may cut the dividend first.
5. **Tax.** In India, dividends are added to your income and taxed at your slab rate, and the company may deduct TDS above a yearly limit. Check the current rules or ask a tax professional.

> **Education, not advice:** This page explains how yield works. It does not say any share is worth buying. Prices can fall, dividends can be cut, and past dividends do not guarantee future ones. Talk to a qualified adviser before investing.

## Frequently asked questions

### What is dividend yield?

It is the yearly dividend per share divided by the share price, shown as a percentage. A ₹20 yearly dividend on a ₹500 share is a 4% yield.

### Is a high dividend yield good?

Not always. A high yield can mean the price has fallen because the market expects a dividend cut. Check the payout ratio, debt and dividend history before drawing a conclusion.

### What is a good payout ratio?

There is no single answer. Many investors treat 30 to 60 percent as balanced for ordinary companies, but banks, utilities and fast-growing firms differ. A ratio above 100 percent means the dividend is not covered by profit.

### What is yield on cost?

It is the yearly dividend divided by the price you originally paid. It shows how your income has grown relative to your own investment, not what the stock yields to a new buyer today.

### Why did my stock's yield change when it paid the same dividend?

Because the price changed. Yield is dividend divided by price, so a lower price raises the yield and a higher price lowers it.

### Are dividends taxed in India?

Yes. Dividends are taxable income in the hands of the shareholder at their slab rate, and TDS may apply above a yearly threshold per company. Rules and limits change, so check the latest guidance.



---

# Down payment calculator: how much to put down on a home and what it does to your EMI and total interest

> The down payment decides how much you borrow, which sets the EMI and the interest you pay over 20 years. This guide shows the numbers for 10% to 40% down, the rules lenders follow, and the other costs on top, so you can choose an amount you can live with.

Published 2026-10-09 by Toolcloak. 3 minute read. Original: https://toolcloak.pages.dev/articles/down-payment-calculator-how-much-to-put-down-on-a-home-and-what-it-does-to-your-emi-and-total-interest
Related free tool: [Down Payment Calculator (Home & Car)](https://toolcloak.pages.dev/tools/down-payment-calculator)

The **down payment** is the part of the price you pay from your own money on day one. The rest comes from the loan. A bigger down payment means a smaller loan, a lower EMI and much less interest, but it also uses up cash that you may need for emergencies, furnishing or other goals. The decision is a balance, and the calculator shows both sides.

> **In short**
> - **Loan = price × (1 − down payment%).** On a ₹50 lakh home with 20% down, you pay **₹10,00,000** now and borrow **₹40,00,000**.
> - At 8.75% for 20 years the EMI is about **₹35,348**, total interest **₹44.84 lakh** and total cost **₹94.84 lakh**.
> - Lenders cap the **loan-to-value (LTV)** ratio, so there is a minimum down payment. Check your lender's current rule.
> - Registration, stamp duty, brokerage and furnishing come **on top** of the down payment.

## How the numbers are worked out

**Example: ₹50 lakh home, 20% down, 8.75% for 20 years**

1. **Work out the down payment.** ₹50 lakh × 20% = **₹10,00,000**.
2. **Find the loan.** ₹50 lakh − ₹10,00,000 = **₹40,00,000**.
3. **Calculate the EMI.** Loan ₹40,00,000 at 8.75% for 240 months gives **₹35,348** a month.
4. **Add up the totals.** EMI × 240 = ₹84.84 lakh. Interest = **₹44.84 lakh**. Total cost with the down payment = **₹94.84 lakh**.

*The tool returns the down payment, loan amount, monthly EMI, total interest and total cost.*

Try it: [Down Payment Calculator (Home & Car)](https://toolcloak.pages.dev/tools/down-payment-calculator). Enter the price, the down payment percentage, the interest rate and the tenure to see the loan, EMI and total interest.

## What each down payment does

| Down payment | Paid now | Loan amount | Monthly EMI | Total interest |
| --- | --- | --- | --- | --- |
| 10% | ₹5,00,000 | ₹45,00,000 | ₹39,767 | ₹50.44 lakh |
| 15% | ₹7,50,000 | ₹42,50,000 | ₹37,558 | ₹47.64 lakh |
| 20% | ₹10,00,000 | ₹40,00,000 | ₹35,348 | ₹44.84 lakh |
| 30% | ₹15,00,000 | ₹35,00,000 | ₹30,930 | ₹39.23 lakh |
| 40% | ₹20,00,000 | ₹30,00,000 | ₹26,511 | ₹33.63 lakh |

*All rows use a ₹50 lakh price, 8.75% interest and 20 years. A bigger down payment cuts the EMI and the total interest in direct proportion to the loan.*

**Total interest by down payment (₹ lakh)**

|  | Value |
| --- | --- |
| 10% | ₹50.4 L |
| 20% | ₹44.8 L |
| 30% | ₹39.2 L |
| 40% | ₹33.6 L |

*Moving from 20% to 30% down saves about ₹5.60 lakh of interest and lowers the EMI by about ₹4,419 a month.*

## Lender rules set the minimum

Lenders cannot finance the whole price. The Reserve Bank of India's loan-to-value limits for home loans have been tiered by loan size: up to 90% for loans of ₹30 lakh or less, up to 80% above that and up to ₹75 lakh, and up to 75% for larger loans. In plain terms that is a minimum down payment of roughly 10%, 20% or 25%. Check the current rule with your lender, because limits can change.

**Minimum down payment by loan size (rough, check with your lender)**

|  | Value | Note |
| --- | --- | --- |
| Loan up to ₹30 lakh | 10% | Up to 90% financed |
| ₹30 lakh to ₹75 lakh | 20% | Up to 80% financed |
| Above ₹75 lakh | 25% | Up to 75% financed |

*Banks may ask for more depending on your income, credit score and the property. These are upper limits on what they can lend, not promises.*

> **The down payment is not the only cash you need:** Stamp duty, registration, legal checks, brokerage, society transfer fees, loan processing fees and moving or furnishing costs are separate. They can add several percent of the price, so keep a cash buffer beyond the down payment.

## Putting more down or investing the difference

Paying an extra ₹10 lakh down reduces your loan interest by ₹5.60 lakh over 20 years at 8.75%. Investing the same ₹10 lakh at an assumed 10% a year would grow to about ₹67.3 lakh in 20 years, but that is not guaranteed and is taxed, while the interest you avoid is a certain saving. A loan rate close to or above your realistic investment return favours a larger down payment. A much lower loan rate favours keeping cash invested, as long as you keep an emergency fund.

**Extra ₹10 lakh: down payment or invest?**

| Put it into the down payment | Keep it invested |
| --- | --- |
| Saves ₹5.60 lakh | Grows to about ₹67.3 lakh |
| EMI falls by about ₹4,419 a month | Assumes 10% a year, not guaranteed |
| A sure saving at the loan rate | Stays available in emergencies |
| Cash is locked in the property | Returns are taxed and can fall |

*Neither is always better. A certain saving is worth more than an uncertain gain of the same size.*

## Choosing your own number

1. **Find the minimum** your lender requires for the loan size you need.
2. **Keep an emergency fund** of at least six months of expenses outside the down payment.
3. **Add the one-time costs** such as stamp duty and registration to the cash you need.
4. **Check the EMI** against your income. Many lenders look at EMIs of up to about 40% to 50% of take-home pay.
5. **Try three down payments** in the calculator and compare the EMI and total interest.

## Frequently asked questions

### How much down payment should I make on a home loan?

At least the minimum your lender requires, often 10% to 25% depending on the loan size. A larger down payment lowers the EMI and the total interest. Keep enough cash for emergencies and one-time costs.

### How do I calculate the loan amount from the down payment?

Multiply the price by the down payment percentage, then subtract from the price. For a 50 lakh home with 20% down, the down payment is 10 lakh and the loan is 40 lakh.

### How does the down payment affect my EMI?

The EMI is proportional to the loan amount at a fixed rate and tenure. Cutting the loan by 25%, for example by raising the down payment from 20% to 40%, cuts the EMI by 25% too.

### Does the down payment include stamp duty and registration?

No. Those are separate charges paid on top of the down payment, and the loan usually does not cover them.

### Is it better to pay a bigger down payment or invest the money?

It depends on the loan rate, your expected returns, taxes and how much cash you need to keep. A bigger down payment is a certain saving at the loan rate. Investment returns are not guaranteed.

### Can I use the calculator for a car or other loan?

Yes. Enter the price, the down payment percentage, the loan rate and the tenure. The [Down Payment Calculator](https://toolcloak.pages.dev/tools/down-payment-calculator) works for any loan with equal monthly instalments.



---

# EPF: how your provident fund grows over 30 years

> A few thousand rupees leave your salary every month and you barely feel it. Thirty years later it can be a crore or more. Here is what actually goes into your EPF, how it compounds, and the one habit that quietly costs people lakhs.

Published 2026-10-09 by Toolcloak. 3 minute read. Original: https://toolcloak.pages.dev/articles/epf-how-your-provident-fund-grows-over-30-years
Related free tool: [EPF Calculator](https://toolcloak.pages.dev/tools/epf-calculator)

If you are a salaried employee in India, a slice of your pay goes into the Employees' Provident Fund every month, matched by your employer. Most people never look at it until they change jobs. It is worth understanding, because it is one of the few savings plans that runs on autopilot and compounds for decades.

> **In short**
> - You put in **12% of basic pay**. Your employer also pays 12%, but part of that goes to the pension scheme (EPS), so less lands in your EPF.
> - On a ₹30,000 basic, about **₹5,950 a month** reaches the EPF. That is ₹71,400 a year.
> - At about 8.25% interest, 30 years can build around **₹94 lakh** with no raises, and about **₹1.9 crore** with 6% yearly raises.
> - Withdrawing at every job change breaks the compounding and can cut the final amount by two thirds.

## Where the money comes from

Your own contribution is 12 percent of your basic salary (plus dearness allowance, if any). The employer pays another 12 percent. Of the employer's 12 percent, 8.33 percent goes to the Employees' Pension Scheme (EPS), but the pension part is calculated only on wages up to ₹15,000, so it is capped at **₹1,250 a month**. The rest of the employer share goes to your EPF.

**The employer's share on a ₹30,000 basic (₹ per month)**

| Step | Amount |
| --- | --- |
| Employer 12% | ₹3,600 |
| To pension (EPS) | −₹1,250 |
| Into your EPF | ₹2,350 |

*Add your own 12 percent, which is ₹3,600, and ₹5,950 reaches your EPF each month. The pension part builds a separate monthly pension rather than a lump sum.*

**One month, start to finish**

1. **Salary is paid.** 12% of your basic is deducted before you see it.
2. **Employer adds its share.** Another 12%, split between EPF and the pension scheme.
3. **Balance earns interest.** Interest is worked out on your monthly running balance.
4. **Interest is credited.** The rate is announced by the EPFO each year and added to your account.

*The money sits in your account until you withdraw it or retire. It compounds without you doing anything.*

Try it: [EPF Calculator](https://toolcloak.pages.dev/tools/epf-calculator). Enter your basic salary, age and expected raises to estimate your EPF corpus at retirement.

## What 30 years can look like

Let us run two simple cases. In both, you start with nothing, earn a basic of ₹30,000, and the interest rate stays at 8.25 percent, the figure the EPFO declared for the years 2023-24 and 2024-25. Future rates will vary, so check the latest announcement.

**EPF balance over the years (lakh rupees)**

| Years of work | No raises | 6% raise every year |
| --- | --- | --- |
| 0 | 0 | 0 |
| 5 | 4.4 | 5.1 |
| 10 | 11.1 | 14.7 |
| 15 | 21.2 | 32 |
| 20 | 36.4 | 61.7 |
| 25 | 59.3 | 111.3 |
| 30 | 94 | 192.6 |

*Approximate figures from monthly compounding, which is close to but not exactly how the EPFO credits interest. The calculator's result may differ slightly. With raises, the contribution grows every year as well, so the curve bends upward much faster.*

**Where ₹94 lakh comes from (30 years, no raises)**

|  | Share |
| --- | --- |
| Contributions (you + employer) | ₹21.4 L |
| Interest earned over the years | ₹72.5 L |

*You and your employer put in about ₹21 lakh in total. The remaining ₹72 lakh is compounding. Most of that arrives in the last ten years.*

> **The last decade does the heavy lifting:** In the no-raise case, the balance at year 20 is about ₹36 lakh and at year 30 about ₹94 lakh. The final ten years add more than the first twenty years combined. This is why leaving the account alone, even for a few extra years, pays so well.

## How much does the interest rate matter?

**30-year balance at different rates (lakh rupees, no raises)**

|  | Balance |
| --- | --- |
| 7.0% | ₹73.0 L |
| 7.5% | ₹80.7 L |
| 8.0% | ₹89.3 L |
| 8.25% | ₹94.0 L |
| 9.0% | ₹109.7 L |

*Every half percent changes the 30-year result by roughly ₹8 to ₹10 lakh here. You do not control the rate, but it explains why the figure on your passbook can differ from an old estimate.*

## The habit that costs lakhs: withdrawing at each job change

When you change jobs it is tempting to withdraw the balance, because the money is right there. Let us compare two people with the same salary and the same 30 years of work, divided into three jobs of ten years each.

**Total at the end of 30 years (lakh rupees)**

|  | Total |
| --- | --- |
| Withdraws after each of 3 jobs | ₹33.3 L (three payouts) |
| Transfers the account and keeps it | ₹94.0 L |

*Three withdrawals of about ₹11.1 lakh each, spent along the way, against a single pot that kept compounding. The difference is about ₹60 lakh. Transferring your EPF to the new employer keeps it growing, and you can do it online with your UAN.*

> **Check the current rules before you act:** Withdrawal conditions, tax on withdrawals taken early, and tax on interest (interest on your own contributions above a yearly limit, currently ₹2.5 lakh, can be taxable) have changed over the years. Read the latest EPFO and income tax rules or ask a tax professional before you withdraw or put in extra money.

## Five simple moves

1. **Know your UAN** and keep it active. It follows you across jobs.
2. **Transfer, do not withdraw,** when you change employers.
3. **Check your passbook** once a year. Make sure every month's contribution is there.
4. **Think about VPF** (voluntary provident fund) if you want to put in more than 12 percent. It earns the same rate but is locked in like EPF, so be sure you will not need the money soon.
5. **Run an estimate** at least once, so the future number feels real.

## Frequently asked questions

### How much is deducted for EPF from salary?

Your own contribution is 12 percent of basic salary plus dearness allowance. Your employer pays another 12 percent, of which up to ₹1,250 a month goes to the pension scheme and the rest to your EPF.

### What is the EPF interest rate?

The EPFO announces the rate every year. It was 8.25 percent for 2023-24 and for 2024-25. Check the latest announcement because the rate can change.

### How much will my EPF be after 30 years?

On a ₹30,000 basic with no raises and 8.25 percent interest, roughly ₹94 lakh. With 6 percent yearly raises, roughly ₹1.9 crore. These are estimates, not guarantees.

### Should I withdraw EPF when I change jobs?

Usually transferring is better, because the money keeps compounding. In a three-job, 30-year example, withdrawing each time left about ₹33 lakh of payouts against ₹94 lakh from keeping it.

### What is VPF?

VPF stands for voluntary provident fund. It lets you contribute more than the standard 12 percent to your EPF account. It earns the same interest rate, but the money stays locked in like regular EPF.



---

# EV vs petrol savings calculator: running cost per km and how long the extra price takes to pay back

> An electric car costs more to buy and far less to run. This guide works out petrol and electric cost per km, monthly and yearly savings and the payback time on the extra price, and shows how distance driven and home or public charging change the answer.

Published 2026-10-09 by Toolcloak. 3 minute read. Original: https://toolcloak.pages.dev/articles/ev-vs-petrol-savings-calculator-running-cost-per-km-and-how-long-the-extra-price-takes-to-pay-back
Related free tool: [EV vs Petrol Car Running Cost Calculator](https://toolcloak.pages.dev/tools/ev-vs-petrol-savings-calculator)

The headline difference between an electric car and a petrol one is simple: the EV costs more to buy and less to run. Whether it saves you money depends on how far you drive, what your petrol and electricity cost, and how long you keep the car. A short calculation tells you roughly when the extra price is recovered.

> **In short**
> - **Petrol cost per km = price per litre ÷ km per litre.** At ₹105 and 15 km/l that is **₹7.00 per km**.
> - **EV cost per km = price per kWh ÷ km per kWh.** At ₹8 and 7 km/kWh that is **₹1.14 per km**.
> - Driving 1,000 km a month, the car costs **₹7,000** in petrol or **₹1,143** in electricity, a saving of **₹5,857 a month** or **₹70,286 a year**.
> - An extra price of ₹4 lakh is recovered in about **5.7 years**.

## How the calculation works

**Example: 1,000 km a month**

1. **Petrol cost.** 1,000 ÷ 15 = 66.7 litres × ₹105 = **₹7,000**.
2. **Electricity cost.** 1,000 ÷ 7 = 142.9 kWh × ₹8 = **₹1,143**.
3. **Monthly saving.** ₹7,000 − ₹1,143 = **₹5,857**.
4. **Payback on the extra price.** ₹4,00,000 ÷ (₹5,857 × 12) = **5.7 years**.

*The tool also shows the yearly saving and the cost per km for both fuels.*

Try it: [EV vs Petrol Car Running Cost Calculator](https://toolcloak.pages.dev/tools/ev-vs-petrol-savings-calculator). Enter your distance, fuel prices, mileage and the extra price of the EV to see savings and payback.

## Cost per km, side by side

**Running cost of one kilometre**

| Petrol car | Electric car (home charging) |
| --- | --- |
| ₹7.00 per km | ₹1.14 per km |
| ₹105 a litre | ₹8 per kWh |
| 15 km per litre | 7 km per kWh |
| A fuel price rise hits every km | About one sixth of the petrol cost |

*At these prices the EV costs 16% as much per kilometre. Petrol costs 6.1 times more.*

## Distance changes everything

The saving grows with every kilometre, but the extra purchase price does not change. So the more you drive, the faster it pays back. A car that only does a few trips a week may never recover the extra cost before you sell it.

| Distance per month | Petrol cost | EV cost | Saving per month | Payback on ₹4 lakh extra |
| --- | --- | --- | --- | --- |
| 500 km | ₹3,500 | ₹571 | ₹2,929 | 11.4 years |
| 1,000 km | ₹7,000 | ₹1,143 | ₹5,857 | 5.7 years |
| 1,500 km | ₹10,500 | ₹1,714 | ₹8,786 | 3.8 years |
| 2,000 km | ₹14,000 | ₹2,286 | ₹11,714 | 2.8 years |

*Same prices as the example. Doubling the distance halves the payback time.*

**Payback time on the extra price by monthly distance (years)**

|  | Value |
| --- | --- |
| 500 km | 11.4 yrs |
| 1,000 km | 5.7 yrs |
| 1,500 km | 3.8 yrs |
| 2,000 km | 2.8 yrs |

*At 500 km a month the extra price takes over eleven years to recover. At 2,000 km it takes under three.*

## Home charging against public charging

The example uses ₹8 per kWh, the kind of rate you might pay for electricity at home. Public fast chargers are often priced much higher. At ₹18 per kWh the same 1,000 km costs ₹2,571 instead of ₹1,143, which cuts the monthly saving to ₹4,429 and stretches the payback to 7.5 years.

**Home against public charging at 1,000 km a month**

|  | Value | Note |
| --- | --- | --- |
| Home at ₹8 per kWh | ₹1,143 | Monthly cost |
| Public at ₹18 per kWh | ₹2,571 | Monthly cost |
| Saving with home charging | ₹5,857 | Against petrol |
| Saving with public charging | ₹4,429 | Against petrol |

*Even at the higher public price the EV is cheaper to run than petrol here, but the payback is longer. Use your own mix of home and public charging for the best estimate.*

## What the calculation leaves out

- **Insurance, tax and registration,** which can differ between EV and petrol cars.
- **Maintenance,** usually lower for an EV, with no engine oil or exhaust, but tyres and brakes still wear.
- **Battery warranty and replacement,** and how the battery ages over the years you plan to keep the car.
- **Resale value,** which depends on the model and how the market treats used EVs.
- **Subsidies or tax benefits,** which vary by state and time and can reduce the extra price.
- **Fuel price changes,** in either direction, over the years you keep the car.

> **Use real-world efficiency:** The mileage printed in a brochure is a test figure. Real-world petrol mileage and EV efficiency are lower, especially in city traffic and with the air conditioner on. Use the average from your own trip meter if you can.

## Doing the comparison properly

1. **Find your real monthly distance** from the odometer or trip meter, not a guess.
2. **Use your real fuel and electricity prices,** and the split between home and public charging.
3. **Enter the price gap** between the two cars you are comparing, after any subsidy.
4. **Check the payback** against how many years you plan to keep the car.
5. **Run a second case** with higher charging costs or lower distance to see the downside.

## Frequently asked questions

### How do I calculate the running cost per km of a petrol car?

Divide the petrol price per litre by the mileage in km per litre. At 105 rupees a litre and 15 km per litre, the cost is 7 rupees per km.

### How do I calculate the running cost per km of an electric car?

Divide the electricity price per kWh by the efficiency in km per kWh. At 8 rupees per kWh and 7 km per kWh, the cost is about 1.14 rupees per km.

### How long does it take for an EV to pay back its extra price?

Divide the extra price by the yearly fuel saving. With an extra price of 4 lakh rupees and a saving of about 70,286 rupees a year, payback is around 5.7 years.

### Does charging at a public station change the result?

Yes. Public fast chargers usually cost more per kWh than home electricity, so the saving is smaller and the payback longer. Enter a blended price if you use both.

### Does the calculator include maintenance and insurance?

No. It compares fuel and electricity cost and the extra purchase price only. Add insurance, maintenance, battery and resale differences separately if you want a full cost of ownership.

### What if I drive very few kilometres?

Then the saving is small and the extra price may never be recovered while you own the car. Try different distances in the [EV vs Petrol Savings Calculator](https://toolcloak.pages.dev/tools/ev-vs-petrol-savings-calculator) to see how sensitive the payback is.



---

# Your photo may be carrying your home address. Here is how to check.

> Every phone photo can hide the camera model, the exact time and, if location is on, GPS coordinates accurate to a few metres. What EXIF is, how to read it, when it travels with the picture, and how to remove it before you share.

Published 2026-10-09 by Toolcloak. 3 minute read. Original: https://toolcloak.pages.dev/articles/your-photo-may-be-carrying-your-home-address
Related free tool: [Image EXIF Viewer](https://toolcloak.pages.dev/tools/image-exif-viewer)

You take a photo of your new bicycle outside your front door and send it to a stranger who wants to buy it. The photo looks harmless. The file may contain something the picture does not show: the **latitude and longitude of where you were standing**.

This is not a hack and it is not rare. It is a standard feature of how cameras save files. Most people never see it, which is exactly why it is worth ten minutes of attention.

## What EXIF actually is

EXIF stands for exchangeable image file format. It is a block of small notes the camera writes inside the photo file, next to the pixels. Opening the picture shows only the pixels. The notes sit behind them until something reads them.

| Field | Example | Why it matters |
| --- | --- | --- |
| Camera or phone model | Pixel 8, iPhone 15 | Shows what device you own |
| Date and time taken | 2026-10-04 18:42 | Shows when you were there, to the minute |
| Exposure settings | 1/120 s, f/1.8, ISO 100 | Harmless, useful for photographers |
| Orientation | Rotated 90° | Tells apps which way is up |
| GPS position | Latitude and longitude | Shows where you were, to a few metres |
| Software | Editor name and version | Shows what you edited it with |

*Typical fields in a phone photo. GPS is only present when the camera app has permission to use your location.*

## How precise is the location?

Coordinates are written as decimal degrees, and each extra decimal place sharpens them by a factor of ten. A change of 0.0001 degrees of latitude is about 11 metres on the ground.

**How small an area each coordinate precision points to**

|  | Metres |
| --- | --- |
| 2 decimals | ≈ 1.1 km |
| 3 decimals | ≈ 111 m |
| 4 decimals | ≈ 11 m |
| 5 decimals | ≈ 1 m |

*Phone photos usually record several decimal places, so the point on the map can be a particular house or doorway, not just a neighbourhood.*

A phone with a good view of the sky typically fixes its position within a few metres. That is enough to tell your house from your neighbour's.

## When the data travels, and when it does not

Whether EXIF survives depends on how the picture is shared. Many large social apps remove location from what other people can see when they display a post, but this is a policy of each app, it can change, and it does not cover every way of sharing. Treat it as something to check, not something to assume.

| How you share | Does the original file go? | Usual risk |
| --- | --- | --- |
| Email attachment | Yes, the file as it is | Metadata travels |
| Messaging app, sent as file or document | Yes, usually | Metadata travels |
| Messaging app, sent as a photo | Often re-compressed | Varies by app |
| Upload to your own website or a marketplace | Often yes | Check first |
| Screenshot of the photo | No, it is a new image | Original EXIF is gone |

*A general guide, not a promise. Apps change how they handle photos, so check a sample before you rely on it.*

> **The risky ones:** Selling things online, posting photos of your home, your children's school, your car in front of your gate, a picture sent to someone you do not know. These are the moments when a precise location and a timestamp together tell a stranger where you live and when you are not there.

## Check a photo in three steps

**Look inside a photo before you share it**

1. **Open it in the viewer.** Choose the photo. Nothing is uploaded: the file is read in your browser.
2. **Look for GPS.** If you see latitude and longitude, the photo knows where it was taken.
3. **Remove it and save a clean copy.** Save a version without the metadata and share that one instead of the original.

Try it: [Image EXIF Viewer](https://toolcloak.pages.dev/tools/image-exif-viewer). Drop a photo in to see the camera, the time and the GPS position it carries, and save a copy with the metadata removed.

## Turn the problem off at the source

- In your camera app's settings, switch off location tagging, or give the camera app location access only when it is needed.
- For one-off sharing, send a clean copy made with an EXIF remover instead of the original.
- Keep the original for yourself. The metadata is useful when you search your own library by place or date.
- Remember that removing the data from a photo does not remove what the picture shows: a street sign or house number is still there.

Removing metadata also goes well with shrinking the photo. A smaller picture shared online loads faster, and [here is why a 4 MB photo is often really a 400 KB one](https://toolcloak.pages.dev/articles/why-your-4mb-photo-is-really-a-400kb-photo).

## Frequently asked questions

### What is EXIF data?

A set of notes stored inside a photo file: camera model, date and time, exposure settings and, if location was on, GPS coordinates. It is invisible when you look at the picture but can be read by software.

### Does every photo contain my location?

Only if the camera app was allowed to use your location when the photo was taken. Photos from the web, screenshots and many edited images carry no GPS data.

### How accurate is the GPS in a photo?

Usually within a few metres on a phone with a clear view of the sky. Each decimal place in the coordinates is ten times more precise, and five decimals point to about one metre.

### Does taking a screenshot remove EXIF?

A screenshot is a new image, so the original photo's camera and location notes are not copied into it. It also reduces quality, so for a clean copy of the real photo, an EXIF remover is better.

### Is removing EXIF safe for the picture?

Yes. The pixels are not changed. Only the hidden notes are dropped, so the photo looks the same.



---

# Final grade calculator: how to find the score you need on the final exam to reach your target grade

> Before an exam, the useful question is not what your grade is but what you need to score to reach the grade you want. This guide shows the formula, how the exam's weight changes the answer, and what to do when the number needed is above 100%.

Published 2026-10-09 by Toolcloak. 3 minute read. Original: https://toolcloak.pages.dev/articles/final-grade-calculator-how-to-find-the-score-you-need-on-the-final-exam-to-reach-your-target-grade
Related free tool: [Final Grade Calculator](https://toolcloak.pages.dev/tools/final-grade-calculator)

Your course grade is a weighted mix of everything you have done and the final exam still to come. If you know your **current grade**, the **weight of the final** and the **grade you want**, you can work out the exact score you need on the exam. It is a short calculation, and it turns a vague worry into a clear target.

> **In short**
> - **Score needed = (goal − current × (1 − weight)) ÷ weight,** with the weight written as a decimal.
> - With a current grade of **82%**, a final worth **30%** and a goal of **85%**, you need **92.0%** on the final.
> - If the answer is above 100%, the goal cannot be reached without extra credit. If it is 0 or below, you already have the grade.
> - The more the final is worth, the more room you have to move your grade, up or down.

## Where the formula comes from

Your final course grade is the grade you have now, counted for the share of the course already finished, plus your exam score counted for the share the exam is worth. In symbols: goal = current × (1 − weight) + exam × weight. Rearranging to find the exam score gives the formula above.

**Example: current 82%, final worth 30%, goal 85%**

1. **Find the weight kept.** 100% − 30% = 70%, or **0.70** as a decimal.
2. **Count what you have already earned.** 82 × 0.70 = **57.4** points of the final grade.
3. **Find the gap to the goal.** 85 − 57.4 = **27.6** points still needed.
4. **Divide by the exam's weight.** 27.6 ÷ 0.30 = **92.0%** needed on the final.

*The tool shows the score needed, a status (achievable, not possible, or already reached) and the share of the grade you keep.*

Try it: [Final Grade Calculator](https://toolcloak.pages.dev/tools/final-grade-calculator). Enter your current grade, the weight of the final and the grade you want to see the score you need.

## How the target changes the score

| Grade you want | Score needed on the final | Status |
| --- | --- | --- |
| 80% | 75.3% | Achievable |
| 85% | 92.0% | Achievable |
| 90% | 108.7% | Not possible without extra credit |

*Same current grade of 82% and a final worth 30%. Each extra 5 points of goal needs about 16.7 more points on the exam, because the exam counts for only 30% of the course.*

**Score needed on the final for a goal of 85%**

| Zone | From | To |
| --- | --- | --- |
| Possible | 60 | 100 |
| Above 100% | 100 | 120 |

*A goal of 90% would need about 108.7%, which is not possible without extra credit.*

## The weight of the final matters

A final worth 50% of the course gives you a lot of room. A final worth 20% gives you very little, and a high goal may be out of reach even with a perfect score. Check the syllabus for the exact weight before you plan.

| Final exam weight | Score needed for a goal of 85% |
| --- | --- |
| 20% | 97.0% |
| 30% | 92.0% |
| 40% | 89.5% |
| 50% | 88.0% |

*Same current grade of 82%. When the final is worth less, you need a higher score to lift your grade by the same amount.*

**Score needed by exam weight (%)**

|  | Value |
| --- | --- |
| 20% | 97% |
| 30% | 92% |
| 40% | 89.5% |
| 50% | 88% |

*A heavier final lowers the score you need when your goal is above your current grade.*

## And your current grade matters too

| Current grade | Score needed for a goal of 85% |
| --- | --- |
| 70% | 120.0% |
| 76% | 106.0% |
| 82% | 92.0% |
| 88% | 78.0% |
| 94% | 64.0% |

*Final worth 30%. A current grade already above the goal would give a score below the goal.*

> **Make sure the current grade is the right number:** Use your grade for the work counted so far, as a percentage of the marks that were available, not the total marks including the exam. If your course drops the lowest score or uses categories with their own weights, work out the current grade the same way your teacher does first.

## What if the answer is above 100%?

- **Check the numbers.** A wrong weight is the commonest cause.
- **Ask about extra credit,** re-submission or a dropped lowest score.
- **Set a nearer goal,** such as the next lower grade boundary, and find the score for that.
- **Aim for the best score you can** and plan the next course differently.

## A study plan from the number

1. **Find the score needed** with the calculator.
2. **Compare it with your past exam scores** in this course. A target far above them needs a plan, not hope.
3. **List the topics** with the most marks and the weakest results.
4. **Work backwards from the exam date** to give each topic time. The [Days Until Date Calculator](https://toolcloak.pages.dev/tools/days-until-date-calculator) shows how many days you have.
5. **Re-check after any mock test** with the new current grade if marks are still being added.

## Frequently asked questions

### How do I calculate the score I need on my final exam?

Subtract your current grade times the share already finished from your goal, then divide by the exam's weight. For a current 82%, a final worth 30% and a goal of 85%, you need 92%.

### What does the weight of the final mean?

It is the share of the course grade that the final exam decides. A weight of 30 means the exam is 30% of the course and everything before it is the other 70%.

### What does it mean if the score needed is above 100%?

The goal cannot be reached by the exam alone. You would need extra credit or a lower goal.

### What if the score needed is zero or negative?

You already have the grade you want, even with a score of zero. The tool shows 0% and says you already have it.

### Does this work for a midterm or any other single assessment?

Yes. Enter the weight of that assessment as the final's weight and your grade so far for the rest. The formula is the same.

### Is this the same as a GPA calculator?

No. This finds a score inside one course. To combine several courses into a grade point average, use the [GPA Calculator](https://toolcloak.pages.dev/tools/gpa-calculator).



---

# Freelance rate calculator: how to set an hourly rate that covers tax, expenses and time off

> Dividing your wished-for salary by 2,080 working hours gives a rate that is far too low for a freelancer. This guide builds the rate from take-home goal, expenses, tax, billable hours and weeks off, with tables showing how each choice changes the number.

Published 2026-10-09 by Toolcloak. 3 minute read. Original: https://toolcloak.pages.dev/articles/freelance-rate-calculator-how-to-set-an-hourly-rate-that-covers-tax-expenses-and-time-off
Related free tool: [Freelance Hourly Rate Calculator](https://toolcloak.pages.dev/tools/freelance-rate-calculator)

A **freelance rate** has to pay for much more than your time. It covers your take-home income, your business costs, your tax, and the many hours you spend finding clients, sending invoices and waiting for payments, which no one pays you for. Start from what you need, not from what you think the market will pay, and the right number follows.

> **In short**
> - **Yearly billing needed = (take-home goal + expenses) ÷ (1 − tax rate).**
> - **Hourly rate = yearly billing ÷ (billable hours per week × weeks worked).**
> - A goal of **₹12 lakh** take-home, **₹1,20,000** expenses, 15% tax, 25 billable hours and 6 weeks off gives **₹1,350 an hour**, **₹10,803 a day** and **₹1,29,412 a month**.
> - Billable hours are fewer than working hours. Plan for 20 to 30 a week, not 40.

## The formula, step by step

**Example: ₹12 lakh take-home goal**

1. **Add expenses to the goal.** ₹12,00,000 + ₹1,20,000 = **₹13,20,000** needed after tax.
2. **Gross it up for tax.** ₹13,20,000 ÷ (1 − 0.15) = **₹15,52,941** to bill.
3. **Count the billable hours.** 25 hours × (52 − 6) weeks = **1,150** hours a year.
4. **Divide.** ₹15,52,941 ÷ 1,150 = **₹1,350** an hour.

*The tool also shows the day rate (8 hours), the yearly billing needed (₹15,52,941), the billable hours per year (1,150) and the monthly billing (₹1,29,412).*

Try it: [Freelance Hourly Rate Calculator](https://toolcloak.pages.dev/tools/freelance-rate-calculator). Enter your take-home goal, expenses, tax rate, billable hours and weeks off to get an hourly and day rate.

## Why billable hours are not working hours

A freelancer spends time on sales, emails, accounting, learning, revisions and unpaid calls. Out of a 40-hour week, only 20 to 30 hours can usually be billed to clients. Using 40 hours in the calculation gives a rate that looks competitive but leaves you short every month.

| Billable hours per week | Billable hours per year | Hourly rate needed |
| --- | --- | --- |
| 15 | 690 | ₹2,251 |
| 20 | 920 | ₹1,688 |
| 25 | 1,150 | ₹1,350 |
| 30 | 1,380 | ₹1,125 |
| 35 | 1,610 | ₹965 |

*Same goal, expenses, tax and 6 weeks off. Dropping from 35 billable hours to 15 more than doubles the hourly rate you need.*

**Hourly rate needed by billable hours per week (₹)**

|  | Hourly rate |
| --- | --- |
| 15 hours | ₹2,251 |
| 20 hours | ₹1,688 |
| 25 hours | ₹1,350 |
| 30 hours | ₹1,125 |
| 35 hours | ₹965 |

*The rate falls as you bill more hours, but only if the work is steady. A lower rate with unfilled weeks is the worst of both.*

## Time off is part of the price

An employee gets paid holidays and sick leave. A freelancer pays for them through the rate. Every week off is a week with no income, so it must be recovered in the other weeks.

| Weeks off per year | Billable hours per year | Hourly rate needed |
| --- | --- | --- |
| 2 | 1,250 | ₹1,242 |
| 4 | 1,200 | ₹1,294 |
| 6 | 1,150 | ₹1,350 |
| 8 | 1,100 | ₹1,412 |
| 10 | 1,050 | ₹1,479 |

*All rows assume 25 billable hours a week. Each extra week off adds about ₹30 to the hourly rate here.*

## The target income changes the rate

| Take-home goal | Hourly rate | Day rate (8 hours) |
| --- | --- | --- |
| ₹6,00,000 | ₹737 | ₹5,893 |
| ₹9,00,000 | ₹1,043 | ₹8,348 |
| ₹12,00,000 | ₹1,350 | ₹10,803 |
| ₹18,00,000 | ₹1,964 | ₹15,714 |
| ₹24,00,000 | ₹2,578 | ₹20,624 |

*Same expenses, 15% tax, 25 billable hours and 6 weeks off. Doubling the goal does not exactly double the rate because the expenses stay fixed.*

**What the example rate means in practice**

|  | Value | Note |
| --- | --- | --- |
| Hourly rate | ₹1,350 | 25 billable hours a week |
| Day rate | ₹10,803 | 8 hours |
| Monthly billing | ₹1,29,412 | Before tax and expenses |
| Billable hours a year | 1,150 | 25 × 46 weeks |

*A rate that looks high next to a salary-equivalent hourly figure is normal. It covers tax, equipment, software, idle time and the risk that work dries up.*

## A note on how tax is handled

The tool adds your expenses to your goal and then grosses up the whole amount for tax. In real accounts, tax is paid on profit, which is income minus expenses, so the exact figure is slightly lower. On these numbers the difference is about ₹21,176 a year. The tool's answer is a cautious one. Treat it as a floor, then check with an accountant, especially if you may pay GST or advance tax.

> **GST and other charges:** If you are registered for GST you collect it on top of your rate and pass it to the government. It is not extra income. Quote your rate before GST and add it as a separate line on the invoice.

## Project price against hourly price

**Two ways to quote the same job**

| Hourly | Fixed price |
| --- | --- |
| ₹1,350 an hour | Hours × rate + buffer |
| Fair when scope is unclear | Rewards speed and skill |
| Client sees the time cost | Needs a clear scope |
| Faster work earns less | Add 15% to 25% for revisions |

*Use your hourly rate to estimate the hours, then price the project. A fixed price should never come out lower than the hourly figure for the same work.*

## Setting your rate: a checklist

1. **Decide the take-home income** you need, after tax and after business costs.
2. **List yearly expenses:** software, equipment, internet, coworking, insurance, accounting.
3. **Choose a realistic number of billable hours,** usually 20 to 30 a week.
4. **Set weeks off honestly,** including holidays, illness and slow periods.
5. **Calculate, then round up** to a clean number, and revisit it once a year.

## Frequently asked questions

### How do I calculate my freelance hourly rate?

Add your yearly expenses to the take-home income you want, divide by one minus your tax rate, then divide by the billable hours you expect in a year (billable hours a week times weeks worked).

### How many billable hours should I assume?

Most freelancers can bill 20 to 30 hours a week. The rest goes to finding clients, admin, learning and unpaid revisions. Using 40 hours leads to a rate that is too low.

### Why is a freelance rate higher than an employee's hourly pay?

It must cover tax, equipment, software, unpaid time between projects, holidays and the risk of gaps between clients. An employee's employer bears many of these costs.

### What is a day rate?

A day rate is the hourly rate times the number of hours in a working day. The tool uses 8 hours. Many freelancers quote day rates for consulting and long projects.

### Does the calculator include GST?

No. GST is collected from the client on top of your rate and passed to the government, so it is not part of your income. Add it as a separate line on your invoice.

### Should I charge by the hour or by the project?

Use the hourly rate to estimate time, then choose. Hourly billing suits unclear scope. A fixed project price suits clear work, and should be at least the hourly figure. The [Freelance Rate Calculator](https://toolcloak.pages.dev/tools/freelance-rate-calculator) gives the hourly floor.



---

# GPA calculator: how to work out your grade point average, weighted by credits, and plan the next term

> A GPA is not the plain average of your grades. Each course counts in proportion to its credits. This guide shows the 4.0 scale, a worked example, the mistake of ignoring credits, how to combine a new term with your existing GPA and the grade you need to reach a target.

Published 2026-10-09 by Toolcloak. 3 minute read. Original: https://toolcloak.pages.dev/articles/gpa-calculator-how-to-work-out-your-grade-point-average-weighted-by-credits-and-plan-the-next-term
Related free tool: [GPA Calculator](https://toolcloak.pages.dev/tools/gpa-calculator)

Your **GPA** (grade point average) is a single number that summarises your grades. Each letter grade maps to **grade points**, and each course counts in proportion to its **credits**. A 4-credit course affects the result twice as much as a 2-credit one, which is why a plain average of grades is wrong.

> **In short**
> - **GPA = total grade points ÷ total credits,** where grade points for a course = points for the grade × credits.
> - For A in 4 credits, B+ in 3, A− in 3 and B in 2, total points are **43.0** over **12** credits, so the GPA is **3.58**.
> - A plain average of the four grade values gives **3.50**, which is wrong because it ignores credits.
> - To combine terms, add the points and credits of all of them and divide again. Do not average the term GPAs.

## The 4.0 scale used by the calculator

| Grade | Points |
| --- | --- |
| A+ | 4.0 |
| A | 4.0 |
| A- | 3.7 |
| B+ | 3.3 |
| B | 3.0 |
| B- | 2.7 |
| C+ | 2.3 |
| C | 2.0 |
| C- | 1.7 |
| D+ | 1.3 |
| D | 1.0 |
| D- | 0.7 |
| F | 0.0 |

*A+ is counted as 4.0 here. Some universities give an A+ more than 4.0 or use a different scale, and some do not use plus and minus grades at all. Follow your own institution's table.*

## A worked example

**One term with four courses**

1. **Convert each grade to points.** A = 4.0, B+ = 3.3, A− = 3.7, B = 3.0.
2. **Multiply by the credits.** A × 4 = **16.0** · B+ × 3 = **9.9** · A- × 3 = **11.1** · B × 2 = **6.0**
3. **Add the points and the credits.** 43.0 points over 12 credits.
4. **Divide.** 43.0 ÷ 12 = **3.58**.

*Type one course per line as the grade and then the credits, such as A 4. The tool shows each line's points, the GPA, the total credits and the number of courses.*

Try it: [GPA Calculator](https://toolcloak.pages.dev/tools/gpa-calculator). List your courses with a grade and credits on each line and see your GPA.

## Why credits matter

The weighted result of 3.58 is higher than the plain average of 3.50 because the best grades, A and A−, sit in the courses with more credits. If your highest grade were in a 1-credit course instead, the weighted GPA would drop.

**Grade points earned in each course (grade value × credits)**

|  | Points |
| --- | --- |
| A (4 cr) | 16.0 |
| B+ (3 cr) | 9.9 |
| A- (3 cr) | 11.1 |
| B (2 cr) | 6.0 |

*The 4-credit A contributes 16 of the 43 points, over a third. Heavier courses move your GPA more, up or down.*

**Weighted against plain average**

| Weighted by credits | Plain average |
| --- | --- |
| 3.58 | 3.50 |
| 43.0 points ÷ 12 credits | (4.0 + 3.3 + 3.7 + 3.0) ÷ 4 |
| What a transcript shows | Treats every course the same |
| Large courses count more | Not used by universities |

*Always include the credits. The gap here is small, but it can be large when credits differ a lot.*

## Combining this term with your GPA so far

Say you have a GPA of 3.4 over 60 credits. Your old points are 3.4 × 60 = 204. Add this term's 43 points to get 247, and divide by 72 credits. Your new cumulative GPA is **3.43**.

> **Do not average two GPAs:** Averaging 3.4 and 3.58 gives 3.49, but the real answer is 3.43. The old GPA covers five times as many credits as the new term, so it should carry five times the weight.

## What do I need next term?

To reach a target cumulative GPA, work out the total points you will need after the next term, subtract the points you already have, and divide by the credits of the next term. With 72 credits already counted at 3.43 and a further 12-credit term, the grade you need depends on the target.

| Target cumulative GPA | Term GPA needed over the next 12 credits |
| --- | --- |
| 3.40 | 3.22 |
| 3.45 | 3.57 |
| 3.50 | 3.92 |
| 3.55 | 4.27 |

*These figures start from 3.4 over 60 credits plus the 3.58 term above, which gives 3.43 over 72 credits. A required term GPA above 4.0 means the target cannot be reached in one term.*

**Term GPA needed for a 3.50 cumulative target**

| Zone | From | To |
| --- | --- | --- |
| Reachable | 3 | 4 |
| Not possible on a 4.0 scale | 4 | 4.5 |

*A cumulative target of 3.50 needs a term GPA of about 3.92, close to the top of the scale. A target of 3.55 would need more than 4.0, so it cannot be reached in one term.*

## GPA, CGPA and percentage

- **GPA** is usually on a 4.0 scale (US style). Some countries use 4.3, 5.0 or other scales.
- **CGPA** is the cumulative GPA across all terms, often on a 10-point scale in India. See the [CGPA to Percentage Converter](https://toolcloak.pages.dev/tools/cgpa-percentage-converter) for those conversions.
- **Percentage** needs your marks. A GPA hides the exact marks in each grade band.

## Using the calculator well

1. **Get the official grade table** from your institution, and note the points for plus and minus grades.
2. **List every course** for the term with its grade and credits.
3. **Check the result** against your transcript or portal.
4. **Add earlier terms** the way shown above to see the cumulative GPA.
5. **Plan the next term** by finding the grades you need for your target.

## Frequently asked questions

### How do I calculate my GPA?

Convert each grade to grade points, multiply by the course credits, add all the points, then divide by the total credits. For A in 4 credits, B+ in 3, A- in 3 and B in 2, the GPA is 43 ÷ 12 = 3.58.

### What is the difference between GPA and CGPA?

GPA usually refers to one term, while CGPA is the cumulative average across all terms. Both use credits as weights. Some universities use the two words interchangeably.

### How do I calculate a cumulative GPA?

Add the total grade points of all terms, add the total credits of all terms and divide. Do not average the GPAs of different terms, because terms can have different credit totals.

### Does the calculator use a 4.0 scale?

Yes. A and A+ count as 4.0, A- as 3.7, B+ as 3.3, B as 3.0 and so on down to F as 0. If your university uses a different scale, you can type the grade points directly instead of a letter.

### What GPA do I need next term to reach a target?

Multiply the target GPA by your total credits including next term, subtract your current points, and divide by next term's credits. If the result is above 4.0, the target is not reachable in one term.

### Is the GPA calculator an official result?

No. The [GPA Calculator](https://toolcloak.pages.dev/tools/gpa-calculator) is a quick check. Your official GPA is whatever your institution's transcript shows, using its own grade table and rules.



---

# hreflang tags explained: language codes, x-default and the return-link rule

> If your site exists in more than one language or country version, search engines need to know which page is for whom. hreflang tags say so. This guide covers the code format, the three ways to add the tags, why every page must link back and the mistakes that make the whole set ignored.

Published 2026-10-09 by Toolcloak. 3 minute read. Original: https://toolcloak.pages.dev/articles/hreflang-tags-explained-language-codes-x-default-and-the-return-link-rule
Related free tool: [Hreflang Tag Generator](https://toolcloak.pages.dev/tools/hreflang-tag-generator)

Suppose you have the same page in English, Hindi and Spanish, or in English for the US and for the UK. Without help, a search engine may show the wrong one to the wrong person. **hreflang** annotations connect all the versions and say 'this one is for these readers'.

> **In short**
> - Each tag has a **language** (like `en`) and optionally a **region** (like `GB`): `en-GB`.
> - Every version must list **all versions, including itself**, and each one must link back. Missing return links cause the tags to be ignored.
> - Add one `x-default` entry for visitors who match none of your versions.
> - You can place tags in the page head, in HTTP headers or in the XML sitemap. Pick one method and keep it consistent.

## What a tag looks like

A tag is a `link` element in the `head` of the page with three parts: `rel="alternate"`, an `hreflang` value and an `href` with the full address of that version.

| Part | Example | Notes |
| --- | --- | --- |
| `rel` | `alternate` | Always this value for hreflang links |
| `hreflang` | `en-GB` | Language, or language plus region |
| `href` | `https://example.com/en-gb/page` | Absolute address, with https and the full domain |

*Use the exact canonical address, with the same trailing slash and host name that you use everywhere else.*

Try it: [Hreflang Tag Generator](https://toolcloak.pages.dev/tools/hreflang-tag-generator). Add your versions and get the full set of tags to paste into every page.

## Getting the codes right

The language comes first, in lowercase, from the **ISO 639-1** list. The region is optional, comes second in capitals and uses **ISO 3166-1 alpha-2** country codes. You can have a language alone, but you cannot have a region alone.

| Code | Meaning | Common slip |
| --- | --- | --- |
| `en` | English, any country |  |
| `en-US` | English for the United States |  |
| `en-GB` | English for the United Kingdom | Not `en-UK`: UK is not a valid country code here |
| `hi` | Hindi |  |
| `hi-IN` | Hindi for India |  |
| `es-MX` | Spanish for Mexico | Region alone like `MX` is invalid |
| `pt-BR` | Portuguese for Brazil |  |
| `zh-Hans` | Chinese, simplified script | Script codes are allowed after the language |
| `x-default` | Fallback for everyone else | Not a language; a special value |

*The language code describes the content; the region is about the target audience. A Hindi page for readers in India is `hi-IN`.*

## The return-link rule

hreflang only works if the pages agree with each other. If page A says 'my Hindi version is B', then B must say 'my English version is A'. Without that return link, search engines may ignore both lines.

**A correct set for three versions**

1. **Page EN lists.** en (itself), hi, es and x-default.
2. **Page HI lists.** en, hi (itself), es and x-default.
3. **Page ES lists.** en, hi, es (itself) and x-default.
4. **Check.** The same set appears on all three pages, so every link has a return link.

*The simplest way to stay consistent is to generate the same block for every version and paste it everywhere.*

**Right and wrong**

| Reciprocal | One-way |
| --- | --- |
| Tags are used | Tags may be ignored |
| EN points to HI | EN points to HI |
| HI points back to EN | HI does not point back |
| Each page also lists itself | Self-reference is missing |

*Search Console's international targeting reports and many crawlers flag missing return tags.*

## Why the number of tags grows fast

Because every version lists every version, the total number of annotations is **n × n** for n versions. Three versions mean 9 lines across the site. Ten versions mean 100. Many sites reach for the sitemap method when the count gets large.

**Total hreflang annotations per page set**

|  | Value |
| --- | --- |
| 2 versions | 4 |
| 3 | 9 |
| 5 | 25 |
| 10 | 100 |
| 20 | 400 |

*Each page carries n lines, and there are n pages. Add x-default and each page carries n + 1 lines.*

**Lines in each page's head as versions grow (with x-default)**

| Number of versions | Lines per page |
| --- | --- |
| 2 | 3 |
| 3 | 4 |
| 5 | 6 |
| 8 | 9 |
| 10 | 11 |
| 15 | 16 |
| 20 | 21 |

*Beyond about ten versions, head tags add weight to every page. HTTP headers or the sitemap keep the page lighter.*

## Three ways to add them

| Method | Where it lives | Best for |
| --- | --- | --- |
| HTML link tags | In the head of each page | Most sites, easy to check by viewing source |
| HTTP headers | In the server response | Files that have no head, such as PDFs |
| XML sitemap | In sitemap.xml with xhtml:link entries | Large sites with many versions |

*Choose one method per site. Mixing methods risks conflicts that are hard to debug.*

## What x-default does

**x-default in practice**

| Box | What it means |
| --- | --- |
| Language picker page | Point x-default at a page that lets people choose their language. |
| Main version | Or point it at the version you want for everyone else, usually English. |
| Only one allowed | Use a single x-default per set of versions. |
| Not required | It is a recommendation, but it prevents odd results for unmatched visitors. |

*A visitor in Japan who sees your site in English and Spanish only would get the x-default page.*

> **Common mistakes:** Using `en-UK` instead of `en-GB`, pointing to redirected or noindex pages, using relative addresses, forgetting the self-reference, listing a page in hreflang but setting its canonical to a different version, and translating only the menu while the main text stays in the original language.

## Checklist before publishing

1. **List every version** on every version, including the page itself.
2. **Use valid codes.** Language lowercase, region uppercase, no UK.
3. **Use full, canonical addresses** that return a normal 200 response.
4. **Add one x-default.**
5. **Keep it consistent** with canonical tags and the sitemap.
6. **Test a few pages** by viewing the source, then watch Search Console for errors.

## Frequently asked questions

### What is hreflang?

It is an annotation that tells search engines which language and region a page is meant for, and where to find the other versions of the same page.

### Is it en-UK or en-GB?

Use en-GB. The region part follows ISO 3166-1 alpha-2 country codes, and the code for the United Kingdom is GB.

### Do I need a self-referencing hreflang tag?

Yes. Each page should list itself along with its alternates, so the set is complete and identical on every page.

### What is x-default?

It is a special value for visitors who do not match any of your language or region versions. It usually points to a language selector or your main version.

### Where do I put hreflang tags?

In the head of each page, in HTTP headers, or in the XML sitemap. Choose one method and use it consistently.

### Does hreflang improve rankings?

Not directly. It helps search engines show the right version to the right audience, which can improve clicks and reduce duplicate-content confusion between translated pages.



---

# IBAN explained: how the mod-97 check catches typos and what each part means

> An IBAN looks like a random string, but every character has a job. This guide breaks one apart, walks through the mod-97 check by hand, shows how many typing mistakes it catches and explains why a valid IBAN is not the same as a real, working account.

Published 2026-10-09 by Toolcloak. 3 minute read. Original: https://toolcloak.pages.dev/articles/iban-explained-how-the-mod-97-check-catches-typos-and-what-each-part-means
Related free tool: [IBAN Validator](https://toolcloak.pages.dev/tools/iban-validator)

The **International Bank Account Number (IBAN)** is a standard way to write a bank account so payments can cross borders with fewer errors. Most of Europe, the Middle East and parts of Africa, Asia and South America use it. India does not; Indian transfers use an account number plus an IFSC code.

> **In short**
> - An IBAN has three parts: a **2-letter country code**, **2 check digits** and a country-specific **BBAN** (the domestic account details).
> - Length depends on the country: 15 to 34 characters. Germany is 22, France 27, Norway 15.
> - The **mod-97 check** catches every single-character typo and swapped neighbours in the examples we tested.
> - A valid IBAN only means the characters are consistent. It does not prove the account exists or belongs to the person you think.

## Taking an IBAN apart

The standard example from the specification is **GB82 WEST 1234 5698 7654 32**. Spaces are added only to make it easier to read. In the machine form there are none: GB82WEST12345698765432.

**GB82WEST12345698765432, piece by piece**

1. **GB.** Country code (United Kingdom), two letters from ISO 3166.
2. **82.** Two check digits, calculated from the rest of the IBAN.
3. **WEST.** Bank code (here, four letters identifying the bank).
4. **123456.** Sort code: identifies the branch.
5. **98765432.** The account number itself.

*The last three pieces together are the BBAN, the Basic Bank Account Number. Its layout differs from country to country.*

Try it: [IBAN Validator](https://toolcloak.pages.dev/tools/iban-validator). Paste an IBAN to check its length, format and check digits.

## Lengths by country

Each country fixes one total length. A string of the wrong length cannot be a valid IBAN for that country, which is the first thing a validator checks.

**IBAN length in characters, selected countries**

|  | Value |
| --- | --- |
| NO | 15 |
| BE | 16 |
| NL | 18 |
| DE | 22 |
| GB | 22 |
| ES | 24 |
| FR | 27 |
| MT | 31 |

*Norway has the shortest at 15 and Malta one of the longest at 31. The UAE is 23 and Saudi Arabia 24.*

## The mod-97 check, step by step

The check digits come from a calculation that anyone can repeat. A validator moves the first four characters to the end, turns letters into numbers, and takes the remainder when divided by 97. A valid IBAN always leaves **1**.

**Checking GB82WEST12345698765432**

1. **Move the first four characters to the end.** GB82 goes last: WEST12345698765432GB82
2. **Turn letters into numbers.** A = 10, B = 11 … Z = 35. W=32, E=14, S=28, T=29, G=16, B=11.
3. **Build one long number.** 3214282912345698765432161182
4. **Divide by 97 and keep the remainder.** The remainder is 1, so the IBAN passes.

*The number is too long for ordinary calculators, so software divides it in chunks. The remainder is the same.*

| Step | Working | Result |
| --- | --- | --- |
| Put 00 in place of the check digits | WEST12345698765432GB00 | Number ends in ...1100 |
| Take that number mod 97 | 3214282912345698765432161100 ÷ 97 | Remainder 16 |
| Subtract from 98 | 98 − 16 | Check digits **82** |

*This is how check digits are created: the same method in reverse. Anyone can repeat it, which is why a validator needs no database.*

## What mistakes does it catch?

We tested the example IBAN. Changing any one of its 16 digits to any other digit gives a failing check in all 144 cases. Swapping each of the 14 pairs of different neighbouring digits also fails every time. Real-world mistakes, such as typing one wrong digit or swapping two, are therefore caught almost every time.

**Our test of the example IBAN**

|  | Value | Note |
| --- | --- | --- |
| Single digit changed | 144 of 144 | All detected |
| Neighbouring digits swapped | 14 of 14 | All detected |
| Chance a random string passes | 1 in 97 | About 1.03% |
| Random strings caught | 98.97% | By this check alone |

*The 1-in-97 figure is for random strings. Real typing errors are caught far more reliably than that.*

**Out of 97 random strings of the right length, how many pass?**

| Group | Count |
| --- | --- |
| Fail the check | 96 |
| Pass by luck | 1 |

*Mod 97 leaves one of 97 possible remainders, so about one random string in 97 happens to leave 1.*

## Valid does not mean real

**What a validator can and cannot tell you**

| It can confirm | It cannot confirm |
| --- | --- |
| The format | The account |
| Country code exists and length is right | That the account is open |
| Characters are allowed for that country | That it belongs to the right person |
| Check digits are consistent | That the bank accepts payments there |

*Before a large first payment, confirm the details with the payee through a second channel, such as a phone call to a known number.*

> **Watch for payment fraud:** A common scam changes the IBAN on an invoice by email. The new IBAN will pass every format check. Always verify a changed bank detail by calling the supplier on a number you already trust.

## Writing and typing IBANs well

1. **Group in fours** when writing by hand or on paper: GB82 WEST 1234 5698 7654 32.
2. **Remove spaces and dashes** before pasting into a system.
3. **Use capital letters.** Lowercase is usually accepted but capitals are standard.
4. **Copy and paste** rather than retyping when you can.
5. **Check the first two letters.** They should match the country of the account.

## Frequently asked questions

### What is an IBAN?

An International Bank Account Number is a standardised way to write a bank account number. It starts with a two-letter country code and two check digits, followed by the country's domestic bank and account details.

### How long is an IBAN?

It depends on the country and ranges from 15 to 34 characters. Germany and the United Kingdom use 22, France 27 and Norway 15.

### How does the IBAN check digit work?

The first four characters are moved to the end, letters become numbers (A = 10 to Z = 35), and the resulting number is divided by 97. A valid IBAN leaves a remainder of 1.

### Does a valid IBAN mean the account exists?

No. Validation only proves the format and check digits are consistent. The account could be closed, wrong or fraudulent.

### Does India use IBAN?

No. Indian domestic transfers use an account number and an IFSC code. For international transfers the receiving bank may ask for an account number and a SWIFT/BIC code.

### Are spaces part of an IBAN?

No. Spaces are added only for readability, usually in groups of four. Remove them when entering an IBAN into a form.



---

# Linear regression: the best-fit line, slope, intercept and R² explained

> Study hours against marks, ad spend against sales, temperature against ice-cream orders. Linear regression draws the one straight line that fits the points best and lets you predict the next one. This guide builds it by hand from eight data points and shows where it breaks.

Published 2026-10-09 by Toolcloak. 3 minute read. Original: https://toolcloak.pages.dev/articles/linear-regression-the-best-fit-line-slope-intercept-and-r-squared-explained
Related free tool: [Linear Regression Calculator (Line of Best Fit)](https://toolcloak.pages.dev/tools/linear-regression-calculator)

Linear regression answers one question: if two things move together, what straight line describes that movement best? The line has the form **y = m × x + b**. The slope **m** says how much y changes for each step in x, and the intercept **b** says where the line starts when x is zero.

> **In short**
> - The **least squares** line is the one that makes the total of the squared gaps (residuals) between points and line as small as possible.
> - In our example, each extra study hour adds about **4.26 marks**, and the line is **y = 4.26x + 46.57**.
> - **R²** tells you how much of the ups and downs the line explains. Here it is 0.97, or 97%.
> - Predicting **inside** your data range is reasonably safe. Predicting far outside it is not.

## The example: hours studied and test score

Eight students recorded how many hours they studied and what they scored out of 100. Plotting them shows a clear upward drift, but no single straight line touches every point. Regression finds the best compromise.

| Hours (x) | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 |
| --- | --- | --- | --- | --- | --- | --- | --- | --- |
| Score (y) | 52 | 55 | 61 | 60 | 68 | 72 | 75 | 83 |

*Average hours = 4.5, average score = 65.75. The best-fit line always passes through this point of averages.*

**Real scores and the fitted line**

| Hours studied | Actual score | Fitted line |
| --- | --- | --- |
| 1 | 52 | 50.8 |
| 2 | 55 | 55.1 |
| 3 | 61 | 59.4 |
| 4 | 60 | 63.6 |
| 5 | 68 | 67.9 |
| 6 | 72 | 72.1 |
| 7 | 75 | 76.4 |
| 8 | 83 | 80.7 |

*The two lines stay close together. The biggest miss is at 4 hours, where the student scored 3.6 marks below the line.*

Try it: [Linear Regression Calculator (Line of Best Fit)](https://toolcloak.pages.dev/tools/linear-regression-calculator). Paste your x and y values to get slope, intercept, R² and predictions.

## Building the line in four steps

**From raw data to y = 4.26x + 46.57**

1. **Find the averages.** x̄ = 4.5 hours, ȳ = 65.75 marks.
2. **Measure the pull.** Add up (x − x̄)(y − ȳ) for every point: 179. Add up (x − x̄)²: 42.
3. **Slope.** m = 179 ÷ 42 = 4.2619. Each extra hour adds about 4.26 marks.
4. **Intercept.** b = ȳ − m × x̄ = 65.75 − 4.2619 × 4.5 = 46.57.

*These are the standard least squares formulas. The calculator does the same arithmetic for any number of points.*

## Reading slope and intercept

**What the numbers mean here**

|  | Value | Note |
| --- | --- | --- |
| Slope | 4.26 | Marks gained per extra hour |
| Intercept | 46.57 | Predicted score at zero hours |
| R² | 0.97 | Share explained by hours |
| Correlation r | 0.98 | Strong and positive |

*The intercept is a mathematical starting point, not always a real-life fact. Nobody studies for zero hours and scores exactly 46.57; it only anchors the line.*

## Residuals: how far each point is from the line

A **residual** is the actual value minus the predicted one. Positive means the student did better than the line expected; negative means worse. Least squares makes the residuals add up to zero and keeps their squares as small as possible.

| Hours | Actual | Predicted | Residual |
| --- | --- | --- | --- |
| 1 | 52 | 50.8 | +1.2 |
| 2 | 55 | 55.1 | −0.1 |
| 3 | 61 | 59.4 | +1.6 |
| 4 | 60 | 63.6 | −3.6 |
| 5 | 68 | 67.9 | +0.1 |
| 6 | 72 | 72.1 | −0.1 |
| 7 | 75 | 76.4 | −1.4 |
| 8 | 83 | 80.7 | +2.3 |

*Squared residuals add up to about 24.6, against about 787.5 for the total variation around the average score.*

## R²: how much does the line explain?

R² compares two amounts: the leftover error around the line (24.6) and the total spread of scores around their average (787.5). R² = 1 − 24.6 ÷ 787.5 = **0.969**. About 97% of the differences between students' scores line up with study hours in this tiny sample.

**Where the variation in scores comes from**

|  | Share |
| --- | --- |
| Explained by the line | 96.9% |
| Left over (residuals) | 3.1% |

*A high R² does not prove that studying causes higher scores. Motivation, sleep or prior knowledge could drive both.*

**How to read R²**

| Zone | From | To |
| --- | --- | --- |
| Weak | 0 | 0.25 |
| Moderate | 0.25 | 0.5 |
| Good | 0.5 | 0.75 |
| Strong | 0.75 | 1 |

*These bands are a rough guide. What counts as good depends on the field: 0.5 can be excellent for human behaviour and poor for a physics experiment.*

## Making predictions

Put a new x into the equation. For 9 hours the line says 4.2619 × 9 + 46.57 = **84.9**. That is just past our data (the maximum was 8), so it is a mild stretch. For 20 hours the line gives 131.8 marks, which is impossible on a test out of 100.

**Interpolating against extrapolating**

| Inside the data (1 to 8 hours) | Far outside the data (20 hours) |
| --- | --- |
| Reasonably safe | Unreliable |
| 6 hours gives about 72 marks | The line predicts 131.8 out of 100 |
| The line was built from points around it | Real relationships bend and flatten |
| Still carries an error of a few marks | Nothing in the data supports it |

*A straight line is an approximation over the range you measured. Always ask whether the relationship could plausibly continue.*

## One strange point can tilt the whole line

Regression gives every point a vote, and points far from the rest have the loudest voice. Change only the last student's score from 83 to 40 (say, they fell ill) and the slope collapses from 4.26 to about **0.68** marks per hour. Always plot your data before trusting the line.

**Slope of the best-fit line**

|  | Marks per hour |
| --- | --- |
| All 8 points as measured | 4.26 |
| Last score changed to 40 | 0.68 |

*One altered point out of eight cut the slope by 84%. Check unusual points: are they errors, or real cases the line should explain?*

> **Correlation is not causation:** A line that fits well tells you the two things move together in this data. It does not tell you why. Ice-cream sales and sunburn both rise with temperature; neither causes the other.

## Checklist before you trust a regression

1. **Plot first.** If the points curve, a straight line is the wrong shape.
2. **Look for outliers** and decide whether they are errors or real.
3. **Keep predictions inside the data range** wherever possible.
4. **Use enough points.** Eight points are fine for learning, but too few for serious conclusions.
5. **Do not claim cause** from fit alone.

## Frequently asked questions

### What is linear regression?

It is a method that finds the straight line y = mx + b that best fits a set of points, so you can describe the relationship between two variables and predict new values.

### What is the least squares method?

It chooses the line that minimises the sum of the squared vertical gaps between the points and the line. Squaring stops positive and negative gaps from cancelling and punishes large misses.

### What does R-squared mean?

R² is the share of the variation in y that the line explains. 0 means the line explains nothing and 1 means it explains everything. In the example, 0.97 means about 97 percent.

### What is a residual?

A residual is the actual value minus the value the line predicts. Small, random-looking residuals suggest a good fit; patterns in them suggest a straight line is the wrong model.

### Can I predict values outside my data?

You can, but be careful. Relationships often bend or flatten outside the measured range, so predictions far beyond your data can be badly wrong.

### Is regression the same as correlation?

They are related. Correlation r measures how tightly points follow a line, and for one variable R² equals r squared. Regression goes further and gives the actual equation of the line.



---

# Loan amortization schedule explained: why early EMIs are mostly interest

> On a 20-year home loan, your first EMI sends about 80% of its money to interest. This guide builds the schedule month by month, shows the exact month the balance tips the other way, and tests what extra payments really save.

Published 2026-10-09 by Toolcloak. 3 minute read. Original: https://toolcloak.pages.dev/articles/loan-amortization-schedule-explained-why-early-emis-are-mostly-interest
Related free tool: [Loan Amortization Calculator](https://toolcloak.pages.dev/tools/loan-amortization-calculator)

An **amortization schedule** is the full table of a loan: every instalment, split into the part that pays interest and the part that reduces what you owe. Banks send it as a statement; most borrowers never read it. It explains almost everything that feels odd about a loan, such as why the balance barely moves in year one.

> **In short**
> - Each month, **interest = outstanding balance × monthly rate**. The rest of the EMI reduces the balance.
> - On a ₹30 lakh loan at 8.5% for 20 years, the EMI is **₹26,035**. In month 1, **₹21,250** of it is interest and only **₹4,785** is principal.
> - The split flips in **month 143**, nearly 12 years in, when principal first beats interest.
> - Total interest is **₹32.48 lakh**, more than the loan itself. Extra payments early on save the most.

## How one month is calculated

Take the example loan: **₹30,00,000 at 8.5% a year for 20 years**. The monthly rate is 8.5% ÷ 12 = 0.7083%. The EMI formula gives ₹26,035 a month. Every month then follows the same three steps.

**Month 1 of the example loan**

1. **Interest for the month.** ₹30,00,000 × 0.7083% = **₹21,250**.
2. **Principal repaid.** EMI ₹26,035 − interest ₹21,250 = **₹4,785**.
3. **New balance.** ₹30,00,000 − ₹4,785 = **₹29,95,215**.
4. **Repeat.** Month 2 charges interest on ₹29,95,215, so interest dips slightly to ₹21,216 and principal rises to ₹4,819.

*The EMI never changes. Only the split moves, a little each month, because the balance shrinks.*

Try it: [Loan Amortization Calculator](https://toolcloak.pages.dev/tools/loan-amortization-calculator). Enter amount, rate and tenure to see every month's interest, principal and balance.

## The schedule at a glance

| Month | Interest | Principal | Balance left |
| --- | --- | --- | --- |
| 1 | ₹21,250 | ₹4,785 | ₹29,95,215 |
| 12 | ₹20,864 | ₹5,171 | ₹29,40,293 |
| 60 (year 5) | ₹18,778 | ₹7,256 | ₹26,43,815 |
| 120 (year 10) | ₹14,952 | ₹11,083 | ₹20,99,815 |
| 180 (year 15) | ₹9,108 | ₹16,926 | ₹12,68,962 |
| 240 (last) | ₹183 | ₹25,852 | ₹0 |

*After five years you have paid about ₹15.6 lakh in EMIs but cleared only about ₹3.56 lakh of the loan.*

**Where each EMI goes, month by month (₹ thousand)**

| Month | Interest part | Principal part |
| --- | --- | --- |
| 1 | 21250 | 4785 |
| 24 | 20407 | 5628 |
| 48 | 19368 | 6667 |
| 72 | 18137 | 7898 |
| 96 | 16679 | 9356 |
| 120 | 14952 | 11083 |
| 144 | 12906 | 13128 |
| 168 | 10483 | 15552 |
| 192 | 7612 | 18422 |
| 216 | 4212 | 21823 |
| 240 | 183 | 25852 |

*The two lines cross at month 143. Before that point, most of every payment is the cost of borrowing.*

## The year-by-year view

**Interest paid in each year (₹ thousand)**

|  | Value |
| --- | --- |
| Y1 | ₹253k |
| Y3 | ₹242k |
| Y5 | ₹229k |
| Y8 | ₹204k |
| Y10 | ₹184k |
| Y13 | ₹147k |
| Y16 | ₹100k |
| Y20 | ₹14k |

*Year 1 interest is ₹2.53 lakh. In year 20 it is only ₹14,000. The loan gets cheaper to carry the longer you hold it.*

**The example loan in four numbers**

|  | Value | Note |
| --- | --- | --- |
| Monthly EMI | ₹26,035 | Fixed for 240 months |
| Total paid | ₹62.48 L | EMI × 240 |
| Total interest | ₹32.48 L | 108% of the loan |
| Principal passes interest | Month 143 | About 11 years 11 months |

*Total interest is the sum of all 240 interest figures. It is not 8.5% × 20 years, because the balance falls as you pay.*

## Why the first years are so expensive

Interest is charged on what you still owe, and at the start you owe everything. A larger balance means more interest, which leaves less of the fixed EMI to reduce the balance. The balance then falls slowly at first, and faster later, like a snowball in reverse.

**Where the first year's EMIs go**

|  | Share |
| --- | --- |
| Interest | ₹2.53 L (81%) |
| Principal | ₹0.60 L (19%) |

*Year 1 EMIs total ₹3,12,416. Only about 19 paise of each rupee reduces the loan.*

> **Why this matters if you sell early:** If you sell or switch lenders in year 3 or 4, your balance will still be about ₹27–28 lakh on a ₹30 lakh loan. Home loan borrowers are often surprised by how little equity they have built early on.

## Tenure changes the total cost a lot

A longer tenure lowers the EMI but raises total interest sharply. Same loan, same 8.5% rate:

| Tenure | EMI | Total interest | Interest as % of loan |
| --- | --- | --- | --- |
| 10 years | ₹37,196 | ₹14.63 L | 49% |
| 15 years | ₹29,542 | ₹23.18 L | 77% |
| 20 years | ₹26,035 | ₹32.48 L | 108% |
| 25 years | ₹24,157 | ₹42.47 L | 142% |
| 30 years | ₹23,067 | ₹53.04 L | 177% |

*Going from 20 to 30 years cuts the EMI by only ₹2,968 a month but adds ₹20.6 lakh of interest.*

**Total interest on ₹30 lakh at 8.5% (₹ lakh)**

|  | Interest |
| --- | --- |
| 10 years | 14.6 |
| 15 years | 23.2 |
| 20 years | 32.5 |
| 25 years | 42.5 |
| 30 years | 53.0 |

*Each extra five years adds more interest than the five years before it.*

## Rate changes

| Rate | EMI | Total interest |
| --- | --- | --- |
| 7.5% | ₹24,168 | ₹28.00 L |
| 8.5% | ₹26,035 | ₹32.48 L |
| 9.5% | ₹27,964 | ₹37.11 L |

*Each full percentage point moves the EMI by about ₹1,900 and the lifetime interest by about ₹4.5 lakh.*

On a floating-rate loan, a rate change usually keeps the EMI the same and changes the tenure instead, or the reverse. Check your sanction letter. The schedule is recalculated from the current balance each time.

## What extra payments save

Any extra amount goes straight to principal, so it removes all the interest that balance would have collected for the rest of the loan. We ran two simple plans on the same loan.

**Two ways to prepay the example loan**

| One extra EMI each year | ₹5,000 extra every month |
| --- | --- |
| Saves ₹6.17 L | Saves ₹12.73 L |
| Loan ends in 201 months (16 years 9 months) | Loan ends in 164 months (13 years 8 months) |
| Total interest ₹26.31 L instead of ₹32.48 L | Total interest ₹20.75 L |
| About 3 years 3 months sooner | About 6 years 4 months sooner |

*₹5,000 a month is ₹60,000 a year, close to two extra EMIs. The monthly plan saves about twice as much because the money arrives earlier and more often.*

> **Check prepayment rules:** Floating-rate home loans to individuals generally carry no prepayment penalty in India, but fixed-rate loans, business loans and some other products may. Ask your lender for the rule in writing and tell them to reduce tenure or EMI as you prefer.

## Reading a real bank statement

1. **Find the outstanding principal.** This is the number that matters if you want to close the loan today.
2. **Check the rate applied** in each period, especially after a rate reset.
3. **Add up interest paid** for the financial year. It is needed for tax certificates and deduction claims.
4. **Look for pre-EMI interest.** On under-construction homes you may pay interest only until possession.
5. **Compare the schedule** with the calculator. Small differences come from the day count and rounding.

## Frequently asked questions

### What is an amortization schedule?

It is a table showing every instalment of a loan, split into interest and principal, with the balance left after each payment.

### Why is most of my early EMI interest?

Interest is charged on the outstanding balance, which is largest at the start. The EMI is fixed, so a bigger interest share leaves less to reduce the principal.

### How is interest calculated each month on a reducing balance loan?

Monthly interest is the opening balance multiplied by the annual rate divided by 12. Principal repaid is the EMI minus that interest.

### When does the principal part overtake the interest part?

It depends on rate and tenure. On a ₹30 lakh loan at 8.5% for 20 years it happens in month 143. Higher rates and longer tenures push it later.

### Does paying one extra EMI a year help?

Yes. On the example loan it shortens the term by about 3 years 3 months and saves about ₹6.2 lakh of interest.

### Is it better to reduce tenure or EMI after prepaying?

Reducing tenure saves more total interest. Reducing the EMI gives more monthly cash flow. Choose based on your budget and how stable your income is.



---

# Home loan: where your EMI goes in year 1 and in year 20

> You pay the same EMI for twenty years, but the money inside it changes sides. In year one about four out of five rupees go to the bank's interest. By year twenty almost nothing does. Seeing that shift is the key to every smart move on a home loan.

Published 2026-10-09 by Toolcloak. 4 minute read. Original: https://toolcloak.pages.dev/articles/home-loan-where-your-emi-goes-in-year-1-and-year-20
Related free tool: [Mortgage Calculator](https://toolcloak.pages.dev/tools/mortgage-calculator)

A ₹50 lakh home loan at 8.5 percent for 20 years has an EMI of **₹43,391**. Over twenty years that adds up to about ₹1.04 crore. The house cost ₹50 lakh of borrowed money, and the bank collects another ₹54 lakh for lending it. This guide shows where that second number comes from, and what you can do about it.

> **In short**
> - The EMI is fixed, but its **split is not**: early EMIs are mostly interest, late EMIs are mostly principal.
> - A longer tenure lowers the EMI but raises the total interest a lot. 30 years costs about 63% more interest than 20.
> - One extra EMI a year, or a steady ₹5,000 extra each month, can cut years off the loan and lakhs off the interest.
> - Check your EMI against take-home pay. Around 30 to 40 percent is where it starts to pinch.

## The loan we will use

**₹50 lakh at 8.5% for 20 years**

|  | Value | Note |
| --- | --- | --- |
| Monthly EMI | ₹43,391 | Same every month |
| Total you pay | ₹1.04 cr | 240 payments |
| Interest alone | ₹54.1 L | Beyond the ₹50 L borrowed |

*All numbers in this guide use this loan unless we say otherwise. They are plain EMI-formula figures: no processing fee, no rate change, no part-payments except where mentioned.*

## Why early EMIs are mostly interest

Each month the bank charges interest on the balance that is still owed. In month one the balance is the full ₹50 lakh, so the interest is about ₹35,400. Your EMI is ₹43,391, so only about ₹7,970 is left to reduce the loan. Next month the balance is a little smaller, the interest a little smaller, and a little more goes to principal. That small gap compounds in your favour over time.

**What a full year of EMIs (₹5.21 lakh) is made of**

|  | Interest to the bank | Principal repaid |
| --- | --- | --- |
| Year 1 | ₹4.21 L interest | ₹1.0 L |
| Year 5 | ₹3.81 L interest | ₹1.4 L |
| Year 10 | ₹3.07 L interest | ₹2.13 L |
| Year 15 | ₹1.95 L | ₹3.26 L principal |
| Year 20 | 23202 | ₹4.97 L principal |

*In the first year only about ₹1 lakh of your ₹5.2 lakh in EMIs reduces what you owe. In the last year nearly all of it does.*

> **What this means in practice:** The best time to prepay a home loan is **early**, because each rupee you pay off in year two removes interest that would have been charged for eighteen more years. A rupee prepaid in year eighteen saves very little.

Try it: [Mortgage Calculator](https://toolcloak.pages.dev/tools/mortgage-calculator). Enter your loan amount, rate and term to see your EMI, total interest and the effect of extra payments.

## Tenure: lower EMI, much higher total

A longer loan feels gentler every month, and that is exactly why it costs more. You borrow the same ₹50 lakh for more months, so interest has more time to be charged.

**Same ₹50 lakh at 8.5%, different tenures**

|  | Value |
| --- | --- |
| 10 yrs | ₹74 L |
| 15 yrs | ₹89 L |
| 20 yrs | ₹104 L |
| 25 yrs | ₹121 L |
| 30 yrs | ₹138 L |

*Totals in lakh rupees. The EMIs are ₹61,993 (10 yrs), ₹49,237 (15), ₹43,391 (20), ₹40,261 (25) and ₹38,446 (30). Going from 20 to 30 years lowers the EMI by only ₹4,945 but adds ₹34 lakh of interest.*

Notice how little the EMI falls once you pass 20 years. The first few years of tenure buy you a lot of relief. The last ten buy you very little.

## A one-point rate change is bigger than it looks

**Monthly EMI for ₹50 lakh over 20 years, by interest rate**

|  | EMI |
| --- | --- |
| 7.5% | ₹40,280 |
| 8.0% | ₹41,822 |
| 8.5% | ₹43,391 |
| 9.0% | ₹44,986 |
| 9.5% | ₹46,607 |
| 10.0% | ₹48,251 |

*Each extra half percent adds roughly ₹1,500 to ₹1,650 a month. Between 7.5% and 8.5% the total interest differs by about ₹7.5 lakh over the loan.*

If your loan is on a floating rate, this table is your risk map. A two-point rise from 8.5% to 10.5% is not a rounding error. Ask your lender whether a rate rise lengthens your tenure or raises your EMI, because the two feel very different.

## Three ways to prepay, compared

**Interest paid on ₹50 lakh at 8.5% (lakh rupees)**

|  | Interest |
| --- | --- |
| No prepayment, 20 years | ₹54.1 L |
| One extra EMI each year, done in 16 yr 9 mo | ₹43.8 L |
| ₹5,000 extra every month, done in 15 yr 7 mo | ₹40.2 L |

*An extra EMI a year is about ₹43,400 of additional money in a year. It saves around ₹10.3 lakh and ends the loan 39 months early. A steady ₹5,000 monthly (₹60,000 a year) saves about ₹13.9 lakh and ends it 53 months early.*

The monthly route wins slightly because the money arrives earlier. But the best plan is the one you will actually follow: a yearly bonus used for one lump-sum prepayment works as well as a standing top-up if you do it every year.

> **Check two things before you prepay:** Ask whether your loan has any prepayment charge (floating-rate loans to individuals usually do not, fixed-rate ones may). And make sure you still have your emergency fund. Paying down an 8.5% loan is good, but not at the cost of borrowing on a card at 36% when something breaks.

## Down payment: a bigger first cheque pays twice

Suppose the home costs ₹70 lakh and you take a 20-year loan at 8.5%. Here is what changes with the down payment:

| Down payment | Loan | EMI | Total interest |
| --- | --- | --- | --- |
| 10% (₹7 L) | ₹63 L | ₹54,673 | ₹68.2 L |
| 20% (₹14 L) | ₹56 L | ₹48,598 | ₹60.6 L |
| 30% (₹21 L) | ₹49 L | ₹42,523 | ₹53.1 L |

*Every extra ₹7 lakh paid up front lowers the EMI by about ₹6,100 and saves around ₹7.5 lakh of interest over the loan.*

## Is the EMI too big for your income?

A common rule of thumb is to keep all your EMIs together below roughly 40 percent of your take-home pay. Lenders themselves often look at somewhere between 40 and 50 percent of income when they approve a loan, but being approved is not the same as being comfortable. If your take-home is ₹1.2 lakh a month, this loan's EMI of ₹43,391 is about 36 percent.

**EMI as a share of monthly take-home pay**

| Zone | Range |
| --- | --- |
| Comfortable | under 30% |
| Stretched | 30–40% |
| Risky | over 40% |

*These bands are a rough guide, not an official standard. Add other EMIs, school fees and rent you still pay. Your own fixed costs matter more than any rule.*

## A five-step checklist before you sign

1. **Work out the total, not only the EMI.** Multiply the EMI by the number of months and subtract the loan amount.
2. **Try two or three tenures.** Pick the shortest one whose EMI you can pay without borrowing elsewhere.
3. **Test a higher rate.** Run the numbers at two points above today's rate and see if you can still manage.
4. **Plan your prepayment.** Decide now what part of every bonus or raise goes to the loan.
5. **Keep six months of expenses aside** before you pour spare money into the loan.

> **Try it:** Open the mortgage calculator, enter ₹50,00,000, 8.5 percent and 20 years, then add ₹5,000 as the extra monthly payment. You should see the term fall to about 15 years and 7 months.

## Frequently asked questions

### How is the EMI on a home loan calculated?

EMI = P × r × (1 + r)^n ÷ ((1 + r)^n − 1), where P is the loan, r is the monthly interest rate (yearly rate ÷ 12) and n is the number of months. For ₹50 lakh at 8.5% over 20 years it comes to about ₹43,391.

### Why is most of my early EMI interest?

Interest is charged on the balance you still owe. At the start that balance is the largest, so interest is largest too. As the balance falls, the interest part of each EMI falls and the principal part rises.

### Is a longer home loan tenure better?

It gives a lower EMI but raises the total interest a lot. On ₹50 lakh at 8.5%, a 30-year loan costs about ₹88 lakh in interest against about ₹54 lakh for 20 years.

### How much can I save by prepaying a home loan?

On ₹50 lakh at 8.5% for 20 years, paying one extra EMI each year saves about ₹10.3 lakh in interest and ends the loan about 39 months early. An extra ₹5,000 every month saves about ₹13.9 lakh.

### How much of my salary should go to EMI?

A common guide is to keep total EMIs under about 30 to 40 percent of take-home pay. It is a rule of thumb, not a regulation, so adjust it for your own expenses.



---

# How to calculate your net worth: assets minus debts, and what the number really tells you

> Net worth is the single number that shows where you stand financially. This guide explains what to count, how to value each item honestly, how to read the debt-to-asset ratio and why a property-heavy net worth can still leave you short of cash.

Published 2026-10-09 by Toolcloak. 3 minute read. Original: https://toolcloak.pages.dev/articles/how-to-calculate-your-net-worth-assets-minus-debts-and-what-the-number-really-tells-you
Related free tool: [Net Worth Calculator](https://toolcloak.pages.dev/tools/net-worth-calculator)

Your **net worth** is everything you own minus everything you owe. It is a snapshot, like a personal balance sheet, and it is a better measure of financial health than salary because it shows what you have actually kept. A high income with large debts can leave a low net worth, and a modest income saved steadily can build a high one.

> **In short**
> - **Net worth = total assets − total debts.**
> - With assets of **₹31,00,000** and debts of **₹15,30,000**, net worth is **₹15,70,000**.
> - The **debt-to-asset ratio** is debts ÷ assets, here **49.4%**. Lower is safer.
> - Most of that net worth is a house. Cash and investments minus short-term dues are only **₹5,20,000**, so check your **liquid** net worth too.

## The formula and a worked example

Add up every asset, add up every debt, then subtract. The calculator groups assets into four buckets and debts into two, which is enough for a clear picture without needing a long spreadsheet.

**Example: a typical early-career balance sheet**

1. **Add the assets.** ₹1,50,000 cash + ₹4,00,000 investments + ₹25,00,000 property and vehicles + ₹50,000 other = **₹31,00,000**.
2. **Add the debts.** ₹15,00,000 loans + ₹30,000 card dues = **₹15,30,000**.
3. **Subtract.** ₹31,00,000 − ₹15,30,000 = **₹15,70,000** net worth.
4. **Work out the ratio.** ₹15,30,000 ÷ ₹31,00,000 = **49.4%** debt-to-asset.

*The tool returns four numbers: net worth, total assets, total debts and the debt-to-asset ratio.*

Try it: [Net Worth Calculator](https://toolcloak.pages.dev/tools/net-worth-calculator). Fill in your assets and debts to see your net worth and debt-to-asset ratio.

## What counts as an asset

| Asset | How to value it | Common mistake |
| --- | --- | --- |
| Cash and bank balances | Today's balance across savings and current accounts | Forgetting accounts you rarely use |
| Investments | Current market value of stocks, mutual funds, FDs, bonds and retirement accounts | Using the amount you invested, not today's value |
| Property and vehicles | A realistic selling price today, not the purchase price or a wish | Valuing a car at its showroom price |
| Other assets | Gold, business stake, money lent out that you expect back | Counting sentimental items you would never sell |

*Use what you would get if you sold the item today. Be a little conservative for items that are hard to sell.*

**What the example assets are made of**

|  | Share |
| --- | --- |
| Cash and bank | ₹1.5 L (4.8%) |
| Investments | ₹4 L (12.9%) |
| Property and vehicles | ₹25 L (80.6%) |
| Other assets | ₹0.5 L (1.6%) |

*Property and vehicles are over 80% of the assets. That is typical, and it explains why net worth can feel high while cash feels tight.*

## What counts as a debt

- **Loans:** home loan, car loan, education loan, personal loan, loans from family that you must repay. Use the **outstanding principal**, not the original amount.
- **Credit card and other dues:** the full unpaid balance, plus any bills you owe such as tax or rent arrears.
- **Not a debt:** next month's regular bills that you pay out of salary, unless they are overdue.

> **Use the outstanding balance:** A home loan of ₹15 lakh outstanding counts as ₹15 lakh, even if the original loan was ₹30 lakh. Your lender's statement shows the current principal. The [Loan Amortization](https://toolcloak.pages.dev/tools/loan-amortization-calculator) schedule lets you see how that balance falls over time.

## Reading the debt-to-asset ratio

The ratio shows how much of what you own is financed by borrowing. At 49.4% in the example, about half the assets are covered by debt. A ratio under 100% means net worth is positive. A ratio above 100% means you owe more than you own.

**Debt-to-asset ratio, a rough way to read it**

| Zone | From | To |
| --- | --- | --- |
| Low debt | 0 | 30 |
| Moderate | 30 | 60 |
| High | 60 | 100 |
| Negative net worth | 100 | 120 |

*These bands are a rule of thumb, not an official standard. A home loan early in life often puts people in the moderate band, and the ratio usually falls as loans are repaid.*

## Net worth is not the same as spendable money

Most of the example net worth sits in a house. You cannot spend a house on a bad month. A second figure, **liquid net worth**, counts only cash and investments and subtracts short-term dues.

**Two ways to read the same finances**

| Total net worth | Liquid net worth |
| --- | --- |
| ₹15,70,000 | ₹5,20,000 |
| All assets minus all debts | Cash and investments (₹5,50,000) minus card dues (₹30,000) |
| Includes the house and the car | What you could use within days |
| The long-term picture | The safety-net picture |

*Both numbers are useful. The first shows long-term wealth, the second shows how well you could handle a job loss or a medical bill.*

## How net worth grows over time

Net worth grows from two sources: new savings and the returns on what you already own. Suppose you start with the example net worth, save ₹2 lakh a year and earn 7% a year on your net worth. These are assumptions for illustration, not a forecast.

**Net worth over 10 years with ₹2 lakh saved a year and 7% growth (value in lakh rupees)**

| Years from now | Projected net worth |
| --- | --- |
| 0 | 15.7 |
| 1 | 18.8 |
| 2 | 22.12 |
| 3 | 25.67 |
| 4 | 29.47 |
| 5 | 33.53 |
| 6 | 37.88 |
| 7 | 42.53 |
| 8 | 47.51 |
| 9 | 52.84 |
| 10 | 58.54 |

*Net worth rises from ₹15.7 L to about ₹58.54 L in ten years. Real returns vary year to year and are never guaranteed.*

## Tracking it once a year

1. **Pick a date** and do the calculation on the same date each year, such as 1 April or 31 December.
2. **Collect statements** for every bank account, investment and loan.
3. **Value property and vehicles conservatively** and keep the same method each year.
4. **Enter the totals in the calculator** and save the four result numbers.
5. **Compare with last year.** Look at net worth, the debt-to-asset ratio and liquid net worth separately.

## Frequently asked questions

### What is net worth?

Net worth is the total value of everything you own (assets) minus everything you owe (debts). It shows what you would be left with if you sold all your assets and paid off every debt.

### How do I calculate my net worth?

List all your assets at today's value, list all your debts at the outstanding balance, then subtract the debts from the assets. The Net Worth Calculator does the addition and subtraction for you.

### Should I include the house I live in?

Many people do, because it is a real asset that you could sell. Others leave it out to see the money that works for them. The best approach is to calculate both: total net worth with the house and liquid net worth without it.

### Can net worth be negative?

Yes. If your debts are larger than your assets, net worth is negative. This is common for people with student loans or a fresh home loan, and it usually improves as loans are repaid.

### What is a good debt-to-asset ratio?

There is no official standard, but lower is safer. A ratio under about 30% means little of what you own is financed by debt, and a ratio above 100% means you owe more than you own.

### How often should I check my net worth?

Once or twice a year is enough for most people. Checking too often makes you react to normal market ups and downs. The [Net Worth Calculator](https://toolcloak.pages.dev/tools/net-worth-calculator) takes a minute with your latest statements.



---

# Ohm's law explained: volts, amps, ohms and watts with worked examples

> Voltage, current, resistance and power are linked by two short formulas. Learn the triangle, the twelve-formula wheel, how to size an LED resistor, why a kettle draws 8.7 amps, and how series and parallel resistors change the numbers.

Published 2026-10-09 by Toolcloak. 4 minute read. Original: https://toolcloak.pages.dev/articles/ohms-law-explained-volts-amps-ohms-and-watts-with-worked-examples
Related free tool: [Ohm’s Law Calculator (Voltage, Current, Resistance, Power)](https://toolcloak.pages.dev/tools/ohms-law-calculator)

Every circuit, from a torch to a train, obeys the same short rule: the current through a conductor depends on the voltage pushing it and the resistance holding it back. **Ohm's law** puts that in one line, and with a second formula for power you can answer most everyday electrical questions.

> **In short**
> - **V = I × R**: voltage (volts) = current (amps) × resistance (ohms).
> - **P = V × I**: power (watts) = voltage × current. Combine the two and you get P = I²R and P = V²/R.
> - Know any two of the four quantities (V, I, R, P) and you can find the other two.
> - Resistors in **series** add up. In **parallel** the total is always smaller than the smallest one.

## The three-way triangle

Cover the quantity you want and the other two show how to get it. Voltage is on top; current and resistance sit below.

**One law, three forms**

1. **Find voltage.** V = I × R. A 0.5 A current through 10 Ω needs 5 V.
2. **Find current.** I = V ÷ R. A 12 V supply across 4 Ω drives 3 A.
3. **Find resistance.** R = V ÷ I. 9 V and 0.03 A means 300 Ω.

*Use base units: volts, amps and ohms. Convert milliamps by dividing by 1,000 (20 mA = 0.02 A) and kilo-ohms by multiplying by 1,000.*

Try it: [Ohm’s Law Calculator (Voltage, Current, Resistance, Power)](https://toolcloak.pages.dev/tools/ohms-law-calculator). Enter any two values to get the other two, with the formula used.

## Adding power

Power is how fast electrical energy is turned into heat, light or motion. Since V = IR, you can swap terms to write power in three useful ways, depending on which two numbers you know.

| You know | Power formula | Example |
| --- | --- | --- |
| Voltage and current | P = V × I | 230 V × 8.7 A ≈ 2,000 W |
| Current and resistance | P = I² × R | 3 A and 4 Ω → 9 × 4 = 36 W |
| Voltage and resistance | P = V² ÷ R | 12 V and 4 Ω → 144 ÷ 4 = 36 W |

*All three answers can agree. The 12 V, 4 Ω, 3 A circuit above uses 36 W by every route.*

## Worked example 1: an LED resistor

An LED needs a resistor in series, or it draws too much current and burns out. Suppose a 5 V supply, a red LED that drops about 2 V, and a target current of 20 mA.

**Sizing the resistor**

1. **Voltage left for the resistor.** 5 V − 2 V = **3 V**.
2. **Resistance.** R = V ÷ I = 3 ÷ 0.02 = **150 Ω**.
3. **Power in the resistor.** P = V × I = 3 × 0.02 = **0.06 W**.
4. **Pick a part.** A 150 Ω, quarter-watt (0.25 W) resistor has plenty of margin.

*If you cannot find 150 Ω, use the next higher standard value. A slightly higher resistance gives a slightly dimmer, cooler LED, which is safe.*

## Worked example 2: a 2,000 W kettle

A kettle is rated 2,000 W on a 230 V supply, the standard in India and much of the world. Rearranging P = V × I gives the current, and Ohm's law then gives the resistance of the heating element.

**Inside a 2,000 W, 230 V kettle**

|  | Value | Note |
| --- | --- | --- |
| Current drawn | 8.7 A | 2,000 ÷ 230 |
| Element resistance | 26.5 Ω | 230 ÷ 8.7 |
| Energy in one hour | 2.0 kWh | 2 kW × 1 h |
| Typical socket rating | 10 A | Small margin left |

*A 16 A socket is common for heavy appliances. Plug a kettle and a heater into the same extension board and the total can exceed what the cable allows.*

## Worked example 3: light bulbs

Higher wattage at the same voltage means more current and lower resistance. A thicker filament, not a thinner one, makes a brighter incandescent bulb.

| Bulb (230 V) | Current | Resistance when hot |
| --- | --- | --- |
| 60 W | 0.26 A | 882 Ω |
| 100 W | 0.43 A | 529 Ω |
| 1,500 W heater | 6.52 A | 35.3 Ω |

*Resistance of a filament rises as it heats, so a cold bulb measures much lower than these hot figures.*

**Current drawn at 230 V (amps)**

|  | Amps |
| --- | --- |
| 60 W bulb | 0.26 |
| 100 W bulb | 0.43 |
| 1,500 W heater | 6.52 |
| 2,000 W kettle | 8.70 |

*Current climbs in step with power when voltage is fixed. Heating appliances use by far the most.*

## Series and parallel resistors

Real circuits use more than one resistor. How you connect them decides how the values combine.

**100 Ω and 220 Ω resistors**

| In series (one after the other) | In parallel (side by side) |
| --- | --- |
| 320 Ω | 68.75 Ω |
| Totals add: 100 + 220 | 1/R = 1/100 + 1/220 |
| Same current through both | Smaller than the smallest resistor |
| Voltage splits between them | Same voltage across both |

*For two resistors in parallel, a handy shortcut is (R1 × R2) ÷ (R1 + R2) = 22,000 ÷ 320 = 68.75 Ω.*

A voltage divider is a series pair used to get a smaller voltage. With a 9 V battery across 100 Ω and 220 Ω, the current is 9 ÷ 320 = 0.028 A. The voltage across the 220 Ω resistor is 0.028 × 220 = 6.2 V, and across the 100 Ω resistor 2.8 V.

## Where the law stops being simple

- **Non-ohmic parts.** Diodes, LEDs and transistors do not have a fixed resistance. Their current does not rise in a straight line with voltage.
- **Temperature.** Metal resistance rises with heat, so a hot wire behaves differently from a cold one.
- **AC circuits.** With alternating current, coils and capacitors add reactance, and the law uses impedance instead of resistance.
- **Real batteries.** Voltage drops under load because of internal resistance.

> **Mains voltage is dangerous:** These formulas are for understanding and low-voltage projects. Never work on live mains wiring. If an appliance trips the breaker, get a qualified electrician to check it.

## Quick sanity checks

1. **Convert units first.** Milliamps and kilo-ohms are the usual slip.
2. **Check the power rating** of any resistor: P = I²R must be below its watt rating, with margin.
3. **Voltage drops add up** around a loop to equal the supply voltage.
4. **Parallel total below the smallest.** If yours is larger, recheck the calculation.
5. **Wire gauge matters.** Higher current needs thicker wire to avoid heating.

## Frequently asked questions

### What is Ohm's law?

It states that voltage equals current multiplied by resistance, V = I × R. It describes how much current flows through a resistor for a given voltage.

### How do I calculate current from power and voltage?

Divide power by voltage: I = P ÷ V. A 2,000 W kettle on 230 V draws about 8.7 amps.

### What is the formula for electrical power?

P = V × I. Using Ohm's law you can also write it as P = I² × R or P = V² ÷ R.

### How do I choose a resistor for an LED?

Subtract the LED's forward voltage from the supply voltage and divide by the desired current. For 5 V, a 2 V LED and 20 mA, the resistor is 150 ohms.

### How do resistors add in parallel?

The reciprocals add: 1/R = 1/R1 + 1/R2. The total is always smaller than the smallest resistor. For 100 and 220 ohms it is 68.75 ohms.

### Does Ohm's law work for all components?

No. It holds for ohmic parts such as ordinary resistors. LEDs, diodes and transistors have a current that does not rise in proportion to voltage.



---

# One rep max: estimate your max lift without risking it

> You lifted 80 kg for five clean reps. What could you lift for one? A few well-known formulas turn a set you already did into an estimate of your max, and a percentage table tells you what to load on the bar next week. No heavy single needed.

Published 2026-10-09 by Toolcloak. 3 minute read. Original: https://toolcloak.pages.dev/articles/one-rep-max-estimate-your-max-lift-without-risking-it
Related free tool: [One Rep Max (1RM) Calculator](https://toolcloak.pages.dev/tools/one-rep-max-calculator)

Your **one rep max**, or 1RM, is the heaviest weight you could lift for a single good repetition. Testing it directly means loading a bar until it nearly crushes you. A far safer way is to take a set you already did, say 80 kg for 5 reps, and let a formula estimate the rest.

> **In short**
> - Two common formulas: **Epley**, 1RM = weight × (1 + reps ÷ 30), and **Brzycki**, 1RM = weight × 36 ÷ (37 − reps).
> - 80 kg × 5 gives roughly **90 to 94 kg**, depending on the formula.
> - Estimates are best from sets of **2 to 10 reps**. Above about 10 they drift apart.
> - Once you have a 1RM, use percentages of it to plan your training weights. Treat the result as a guide, not a promise.

## Three formulas, one set

Researchers have fitted many formulas to lifting data. Three are common: Epley, Brzycki and Lombard. For the same set of 80 kg × 5 reps they give slightly different answers.

**Estimated 1RM from 80 kg × 5 reps**

|  | Value | Note |
| --- | --- | --- |
| Brzycki | 90.0 kg | 80 × 36 ÷ 32 |
| Epley | 93.3 kg | 80 × (1 + 5 ÷ 30) |
| Lombard | 94.0 kg | 80 × 5^0.10 |

*A spread of about 4 kg on a 90 kg lift. In practice the safest idea is to use the middle or the lower value when you plan heavy work.*

Try it: [One Rep Max (1RM) Calculator](https://toolcloak.pages.dev/tools/one-rep-max-calculator). Enter the weight and the reps you did and get your estimated 1RM from all three formulas.

## How heavy can you go, for how many reps?

Flip the formula around and you get a useful table: the percentage of your 1RM that you can typically lift for a given number of reps. The Brzycki version is easy to remember. Every extra rep costs you around 3 percent at the start and a bit more as the reps climb.

**Share of your 1RM you can usually lift, by reps (Brzycki)**

| Reps in the set | Percent of 1RM |
| --- | --- |
| 1 | 100 |
| 2 | 97.2 |
| 3 | 94.4 |
| 5 | 88.9 |
| 8 | 80.6 |
| 10 | 75 |
| 12 | 69.4 |
| 15 | 61.1 |

*Five reps is about 89 percent of your max, ten reps about 75 percent. Individuals vary, and some people get far more reps at a given percentage than others.*

## A 1RM turned into training weights

Say your estimated 1RM is **90 kg**. Here is what common percentages look like on the bar:

| % of 1RM | Weight | Typical use |
| --- | --- | --- |
| 95% | 85.5 kg | Very heavy singles and doubles |
| 90% | 81 kg | Heavy sets of about 3 |
| 85% | 76.5 kg | Strength sets of about 5 |
| 80% | 72 kg | Solid work sets of about 8 |
| 75% | 67.5 kg | Sets of about 10 |
| 70% | 63 kg | Volume work, sets of 10 to 12 |
| 60% | 54 kg | Warm-ups and technique work |

*The 'typical use' column reflects widely used coaching guidelines, not a prescription. Your programme, experience and recovery decide what you actually do.*

**What different rep ranges are usually used for**

| Box | What it means |
| --- | --- |
| 1 to 5 reps | Heavy loads, mostly to build maximal strength. |
| 6 to 12 reps | Moderate loads, common for building muscle size. |
| 12 to 20 reps | Lighter loads, builds endurance and works well for beginners learning form. |
| Warm-up sets | 50 to 70 percent for a few reps, to rehearse the movement. |

*These ranges overlap and there is no single right answer. Muscle can be built across a wide range if the sets are hard enough.*

## When the estimate goes wrong

The formulas were fitted to many people, not to you. They also agree best at low and moderate reps. As the number of reps grows, the formulas disagree and fatigue, technique and conditioning matter more than raw strength.

**Gap between the highest and lowest formula for a 100 kg set (kg)**

|  | Gap |
| --- | --- |
| 3 reps | 5.7 kg |
| 5 reps | 5.0 kg |
| 8 reps | 3.6 kg |
| 10 reps | 7.4 kg |
| 12 reps | 15.8 kg |

*At 12 reps the three formulas differ by nearly 16 kg on a 100 kg set. That is a clear sign to use a heavier, lower-rep set when you want a trustworthy number.*

## Getting a more reliable number

1. **Use a heavy, clean set.** Aim for 3 to 6 reps with good form and a rep or two still left in the tank.
2. **Do the same lift each time.** A bench press 1RM says nothing about your squat.
3. **Track it over time.** A rising estimate matters more than any single reading.
4. **Re-test every few weeks.** Strength changes faster than you expect, especially early on.
5. **Be honest about the set.** If you stopped at 5 reps with none left, it is a better data point than 5 reps you could have done 8 of.

> **Do not test a true max alone:** If you ever want to try a real single, use a spotter or safety bars, warm up properly and stop if form breaks. For most people, the estimate from a solid set of 3 to 6 reps is accurate enough for planning and carries much less risk of injury.

## Frequently asked questions

### How do I calculate my one rep max?

Use a formula on a set you have done. Epley: 1RM = weight × (1 + reps ÷ 30). Brzycki: 1RM = weight × 36 ÷ (37 − reps). For 80 kg × 5 they give about 93 kg and 90 kg.

### Which 1RM formula is most accurate?

None is best for everyone. They tend to agree for sets of about 2 to 10 reps and drift apart above that, so use a heavy, low-rep set for the most reliable estimate.

### What percent of my 1RM is five reps?

Roughly 89 percent using the Brzycki formula. Ten reps is about 75 percent. Individual differences can be large.

### Is it safe to test my true one rep max?

It carries more risk than lighter work. Use a spotter or safety bars, warm up well and stop if form breaks. An estimate from a set of 3 to 6 reps is usually enough.

### How often should I recalculate my 1RM?

Every few weeks is common, or after a training block. Beginners often see their estimated max change quickly, so re-test more often.



---

# Overtime pay calculator: how to work out your hourly rate, the multiplier and what extra hours are worth

> Overtime is paid at a higher rate than normal hours, but the final amount depends on how your hourly rate is worked out. This guide shows the formula, how the choice of monthly hours changes the result, and what common multipliers mean.

Published 2026-10-09 by Toolcloak. 3 minute read. Original: https://toolcloak.pages.dev/articles/overtime-pay-calculator-how-to-work-out-your-hourly-rate-the-multiplier-and-what-extra-hours-are-worth
Related free tool: [Overtime Pay Calculator](https://toolcloak.pages.dev/tools/overtime-pay-calculator)

**Overtime** is work beyond your normal hours, and most rules pay it at a higher rate than ordinary time. The calculation has two parts: your normal hourly rate, and a multiplier such as 1.5 or 2. The first part is the one people argue about, because it depends on how many hours a month your salary is meant to cover.

> **In short**
> - **Hourly rate = monthly salary ÷ working hours in the month.**
> - **Overtime pay = hourly rate × multiplier × overtime hours.**
> - On ₹30,000 a month and 176 working hours, the rate is **₹170.45**. Ten overtime hours at 1.5× pay **₹2,556.82**, for a total of **₹32,556.82**.
> - The divisor matters. The same salary and hours give very different pay if the month is counted as 160, 176, 208 or 240 hours.

## The formula, step by step

**Example: ₹30,000 a month, 10 hours of overtime at 1.5×**

1. **Work out the normal hourly rate.** ₹30,000 ÷ 176 hours = **₹170.45**.
2. **Apply the multiplier.** ₹170.45 × 1.5 = **₹255.68** per overtime hour.
3. **Multiply by the overtime hours.** ₹255.68 × 10 = **₹2,556.82**.
4. **Add to the salary.** ₹30,000 + ₹2,556.82 = **₹32,556.82** for the month.

*If you are paid by the hour, enter the hourly rate and set the monthly salary to 0. The tool then skips the first step.*

Try it: [Overtime Pay Calculator](https://toolcloak.pages.dev/tools/overtime-pay-calculator). Enter your monthly salary (or hourly rate), working hours, overtime hours and the multiplier.

## What the multiplier means

| Multiplier | Overtime hourly rate | Pay for 10 hours | Total with ₹30,000 salary |
| --- | --- | --- | --- |
| 1.25× | ₹213.07 | ₹2,130.68 | ₹32,130.68 |
| 1.5× | ₹255.68 | ₹2,556.82 | ₹32,556.82 |
| 2× | ₹340.91 | ₹3,409.09 | ₹33,409.09 |
| 2.5× | ₹426.14 | ₹4,261.36 | ₹34,261.36 |

*A multiplier of 1 would pay the normal rate. Many arrangements use 1.25, 1.5 or 2, and some use different multipliers for weekdays, weekends and holidays.*

**Pay for 10 overtime hours by multiplier (₹)**

|  | Overtime pay |
| --- | --- |
| 1.25× | ₹2,131 |
| 1.5× | ₹2,557 |
| 2× | ₹3,409 |
| 2.5× | ₹4,261 |

*Going from time-and-a-half (1.5×) to double time (2×) adds about 33% to the overtime pay for the same hours.*

> **Rules differ by country and job:** In the United States, many hourly workers who are not exempt get 1.5 times their regular rate for hours beyond 40 in a week. Under India's Factories Act, factory workers are entitled to twice the ordinary rate for overtime. Shops, offices and private employers follow state laws, contracts or company policy, and some salaried roles have no overtime at all. Check your own contract and local law.

## The hidden choice: how many hours is a month?

A salary of 30,000 a month does not tell you the hourly rate until you decide how many hours it pays for. The tool's default is 176, which is 22 working days of 8 hours. A company that counts 26 days (208 hours) or a standard 30-day month (240 hours) gets a lower hourly rate from the same salary.

| Hours counted in the month | Normal hourly rate | 10 overtime hours at 1.5× |
| --- | --- | --- |
| 160 hours | ₹187.50 | ₹2,812.50 |
| 176 hours | ₹170.45 | ₹2,556.82 |
| 208 hours | ₹144.23 | ₹2,163.46 |
| 240 hours | ₹125.00 | ₹1,875.00 |

*Same salary, same 10 hours. The pay ranges from about ₹1,875 to about ₹2,813 depending only on the divisor.*

> **Ask which divisor your employer uses:** Check your appointment letter or the wage rules that apply to you. If overtime is paid per month on a different basis from the one you assumed, your own calculation will not match the payslip.

## How hours add up

| Overtime hours | Overtime pay (1.5×) | Total with salary |
| --- | --- | --- |
| 5 | ₹1,278.41 | ₹31,278.41 |
| 10 | ₹2,556.82 | ₹32,556.82 |
| 20 | ₹5,113.64 | ₹35,113.64 |
| 40 | ₹10,227.27 | ₹40,227.27 |

*Pay grows in a straight line with hours, at about ₹256 per extra hour for this salary.*

**The example month in four numbers**

|  | Value | Note |
| --- | --- | --- |
| Normal hourly rate | ₹170.45 | ₹30,000 ÷ 176 |
| Overtime hourly rate | ₹255.68 | At 1.5× |
| Overtime pay | ₹2,556.82 | 10 hours |
| Total with salary | ₹32,556.82 | Before tax and deductions |

*Overtime is part of your income, so tax and provident-fund rules may apply to it. The tool shows pay before any deductions.*

## Is overtime worth it?

Extra pay is welcome, but rest has a cost too. Compare the overtime rate with what an hour of your own time is worth, and keep an eye on how many extra hours you are doing. If overtime becomes routine, a discussion about your normal hours or a revised salary may serve you better.

## Checking your payslip

1. **Write down the hours you worked** in overtime each day or week.
2. **Find your normal hourly rate** from your contract or the way your company counts hours.
3. **Use the right multiplier** for the day: weekday, weekend or holiday.
4. **Calculate the pay** with the calculator.
5. **Compare it with the payslip** and ask payroll to explain any difference.

## Frequently asked questions

### How do I calculate overtime pay?

Divide your monthly salary by the working hours in the month to get the hourly rate, multiply by the overtime multiplier, then multiply by the overtime hours. For 30,000 a month, 176 hours, 10 overtime hours at 1.5 times, the pay is about 2,557.

### What is time and a half?

It means paying 1.5 times the normal hourly rate for overtime hours. At a normal rate of 200 an hour, time and a half is 300 an hour.

### What is double time?

Double time is a multiplier of 2, so overtime hours are paid at twice the normal rate. It is often used for holidays, rest days or very long hours.

### How many working hours are in a month?

It depends on how your employer counts. 22 days of 8 hours gives 176 hours, 26 days gives 208 and a 30-day month of 8 hours a day gives 240. A bigger number gives a smaller hourly rate.

### Can I use the calculator if I am paid by the hour?

Yes. Set the monthly salary to 0 and enter your hourly rate. The tool then uses the hourly rate directly.

### Does the result include tax?

No. The [Overtime Pay Calculator](https://toolcloak.pages.dev/tools/overtime-pay-calculator) shows gross pay. Tax and other deductions depend on your income, your country and your employer's payroll.



---

# Pascal's triangle: patterns, coin flips and the numbers behind combinations

> Start with a single 1 and add two numbers to get the one below. That tiny rule builds a triangle hiding powers of 2, the Fibonacci sequence, the coefficients of (a + b)ⁿ and the odds of every coin-flip result. This guide walks through each pattern with pictures.

Published 2026-10-09 by Toolcloak. 4 minute read. Original: https://toolcloak.pages.dev/articles/pascals-triangle-patterns-coin-flips-and-the-numbers-behind-combinations
Related free tool: [Pascal’s Triangle Generator](https://toolcloak.pages.dev/tools/pascals-triangle-generator)

**Pascal's triangle** is built with one rule: every number is the sum of the two numbers directly above it, and the edges are all 1. It is named after the French mathematician Blaise Pascal, but it was known in India, Persia and China centuries earlier.

> **In short**
> - Row n (counting from 0) lists the **combinations C(n, k)**: the number of ways to pick k things from n.
> - Each row adds up to **2ⁿ**: 1, 2, 4, 8, 16, 32…
> - Row n gives the coefficients in the expansion of **(a + b)ⁿ**.
> - Row 10 gives the exact odds of getting 0, 1, 2… 10 heads in ten coin flips.

## The first nine rows

| Row | Numbers |
| --- | --- |
| 0 | 1 |
| 1 | 1  1 |
| 2 | 1  2  1 |
| 3 | 1  3  3  1 |
| 4 | 1  4  6  4  1 |
| 5 | 1  5  10  10  5  1 |
| 6 | 1  6  15  20  15  6  1 |
| 7 | 1  7  21  35  35  21  7  1 |
| 8 | 1  8  28  56  70  56  28  8  1 |

*To get the 6 in row 4, add the 3 and 3 above it. To get the 56 in row 8, add 21 and 35 from row 7.*

**Building row 5 from row 4**

1. **Start with row 4.** 1  4  6  4  1
2. **Put a 1 at each end.** 1  _  _  _  _  1
3. **Add neighbours in pairs.** 1+4 = 5, 4+6 = 10, 6+4 = 10, 4+1 = 5
4. **Row 5 is ready.** 1  5  10  10  5  1

*No multiplication is needed. Repeated addition produces every row.*

Try it: [Pascal’s Triangle Generator](https://toolcloak.pages.dev/tools/pascals-triangle-generator). Generate as many rows as you like and highlight patterns such as odd numbers and sums.

## Pattern 1: each row sums to a power of 2

Add up any row and you get 2 raised to the row number. Row 4: 1 + 4 + 6 + 4 + 1 = 16 = 2⁴. The reason is simple: each number feeds two numbers below it, so each row doubles the one before.

**Sum of each row (rows 0 to 10)**

|  | Value |
| --- | --- |
| 0 | 1 |
| 1 | 2 |
| 2 | 4 |
| 3 | 8 |
| 4 | 16 |
| 5 | 32 |
| 6 | 64 |
| 7 | 128 |
| 8 | 256 |
| 9 | 512 |
| 10 | 1024 |

*A perfect doubling staircase. Row 10 adds up to 1,024.*

## Pattern 2: the numbers are combinations

The number in row **n**, position **k** (both counted from 0) is **C(n, k)**, the count of ways to choose k items from n without caring about order. In row 5, the 10 is C(5, 2): there are 10 ways to pick 2 people from 5.

**C(n, k) for small n and k**

| n \ k | 0 | 1 | 2 | 3 | 4 | 5 |
| --- | --- | --- | --- | --- | --- | --- |
| n = 2 | 1 | 2 | 1 |  |  |  |
| n = 3 | 1 | 3 | 3 | 1 |  |  |
| n = 4 | 1 | 4 | 6 | 4 | 1 |  |
| n = 5 | 1 | 5 | 10 | 10 | 5 | 1 |
| n = 6 | 1 | 6 | 15 | 20 | 15 | 6 |
| n = 7 | 1 | 7 | 21 | 35 | 35 | 21 |

Real examples: choosing a 5-card hand from 52 cards has C(52, 5) = **2,598,960** possibilities, and an 6-from-49 lottery has C(49, 6) = **13,983,816** possible tickets. The triangle gets there by repeated addition, and a formula does it in one step.

## Pattern 3: coin flips

Flip a fair coin four times. There are 2⁴ = 16 equally likely sequences, and the number of those with exactly k heads is C(4, k): row 4 of the triangle.

**Ways to get k heads in 4 flips (out of 16 sequences)**

|  | Value |
| --- | --- |
| 0 heads | 1 |
| 1 head | 4 |
| 2 heads | 6 |
| 3 heads | 4 |
| 4 heads | 1 |

*Two heads and two tails is the most common result, with 6 of 16 sequences (37.5%). All heads has just 1 of 16 (6.25%).*

Move to ten flips and row 10 gives the counts out of 1,024. The bell shape appears: results near 5 heads are far more common than extremes.

**Ways to get k heads in 10 flips (out of 1,024 sequences)**

|  | Value |
| --- | --- |
| 0 | 1 |
| 1 | 10 |
| 2 | 45 |
| 3 | 120 |
| 4 | 210 |
| 5 | 252 |
| 6 | 210 |
| 7 | 120 |
| 8 | 45 |
| 9 | 10 |
| 10 | 1 |

*Exactly 5 heads happens 252 times out of 1,024, about 24.6%. All heads or all tails happens 2 times out of 1,024, about 0.2%.*

**Odds in 10 fair flips**

|  | Value | Note |
| --- | --- | --- |
| Exactly 5 heads | 24.6% | 252 of 1,024 |
| 4, 5 or 6 heads | 65.6% | 672 of 1,024 |
| Outside 4 to 6 | 34.4% | 352 of 1,024 |
| Exactly 10 heads | 0.1% | 1 of 1,024 |

*Add the middle three counts, 210 + 252 + 210, to get 672. Even a fair coin gives lopsided results more often than most people expect.*

## Pattern 4: expanding (a + b)ⁿ

Multiply (a + b) by itself n times and the numbers in front of each term are row n of the triangle. For (a + b)⁴ the coefficients are 1, 4, 6, 4, 1.

| Power | Expansion |
| --- | --- |
| (a + b)² | a² + 2ab + b² |
| (a + b)³ | a³ + 3a²b + 3ab² + b³ |
| (a + b)⁴ | a⁴ + 4a³b + 6a²b² + 4ab³ + b⁴ |
| (a + b)⁵ | a⁵ + 5a⁴b + 10a³b² + 10a²b³ + 5ab⁴ + b⁵ |

*Set a = b = 1 and every coefficient adds to 2ⁿ, which is the row-sum pattern again.*

## Pattern 5: hidden Fibonacci numbers

Draw shallow diagonals through the triangle and add the numbers on each one. The sums are 1, 1, 2, 3, 5, 8, 13: the **Fibonacci sequence**. For example, 1 + 6 + 5 + 1 = 13 on one such diagonal.

**Diagonal sums spell out Fibonacci**

| When | What |
| --- | --- |
| Sum 1 | 1. A single 1. |
| Sum 2 | 1 + 1 = 2. Two 1s from neighbouring rows. |
| Sum 5 | 1 + 3 + 1 = 5. Picked from rows 2, 3 and 4. |
| Sum 8 | 1 + 4 + 3 = 8. Picked from rows 3, 4 and 5. |
| Sum 13 | 1 + 5 + 6 + 1 = 13. Picked from rows 3, 4, 5 and 6. |

*Fibonacci numbers: each is the sum of the two before it. The same numbers sit hidden inside the triangle.*

## More patterns to spot

**Quick pattern guide**

| Box | What it means |
| --- | --- |
| Hockey stick | Add a diagonal run from the edge, such as 1 + 3 + 6 + 10 = 20, and the next number down and across is the sum. |
| Triangular numbers | The third diagonal reads 1, 3, 6, 10, 15, 21: the number of dots in a triangle. |
| Powers of 11 | Rows 0 to 4 read as 1, 11, 121, 1331, 14641, which are 11⁰ to 11⁴. |
| Odd and even | Colour odd numbers and a fractal called the Sierpinski triangle appears. |

*From row 5 the entries reach 10 or more and carry over to the next place: row 5 (1, 5, 10, 10, 5, 1) becomes 161,051, which is 11⁵.*

## Frequently asked questions

### What is Pascal's triangle?

It is a triangle of numbers where each number is the sum of the two directly above it, with 1s along both edges. Its rows contain the binomial coefficients.

### How do you find a number in Pascal's triangle?

Add the two numbers above it. Or use the formula C(n, k) = n! ÷ (k! × (n − k)!) for the number in row n, position k, counting both from zero.

### What do the row sums of Pascal's triangle equal?

Each row sums to 2 raised to the row number. Row 5 adds up to 32 and row 10 adds up to 1,024.

### How is Pascal's triangle used in probability?

Row n gives the number of ways to get k successes in n fair yes-or-no trials, such as k heads in n coin flips. Divide each entry by 2ⁿ to get its probability.

### Where is the Fibonacci sequence in Pascal's triangle?

Add the numbers along the shallow diagonals. The sums are 1, 1, 2, 3, 5, 8, 13 and so on.

### Who invented Pascal's triangle?

Blaise Pascal wrote about it in the 1650s, but earlier mathematicians in India, Persia and China studied it long before. In China it is often called Yang Hui's triangle.



---

# Percent error: how far off is your measurement, and how it differs from percent change

> You measured gravity as 9.6 m/s² and the textbook says 9.81. How wrong are you, in percent? This guide shows the one-line formula, three worked examples, how it differs from percent difference and percent change, and what size of error is acceptable.

Published 2026-10-09 by Toolcloak. 3 minute read. Original: https://toolcloak.pages.dev/articles/percent-error-how-far-off-is-your-measurement-and-how-it-differs-from-percent-change
Related free tool: [Percent Error Calculator](https://toolcloak.pages.dev/tools/percent-error-calculator)

**Percent error** tells you how far a measured or estimated value is from the true (accepted) value, as a percentage of that true value. It turns a gap like '0.21 off' into something you can compare across experiments: 2% off is easy to judge, whatever the units.

> **In short**
> - **Percent error = |measured − true| ÷ |true| × 100.** The bars mean you drop any minus sign.
> - Divide by the **true** value, not the measured one. Swapping them gives a different answer.
> - It needs a known true value. If you are comparing two measurements with no reference, use **percent difference** instead.
> - Low percent error means accurate. Repeating the same wrong number is precise but not accurate.

## The formula in three steps

**Measured g = 9.6 m/s², accepted g = 9.81 m/s²**

1. **Find the gap.** 9.6 − 9.81 = −0.21. Drop the sign: 0.21.
2. **Divide by the true value.** 0.21 ÷ 9.81 = 0.0214.
3. **Multiply by 100.** 0.0214 × 100 = **2.14%**.

*The result says your measurement was about 2.14% below the accepted value. Some teachers keep the sign to show whether you were over or under; the calculator can show both.*

Try it: [Percent Error Calculator](https://toolcloak.pages.dev/tools/percent-error-calculator). Enter the measured and true values to get percent error, with the steps shown.

## Eight examples

| Measured | True value | Gap | Percent error |
| --- | --- | --- | --- |
| 9.6 | 9.81 | 0.21 | 2.14% |
| 101.2 | 100 | 1.2 | 1.20% |
| 48 | 50 | 2 | 4.00% |
| 0.52 | 0.5 | 0.02 | 4.00% |
| 298 | 299.792 | 1.792 | 0.60% |
| 3.14 | 3.14159 | 0.00159 | 0.05% |
| 72 | 70 | 2 | 2.86% |
| 1.9 | 2.0 | 0.1 | 5.00% |

*Look at rows 3 and 4. A gap of 2 and a gap of 0.02 both give 4%, because each is 4% of its own true value. That is the point of using a percentage.*

**Percent error in each example (%)**

|  | Value |
| --- | --- |
| g | 2.14% |
| 100 | 1.2% |
| 50 | 4% |
| 0.5 | 4% |
| c | 0.6% |
| π | 0.05% |
| 70 | 2.86% |
| 2.0 | 5% |

*Even 3.14 for pi, which looks rough, is only 0.05% off the real value.*

## Percent error, percent difference and percent change

Three formulas look alike and are used for different jobs. The difference is what you divide by.

| Name | When to use it | Divide by | Example |
| --- | --- | --- | --- |
| Percent error | Compare a measurement with a known true value | The true value | 9.6 vs 9.81 → 2.14% |
| Percent difference | Compare two measurements, neither is 'right' | Their average | 9.6 vs 9.81 → 2.16% |
| Percent change | How much something grew or shrank over time | The old value | 50 to 60 → +20% |

*Percent error and percent difference are close when the numbers are close, but they are not the same thing.*

**Same two numbers, 50 and 60, three different answers**

|  | Percent |
| --- | --- |
| Percent change, 50 to 60 | 20% |
| Percent error, measured 50, true 60 | 16.7% |
| Percent difference, 50 and 60 | 18.2% |

*Percent change divides by 50, error by 60 and difference by the average, 55. Always say which one you mean.*

## Why the direction matters

**Dividing by the wrong number**

| Divide by the true value | Divide by the measured value |
| --- | --- |
| 4.00% | 4.17% |
| Measured 48, true 50 | Same gap 2 ÷ 48 = 0.0417 |
| Gap 2 ÷ 50 = 0.04 | Gives a slightly different, wrong answer |
| This is the correct percent error | Small here, bigger when the error is large |

*The true value is the fixed reference, so it always goes at the bottom of the fraction.*

## How big an error is acceptable?

There is no universal limit. It depends on the instrument and the purpose. A school pendulum experiment with a stopwatch often lands in the 1% to 5% range. A high-precision lab measurement aims for a tiny fraction of 1%.

**A rough guide for a school lab**

| Zone | From | To |
| --- | --- | --- |
| Excellent | 0 | 2 |
| Good | 2 | 5 |
| Check method | 5 | 10 |
| Likely mistake | 10 | 15 |

*Bands vary by teacher and by experiment. Treat them as a conversation starter, not a grade.*

## Accuracy against precision

**Percent error measures accuracy, not precision**

| Box | What it means |
| --- | --- |
| Accurate and precise | Readings cluster together right on the true value. Percent error is small. |
| Precise, not accurate | Readings agree with each other but sit away from the truth, for example a mis-zeroed scale. Percent error is large. |
| Accurate, not precise | Readings scatter around the true value. The average is close; each reading is not. |
| Neither | Scattered and off-target. Check method, tools and recording. |

*Taking several readings and averaging them reduces random scatter. It does not remove a built-in bias.*

## Common sources of error

1. **Reading the instrument,** such as looking at a ruler from an angle.
2. **Zero error,** where a scale or timer does not start at zero.
3. **Rounding too early,** which adds error to every later step.
4. **Reaction time** with hand-held stopwatches.
5. **Wrong true value,** for example using 9.8 for gravity in a place where 9.78 is more accurate.

> **When the true value is zero:** Percent error cannot be calculated when the accepted value is 0, because you would divide by zero. Report the absolute error (the plain gap) instead.

## Frequently asked questions

### What is the percent error formula?

Percent error = |measured value − true value| ÷ |true value| × 100. It expresses the gap as a percentage of the accepted value.

### Can percent error be negative?

With the absolute value bars it is always positive. Some courses omit the bars to show direction: negative means the measurement was below the true value.

### What is the difference between percent error and percent difference?

Percent error compares a measurement with a known true value and divides by that value. Percent difference compares two measurements and divides by their average.

### What is a good percent error?

It depends on the experiment. For many school labs under 5 percent is good and under 2 percent is excellent, while precision science aims for far smaller values.

### Why can't I calculate percent error when the true value is zero?

Because the formula divides by the true value. Dividing by zero is undefined, so report the absolute error instead.

### Is percent error the same as percent change?

No. Percent change compares an old and a new value over time and divides by the old value. Percent error compares a measurement with a true reference value.



---

# The Pomodoro method, with the numbers: what 25 minutes on, 5 off really adds up to.

> Eight rounds is a full working session, but it is four hours on the clock, not three and a half. How the cycle adds up, why 50/10 and 15/3 have the same ratio, and how to set a timer you will actually stick with.

Published 2026-10-09 by Toolcloak. 3 minute read. Original: https://toolcloak.pages.dev/articles/pomodoro-25-minutes-the-numbers-behind-the-method
Related free tool: [Pomodoro Timer (Focus & Break)](https://toolcloak.pages.dev/tools/pomodoro-timer)

The Pomodoro method is almost embarrassingly simple: work for 25 minutes, rest for 5, repeat. People have used it for decades to study, write and finish tasks they were avoiding. It works well when you know what a round really costs and a little less well when you expect it to be magic.

The name is Italian for tomato. Francesco Cirillo, a university student at the time, used a tomato-shaped kitchen timer to push himself through his studies in the late 1980s, and the name stuck. The method is a way of making a big task start, not a scientific constant.

## One cycle, one day

The standard rhythm: four focus rounds of 25 minutes, with a 5-minute break between them, and a longer break of 15 to 30 minutes after the fourth. Here is what that costs on the clock, using a 20-minute long break.

| Piece | Minutes | Running total |
| --- | --- | --- |
| 4 focus rounds of 25 | 100 | 100 |
| 3 short breaks of 5 | 15 | 115 |
| 1 long break | 20 | 135 (2 h 15 min) |

*One full set is 2 hours 15 minutes of clock time and 100 minutes of focus, which is 74% of the time.*

A day of eight rounds is two sets with the last break skipped: 8 × 25 = 200 minutes of focus, plus 3 short breaks, a long break, and 3 more short breaks, which totals 250 minutes. That is 4 hours 10 minutes on the clock for 3 hours 20 minutes of real work: 80% focus.

**A full eight-round day**

|  | Value | Note |
| --- | --- | --- |
| Focused work | 3 h 20 | 8 rounds of 25 minutes |
| Time on the clock | 4 h 10 | Including every break |
| Share that is focus | 80% | 200 of 250 minutes |

*Plan your day by the clock total, not by the focus total, or you will always feel behind.*

## 25 and 5 are a convention, not a law

There is no number that is right for everybody. Shorter rounds suit people who struggle to start, and longer ones suit people who lose their flow when interrupted. Look at the ratios and you will see that the popular variants are close cousins.

| Rhythm | Focus | Break | Share spent focusing |
| --- | --- | --- | --- |
| 15 / 3 | 15 min | 3 min | 83% |
| 25 / 5 | 25 min | 5 min | 83% |
| 50 / 10 | 50 min | 10 min | 83% |
| 25 / 5 over a full set | 100 min | 35 min | 74% |

*The three common rhythms all spend five-sixths of their time on focus. The long break is what pulls a full set down to 74%.*

If 25 minutes feels too short for the work you are doing, try 50 and 10. If it feels too long on a day when you can barely start, try 15 and 3. The rule is not the number, it is that the break is real: standing up, water, looking out of a window. Not a scroll through your phone.

## What makes it work

- **A small, clear start.** “Work for 25 minutes” is easier to begin than “finish the report”.
- **One task per round.** If something else pops into your head, write it on a list and go back.
- **A visible end.** A countdown makes the effort feel limited, which is a relief when you are dreading the task.
- **A break you protect.** Skipping breaks feels productive for an hour and then costs you the afternoon.

> **If you get interrupted:** A round that is broken by something urgent does not count. Note it, deal with the interruption, and start a fresh round afterwards. Do not carry half a round forward and pretend it was a full one.

## Make your own first week

**Setting up a rhythm you will keep**

1. **Start with 25 / 5.** Do not tune it yet. Run four rounds on one real task.
2. **Notice where it breaks.** Too short, you were just getting going. Too long, you were drifting by minute 18.
3. **Adjust once.** Change focus length by 10 minutes at most. Keep the breaks real.
4. **Count rounds, not hours.** Aim for a number of rounds per day that fits your schedule, like 4 or 6.

Pomodoro also pairs well with sleep. A tired brain burns through rounds slowly, and [this article on sleeping in 90-minute cycles](https://toolcloak.pages.dev/articles/sleep-in-90-minute-cycles-when-to-go-to-bed) shows how to time bedtime so the next day's rounds start fresher. And if your work is typing, [what counts as fast typing](https://toolcloak.pages.dev/articles/what-counts-as-fast-typing) helps you set a realistic target per round.

Try it: [Pomodoro Timer (Focus & Break)](https://toolcloak.pages.dev/tools/pomodoro-timer). Set your focus and break lengths, start a round, and let it switch to the break for you. It keeps counting rounds for the day.

## Frequently asked questions

### What is the Pomodoro technique?

A time-management method: work for a fixed block, typically 25 minutes, then take a short 5-minute break. After four rounds you take a longer break of 15 to 30 minutes. It was created by Francesco Cirillo using a tomato-shaped kitchen timer.

### Is 25 minutes the best focus length?

It is the traditional choice, not a proven optimum. Some people prefer 15 minutes, some 50. Pick the shortest length that lets you get properly into the task.

### How long is a full Pomodoro cycle?

Four 25-minute rounds, three 5-minute breaks and a 20-minute long break come to 135 minutes, or 2 hours 15 minutes, of which 100 minutes are focus.

### What should I do in the break?

Stand up, drink water, stretch or look at something far away. Avoid anything that pulls you into another task, such as email or social media.

### What if I finish the task before the timer ends?

Use the rest of the round to review or improve what you just did, or start the next small step. Do not stop the round early to start something unrelated.



---

# Pro-rata calculator: how to work out rent, salary or fees for part of a month or year

> When someone joins, leaves or moves out in the middle of a period, the full amount has to be split fairly. This guide shows how pro-rata works, how the choice of days in the month changes the result, and where it is used for rent, pay and subscriptions.

Published 2026-10-09 by Toolcloak. 3 minute read. Original: https://toolcloak.pages.dev/articles/pro-rata-calculator-how-to-work-out-rent-salary-or-fees-for-part-of-a-month-or-year
Related free tool: [Pro-Rata Calculator](https://toolcloak.pages.dev/tools/pro-rata-calculator)

**Pro rata** means in proportion. If you pay for a whole month but use only part of it, you pay the same share of the amount as the share of the month you use. The method is the same for rent, salary, a gym fee or a software plan: find the amount for one day, then multiply by the days that count.

> **In short**
> - **Pro-rated amount = full amount × days charged ÷ days in the full period.**
> - A full amount of **₹30,000** for **12 days out of 30** is **₹12,000**. That is 40% of the period, at ₹1,000 a day.
> - The remaining amount is ₹18,000.
> - The number of days you divide by is a **choice** that the contract should state. Different choices give different amounts.

## The formula, step by step

**Example: ₹30,000 a month, 12 days charged**

1. **Find the daily amount.** ₹30,000 ÷ 30 days = **₹1,000** a day.
2. **Multiply by the days charged.** ₹1,000 × 12 = **₹12,000**.
3. **Find the share of the period.** 12 ÷ 30 = **40%**.
4. **Find what is left.** ₹30,000 − ₹12,000 = **₹18,000**.

*The tool shows the pro-rated amount, the amount per day, the remaining amount and the share of the period.*

Try it: [Pro-Rata Calculator](https://toolcloak.pages.dev/tools/pro-rata-calculator). Enter the full amount, the days in the full period and the days to charge to get the pro-rated amount.

## How much for different numbers of days

| Days charged (out of 30) | Pro-rated amount | Share of the period |
| --- | --- | --- |
| 5 | ₹5,000 | 16.7% |
| 10 | ₹10,000 | 33.3% |
| 12 | ₹12,000 | 40.0% |
| 15 | ₹15,000 | 50.0% |
| 20 | ₹20,000 | 66.7% |
| 25 | ₹25,000 | 83.3% |

*Same ₹30,000 and a 30-day period. The amount grows in a straight line with the days.*

**Pro-rated amount by days charged (₹)**

|  | Amount |
| --- | --- |
| 5 days | ₹5,000 |
| 10 days | ₹10,000 |
| 15 days | ₹15,000 |
| 20 days | ₹20,000 |
| 25 days | ₹25,000 |

*Each day adds ₹1,000 here. Halving the days halves the amount.*

## The hidden choice: how many days in the period?

A month has 28 to 31 days, and a contract may say to use the real number of days, or a fixed 30, or even the working days. The same ₹30,000 and 12 days give different amounts:

| Days used as the divisor | Amount per day | Amount for 12 days |
| --- | --- | --- |
| 28 days | ₹1,071.43 | ₹12,857.14 |
| 30 days | ₹1,000.00 | ₹12,000.00 |
| 31 days | ₹967.74 | ₹11,612.90 |

*The difference between 28 and 31 days is ₹1,244.24 for the same 12 days. Neither is wrong. The agreement decides.*

> **Agree the method first:** For rent, notice pay or final salary, write down whether you use the actual days in the month, a fixed 30 days or working days. It stops an argument over a few hundred rupees later.

## Where pro rata is used

- **Rent and deposits,** when a tenant moves in or out mid-month.
- **Salary,** for a joiner or leaver who works part of a month.
- **Subscriptions and memberships,** for a plan started or cancelled partway through.
- **Fees and insurance,** where cover or a service covers only part of a term.
- **Annual amounts,** such as a yearly plan used for a number of days.

**Two ways to use the same tool**

| Part of a month | Part of a year |
| --- | --- |
| ₹12,000 | ₹3,287.67 |
| ₹30,000 for 12 of 30 days | ₹12,000 plan used for 100 of 365 days |
| Joiner, leaver or tenant | Set the period to 365 days |
| Per day: ₹1,000 | Per day: ₹32.88 |

*The calculator does not care about the length of the period. Enter the days of the full period, whether 30, 31 or 365.*

**The example month in four numbers**

|  | Value | Note |
| --- | --- | --- |
| Pro-rated amount | ₹12,000 | 12 of 30 days |
| Per day | ₹1,000 | ₹30,000 ÷ 30 |
| Remaining amount | ₹18,000 | The unused part |
| Share of period | 40% | 12 ÷ 30 |

*The pro-rated amount and the remaining amount always add up to the full amount.*

## A short routine

1. **Write down the full amount** and the length of the full period in days.
2. **Count the days to be charged,** and decide whether the first and last days count.
3. **Agree the divisor** (actual days, 30 or working days).
4. **Run the calculator** and check that the amount and the remainder add up.
5. **Round to the nearest rupee** only at the end.

## Frequently asked questions

### How do I calculate pro-rata pay?

Divide the full amount by the number of days in the full period, then multiply by the days to be paid. For 30,000 a month and 12 days out of 30, it is 30,000 ÷ 30 × 12 = 12,000.

### Should I divide by 30 or the real days of the month?

It depends on your contract or policy. The real days give a slightly different amount in months of 28, 29 or 31 days. Agree the method first.

### How do I pro-rate an annual amount?

Enter the annual amount, 365 days (or 366 in a leap year) as the full period, and the days to charge. A 12,000 plan used for 100 days comes to about 3,287.67.

### What does the remaining amount mean?

It is the part of the full amount for the days that are not charged. With 30,000 and 12 days charged out of 30, the remaining amount is 18,000.

### Can days to charge be more than the full period?

No. The tool asks for a number between 0 and the length of the full period.

### Is the pro-rata amount the same as a refund?

Not always. A refund may also include fees or a minimum charge. For a daily figure alone, you can also use the [Time Calculator](https://toolcloak.pages.dev/tools/time-calculator) or the [Date Difference Calculator](https://toolcloak.pages.dev/tools/date-difference-calculator) to count the days.



---

# Profit margin calculator: how to work out margin, profit and markup, and set a price for a target margin

> Margin and markup sound alike but give different percentages for the same sale. This guide shows how to calculate profit, margin and markup from cost and selling price, and how to find the price that gives the margin you want.

Published 2026-10-09 by Toolcloak. 3 minute read. Original: https://toolcloak.pages.dev/articles/profit-margin-calculator-how-to-work-out-margin-profit-and-markup-and-set-a-price-for-a-target-margin
Related free tool: [Profit Margin Calculator](https://toolcloak.pages.dev/tools/profit-margin-calculator)

**Profit margin** tells you what share of the selling price is profit. **Markup** tells you how much you added on top of the cost. They use the same two numbers, cost and selling price, but divide by different things, so the same sale shows a margin of 25% and a markup of 33%. Mixing them up is one of the commonest pricing mistakes.

> **In short**
> - **Profit = selling price − cost.**
> - **Margin = profit ÷ selling price × 100.** **Markup = profit ÷ cost × 100.**
> - Cost ₹600 and selling price ₹800 give profit ₹200, a margin of **25.00%** and a markup of **33.33%**.
> - To get a **30% margin** on a cost of ₹600, the price is cost ÷ (1 − 0.30) = **₹857.14**.

## The formulas, step by step

**Example: cost ₹600, selling price ₹800**

1. **Find the profit.** ₹800 − ₹600 = **₹200**.
2. **Margin on the selling price.** ₹200 ÷ ₹800 × 100 = **25.00%**.
3. **Markup on the cost.** ₹200 ÷ ₹600 × 100 = **33.33%**.
4. **Price for a 30% margin.** ₹600 ÷ (1 − 0.3) = **₹857.14**.

*Enter a target margin in the optional field and the tool also shows the price that would give it. If the selling price is below cost it says you are selling at a loss.*

Try it: [Profit Margin Calculator](https://toolcloak.pages.dev/tools/profit-margin-calculator). Enter your cost and selling price, and an optional target margin, to see profit, margin, markup and the target price.

## Margin and markup are not the same

**The same sale, two percentages**

| Margin | Markup |
| --- | --- |
| 25% | 33.3% |
| ₹200 profit out of ₹800 sold | ₹200 profit on a ₹600 cost |
| Divides by the selling price | Divides by the cost |
| Can never reach 100% | Can go above 100% |

*Margin is always the smaller number for a profitable sale, because the selling price is larger than the cost.*

| Markup on cost | Selling price for a ₹600 cost | Margin you actually get |
| --- | --- | --- |
| 10% | ₹660 | 9.1% |
| 25% | ₹750 | 20.0% |
| 50% | ₹900 | 33.3% |
| 100% | ₹1,200 | 50.0% |

*A 50% markup is a 33.3% margin, and a 100% markup (doubling the cost) is a 50% margin.*

## Going the other way: from the margin you want to a price

If you want a given margin, do not just add that percentage to the cost. A 30% markup gives only a 23% margin. The right formula is **price = cost ÷ (1 − margin)**. That is why the tool asks for the target margin and gives you the price.

| Target margin | Price for a ₹600 cost | Same thing as a markup of |
| --- | --- | --- |
| 20% | ₹750.00 | 25.0% |
| 30% | ₹857.14 | 42.9% |
| 40% | ₹1,000.00 | 66.7% |
| 50% | ₹1,200.00 | 100.0% |

*Targets above 50% need the price to be more than double the cost. A margin of 100% is impossible, and the tool rejects it.*

**Price needed on a ₹600 cost, by target margin (₹)**

|  | Selling price |
| --- | --- |
| 20% | ₹750 |
| 30% | ₹857 |
| 40% | ₹1,000 |
| 50% | ₹1,200 |

*Each extra 10 points of margin needs a bigger price jump than the one before it.*

## What the margin leaves out

- **Other costs.** If cost means only the purchase or production cost, this is a gross margin. Rent, wages, delivery, payment fees and returns come out of it.
- **Tax.** Enter prices without GST, or with it on both sides, so that the tax does not distort the margin.
- **Discounts.** A sale price that is lower than the list price changes the margin. Use the price the customer actually pays.
- **Volume.** A low margin on a fast-moving product can earn more in total than a high margin on a slow one.

> **Check which one a quote means:** When a supplier or a colleague says '30% profit', ask whether it is on cost or on selling price. The same words can mean a price that differs by several percent.

**The example sale in four numbers**

|  | Value | Note |
| --- | --- | --- |
| Profit | ₹200 | ₹800 − ₹600 |
| Margin | 25.00% | Profit ÷ selling price |
| Markup | 33.33% | Profit ÷ cost |
| Price for 30% margin | ₹857.14 | Cost ÷ (1 − 0.30) |

*Use the margin to compare products and the markup to set a price from a known cost.*

## A pricing routine

1. **List the full cost** of one unit, not just the purchase price.
2. **Decide the margin** you need to cover running costs and leave a profit.
3. **Work out the price** with cost ÷ (1 − margin).
4. **Check it against the market,** and against what customers will pay.
5. **Recalculate** whenever your costs or discounts change.

## Frequently asked questions

### How do I calculate profit margin?

Subtract the cost from the selling price to get the profit, divide by the selling price and multiply by 100. For a cost of 600 and a selling price of 800, the margin is 200 ÷ 800 = 25%.

### What is the difference between margin and markup?

Margin divides profit by the selling price. Markup divides profit by the cost. For the same sale the markup is always larger, 33.33% against 25% in this example.

### How do I find the selling price for a target margin?

Divide the cost by one minus the margin as a decimal. For a cost of 600 and a margin of 30%, the price is 600 ÷ 0.70 = 857.14.

### Can the margin be more than 100%?

No. Profit can never be more than the selling price, so margin stays below 100%. Markup can be above 100% when the price is more than double the cost.

### What happens if I sell below cost?

The profit and the margin become negative, and the tool tells you that you are selling at a loss.

### Is this the same as the markup calculator?

It shows both numbers. If you only need to add a markup to a cost, the [Markup and Margin Calculator](https://toolcloak.pages.dev/tools/markup-margin-calculator) is built for that. The [Profit Margin Calculator](https://toolcloak.pages.dev/tools/profit-margin-calculator) also finds the price for a target margin.



---

# Projectile motion: why 45 degrees throws farthest (and when it does not)

> Throw a ball, kick a football or launch a water rocket and the path is the same curve. This guide explains it with one rule, five launch angles, four planets and a few real numbers, so you can predict range, height and flight time before you throw.

Published 2026-10-09 by Toolcloak. 4 minute read. Original: https://toolcloak.pages.dev/articles/projectile-motion-why-45-degrees-throws-farthest-and-when-it-does-not
Related free tool: [Projectile Motion Calculator (Range, Height, Time)](https://toolcloak.pages.dev/tools/projectile-motion-calculator)

Once a ball leaves your hand, nothing pushes it forward any more. It keeps moving sideways at a steady speed while gravity pulls it down, a little more each second. Put those two simple ideas together and you get a smooth curve called a **parabola**.

> **In short**
> - Sideways motion and up-down motion are **independent**. Sideways speed stays constant; vertical speed changes by 9.81 m/s every second.
> - On level ground the range is **R = v² × sin(2θ) ÷ g**. It is largest at **45°** and equal for angles that add up to 90°, like 30° and 60°.
> - Doubling the launch speed makes the range **four times** longer.
> - Air resistance, a launch from a height and a different planet all change the answer. Gravity matters most.

## Two motions at once

**How to split any throw into two easy problems**

1. **Split the speed.** A launch speed v at angle θ gives a sideways part vx = v cos θ and an upward part vy = v sin θ.
2. **Sideways: no force.** vx never changes (we ignore air). Distance sideways = vx × time.
3. **Up and down: gravity.** vy drops by g = 9.81 m/s every second. Up for a while, then down.
4. **Join them.** Use the same time for both. The ball lands when its height is back to zero.

*This is the trick behind every projectile problem in school physics. The calculator below does the same four steps.*

Try it: [Projectile Motion Calculator (Range, Height, Time)](https://toolcloak.pages.dev/tools/projectile-motion-calculator). Set the speed, angle and starting height to see range, peak height and time in air.

## Five angles, one speed

Take a ball thrown at **20 m/s** from the ground on Earth. Only the angle changes. Notice how 15° and 75° land at the same spot, and so do 30° and 60°, but the high throws spend much longer in the air.

| Angle | Range | Peak height | Time in air |
| --- | --- | --- | --- |
| 15° | 20.4 m | 1.4 m | 1.06 s |
| 30° | 35.3 m | 5.1 m | 2.04 s |
| 45° | **40.8 m** | 10.2 m | 2.88 s |
| 60° | 35.3 m | 15.3 m | 3.53 s |
| 75° | 20.4 m | 19.0 m | 3.94 s |

*Range = v² sin(2θ) ÷ g and peak = v² sin²θ ÷ 2g with v = 20 and g = 9.81. Rounded to one decimal.*

**Range of a 20 m/s throw by launch angle (metres)**

|  | Value |
| --- | --- |
| 15° | 20.4 m |
| 30° | 35.3 m |
| 45° | 40.8 m |
| 60° | 35.3 m |
| 75° | 20.4 m |

*The bars rise to a peak at 45° and fall again in a mirror image. The mirror pairs always add up to 90°.*

**Height along the path for three angles (same 20 m/s speed)**

| Distance sideways (m) | 30° | 45° | 60° |
| --- | --- | --- | --- |
| 0 | 0 | 0 | 0 |
| 5 | 2.5 | 4.4 | 7.4 |
| 10 | 4.1 | 7.5 | 12.4 |
| 15 | 5 | 9.5 | 14.9 |
| 20 | 5 | 10.2 | 15 |
| 25 | 4.2 | 9.7 | 12.6 |
| 30 | 2.6 | 7.9 | 7.8 |
| 35 | 0.2 | 5 | 0.5 |
| 40 | 0 | 0.8 | 0 |

*The 30° and 60° paths touch the ground at nearly the same place (about 35 m), but the 60° ball climbs three times higher. The 45° path travels farthest. Values after landing are shown as zero.*

## Why 45 degrees wins

A steeper throw stays in the air longer but spends its speed going up. A flatter throw keeps more sideways speed but lands quickly. The product of the two, sideways speed times time, is biggest when they are balanced, and that balance point is exactly 45° on level ground.

**Flat throw against high throw**

| Low angle (15°) | 45° throw |
| --- | --- |
| Fast but short-lived | Balanced |
| Most speed goes sideways | Speed split equally |
| Hits the ground in about 1 second | About 2.9 seconds in the air |
| Range: 20.4 m | Range: 40.8 m, double the 15° throw |

*A 75° throw has the mirror problem: lots of time, hardly any sideways speed.*

## Speed matters more than angle

Because the range formula has v², small speed gains give big distance gains. At 45° on Earth, 10 m/s gets you about 10 m, but 40 m/s gets you about 163 m, sixteen times more for four times the speed.

**Range at 45° for different launch speeds**

|  | Range |
| --- | --- |
| 10 m/s | 10.2 m |
| 20 m/s | 40.8 m |
| 30 m/s | 91.7 m |
| 40 m/s | 163.1 m |

*Double the speed, four times the range. This is why a small increase in a cricket bowler's or a golfer's speed matters so much.*

## Same throw, different worlds

Gravity sets how fast the ball falls back. With the same 20 m/s, 45° throw, the range is simply v² ÷ g, so a weaker pull means a much longer flight.

**Range of a 20 m/s, 45° throw on different worlds**

|  | Range |
| --- | --- |
| Jupiter (g 24.8) | 16 m |
| Earth (g 9.81) | 41 m |
| Mars (g 3.71) | 108 m |
| Moon (g 1.62) | 247 m |

*On the Moon the same throw lasts about 17.5 seconds and travels nearly 250 metres. These are ideal numbers with no air; Jupiter has no ground to stand on, so treat it as a thought experiment.*

## When 45 degrees is not the best

The 45° rule needs the ball to **start and land at the same height**. Throw from a rooftop, a cliff or a bowler's raised arm and the best angle drops below 45°, because the ball has extra time to fall. A shot-putter releases at about 2 m above ground and the ideal angle is a little lower than 45° (for 20 m/s it is about 43.7°). In real sport, air drag, spin and a body that cannot throw as hard at steep angles pull the best angle lower still: long jumpers take off near 20°, and good javelin throws leave at around 30° to 35°.

**What moves the best angle**

| Box | What it means |
| --- | --- |
| Release higher than landing | Best angle falls below 45°. A rooftop throw prefers a flatter angle. |
| Release lower than landing | Best angle rises above 45°, like throwing up onto a platform. |
| Air drag | Steep paths lose more speed, so practical best angles are lower than 45°. |
| Body mechanics | People can push harder at lower angles, which favours flatter throws. |

*The ideal formula is a starting point. Real throwing sports tune the angle by trial and by feel.*

> **Quick sanity check:** Throw at 45° and the range equals the peak height times four. In the table, 10.2 m peak and 40.8 m range fit that rule. If your numbers break it, one of them has a typing error.

## Using the calculator well

1. **Use consistent units.** Metres and seconds, or feet and seconds, never a mix.
2. **Check the starting height.** Zero means level ground; a positive number means a raised launch.
3. **Try the mirror angle.** Enter 30° then 60° and watch the range stay the same while time and height change.
4. **Treat results as ideal.** Wind, spin and drag will shorten real throws, especially fast ones.

## Frequently asked questions

### What is projectile motion?

It is the motion of an object thrown or launched into the air that moves only under gravity. Its path is a parabola because the sideways speed is constant while gravity changes the vertical speed.

### Why does 45 degrees give the longest range?

On level ground the range is v² sin(2θ) ÷ g. The sine is largest, equal to 1, when 2θ is 90 degrees, so θ is 45 degrees.

### What is the formula for time of flight?

For a launch from the ground, time in the air is 2 × v × sin θ ÷ g. At 20 m/s and 45° that is about 2.88 seconds on Earth.

### Do two different angles give the same range?

Yes. Any two angles that add up to 90 degrees, such as 30° and 60°, give the same range on level ground. The higher one stays in the air longer.

### How does air resistance change things?

Drag slows the object, shortens the range and lowers the best launch angle. The calculator assumes no air, so real throws, especially fast and light ones, fall shorter.

### What is the acceleration due to gravity?

On Earth it is about 9.81 m/s² near the surface, often rounded to 9.8 or 10. It is about 3.71 on Mars and 1.62 on the Moon.



---

# Pythagorean theorem: the 3-4-5 trick, ladders, TV sizes and how to check a right angle

> a² + b² = c² is the most useful line in school geometry. This guide shows what it really says, how carpenters use a 3-4-5 rope to square a room, why a '55-inch' TV is not 55 inches wide and how to test whether any triangle has a right angle.

Published 2026-10-09 by Toolcloak. 4 minute read. Original: https://toolcloak.pages.dev/articles/pythagorean-theorem-3-4-5-triples-and-how-to-check-a-right-angle
Related free tool: [Pythagorean Theorem Calculator](https://toolcloak.pages.dev/tools/pythagorean-theorem-calculator)

In a triangle with one **right angle** (90 degrees), the square of the longest side equals the sum of the squares of the other two. Written as a formula: **a² + b² = c²**, where c is the longest side, called the **hypotenuse**. It works for every right triangle, from a rooftop to a phone screen.

> **In short**
> - c is always the side **opposite the right angle** and always the longest. The two shorter sides are called legs.
> - To find c: add the squares of the legs, then take the square root. To find a leg: subtract, then take the square root.
> - The **3-4-5** triangle (and its multiples 6-8-10, 9-12-15) gives a right angle with nothing but a tape measure.
> - The rule only works for right triangles. If a² + b² is not equal to c², the angle is not 90 degrees.

## What the formula actually says

Draw a square on each side of a right triangle. The area of the big square on the hypotenuse is exactly the sum of the areas of the two smaller squares. For the 3-4-5 triangle that is 9 + 16 = 25. Nothing is approximate here; it is an exact fact of flat geometry.

**The squares on the sides of a 3-4-5 triangle (areas)**

|  | Square on leg a (3 × 3) | Square on leg b (4 × 4) |
| --- | --- | --- |
| Two legs together | 9 | 16 |
| Hypotenuse (5 × 5) | 25 |

*Both bars are the same length: 25. That equality is the whole theorem, drawn as bars.*

## Three steps to the missing side

**Finding the hypotenuse of a triangle with legs 5 and 12**

1. **Square each leg.** 5² = 25 and 12² = 144.
2. **Add them.** 25 + 144 = 169.
3. **Take the square root.** √169 = 13. The hypotenuse is 13.

*To find a leg instead, subtract: if c = 13 and a = 5, then b² = 169 − 25 = 144, so b = 12.*

Try it: [Pythagorean Theorem Calculator](https://toolcloak.pages.dev/tools/pythagorean-theorem-calculator). Enter any two sides and get the third, with the working shown step by step.

## Pythagorean triples: when all three sides are whole numbers

Most right triangles have a messy hypotenuse. Legs of 1 and 1 give √2 = 1.4142… forever. A few special sets of whole numbers fit perfectly. They are called **Pythagorean triples**, and every multiple of a triple is another triple.

| Triple | Check | Handy multiples |
| --- | --- | --- |
| 3, 4, 5 | 9 + 16 = 25 | 6-8-10, 9-12-15, 30-40-50 |
| 5, 12, 13 | 25 + 144 = 169 | 10-24-26, 15-36-39 |
| 8, 15, 17 | 64 + 225 = 289 | 16-30-34 |
| 7, 24, 25 | 49 + 576 = 625 | 14-48-50 |
| 20, 21, 29 | 400 + 441 = 841 | 40-42-58 |
| 9, 40, 41 | 81 + 1,600 = 1,681 | 18-80-82 |

*If your two legs and hypotenuse match one of these (or a multiple), you can skip the square root completely.*

## The carpenter's trick: squaring a corner with a rope

Builders marking a foundation, gardeners laying a bed and tailors cutting cloth all need a true 90-degree corner. Measure **3 units** along one edge and **4 units** along the other. If the straight distance between those two marks is exactly **5 units**, the corner is square. For big jobs use 6-8-10 or 9-12-15 so small errors matter less.

**Squaring a corner: by eye or with 3-4-5**

| By eye | 3-4-5 check |
| --- | --- |
| Maybe 88° | Exact 90° |
| Looks fine from one side | Mark 3 m and 4 m from the corner |
| A 2° error becomes about 5 cm over a 3 m wall | The diagonal between marks must be 5 m |
| Doors and tiles then refuse to line up | Adjust until the tape agrees |

*The check works because a triangle with sides 3, 4, 5 can only be built with a right angle between the short sides.*

## Diagonal of a rectangle: from rooms to screens

A rectangle cut along its diagonal makes two right triangles, so width² + height² = diagonal². A 12 ft by 16 ft room has a diagonal of √(144 + 256) = √400 = **20 ft**, another 3-4-5 in disguise. Screens are sold by their diagonal, which is why the number on the box is bigger than the width.

**Actual width of a 16:9 TV for each advertised diagonal (inches)**

|  | Value |
| --- | --- |
| 32" | 27.9 in |
| 43" | 37.5 in |
| 50" | 43.6 in |
| 55" | 47.9 in |
| 65" | 56.7 in |
| 75" | 65.4 in |
| 85" | 74.1 in |

*For 16:9, width = diagonal × 16 ÷ √337 and height = diagonal × 9 ÷ √337. A 55-inch TV is about 47.9 in wide and 27.0 in tall.*

**A 55-inch 16:9 TV in numbers**

|  | Value | Note |
| --- | --- | --- |
| Diagonal | 55 in | The number on the box |
| Width | 47.9 in | About 122 cm |
| Height | 27.0 in | About 68 cm |
| Screen area | 1,293 sq in | Roughly 1.2 sq m |

*Doubling the diagonal does not double the screen: it makes the area four times larger, because area grows with the square.*

## Ladders: how far from the wall?

A common safety rule is the **4-to-1 rule**: for every 4 units of height you climb, put the foot of the ladder 1 unit away from the wall. That makes an angle of about 76 degrees. Pythagoras gives the numbers: with a 4 m ladder, the base sits about 0.97 m from the wall and the top reaches about 3.88 m.

| Ladder length | Foot from wall | Height reached | Angle |
| --- | --- | --- | --- |
| 3 m | 0.73 m | 2.91 m | about 76° |
| 4 m | 0.97 m | 3.88 m | about 76° |
| 5 m | 1.21 m | 4.85 m | about 76° |
| 6 m | 1.46 m | 5.82 m | about 76° |

*Every row is the same triangle scaled up: base = length ÷ √17 and height = 4 × length ÷ √17. Follow the maker's label on your ladder first; the 4-to-1 rule is a guide, not a replacement for it.*

## Walking blocks versus flying straight

On a street grid you cannot cut the corner. Going 3 blocks east and 4 blocks north means walking 7 blocks, but a bird flies 5. The gap between the two is the saving from the diagonal.

**Distance for a 3-east, 4-north trip**

|  | Blocks |
| --- | --- |
| Along the streets | 7 blocks |
| Straight line | 5 blocks |

*A drone or a cable can take the straight route. A taxi cannot. The straight line is always the shortest possible path.*

## Is this triangle a right triangle?

The theorem also works backwards. Take the longest side as c and compare c² with a² + b². The result tells you the kind of triangle.

**Compare c² with a² + b²**

| Box | What it means |
| --- | --- |
| c² = a² + b² | Exactly equal: a right triangle (90°). Example: 6, 8, 10. |
| c² < a² + b² | Smaller: all angles are sharp, an acute triangle. Example: 6, 7, 8. |
| c² > a² + b² | Larger: one angle is wider than 90°, an obtuse triangle. Example: 4, 5, 8. |
| Not a triangle | If the two short sides add up to less than the long one, no triangle exists. Example: 2, 3, 9. |

*For 6, 7, 8: 36 + 49 = 85, which is more than 64, so the triangle is acute.*

> **Common mistakes:** Using a leg as c, forgetting to take the square root at the end, adding the sides instead of their squares (3 + 4 is 7, not 5), and applying the rule to a triangle that has no right angle. Always check that c is the longest side.

## Quick values worth remembering

| Legs | Hypotenuse | Note |
| --- | --- | --- |
| 1, 1 | √2 ≈ 1.4142 | Diagonal of a unit square |
| 1, 2 | √5 ≈ 2.2361 | Not a whole number |
| 2, 3 | √13 ≈ 3.6056 | Not a whole number |
| 5, 5 | ≈ 7.0711 | 5 × √2 |
| 10, 10 | ≈ 14.1421 | 10 × √2 |

*Any square's diagonal is its side times about 1.4142.*

## Frequently asked questions

### What is the Pythagorean theorem?

In a right triangle, the square of the hypotenuse equals the sum of the squares of the other two sides: a² + b² = c². It lets you find any one side when you know the other two.

### What is the hypotenuse?

The hypotenuse is the side opposite the right angle. It is always the longest side of a right triangle.

### What is a 3-4-5 triangle used for?

It is a quick way to build a perfect right angle. Measure 3 along one edge and 4 along the other; if the distance between the marks is 5, the corner is square.

### Does the Pythagorean theorem work for any triangle?

No. It holds only for right triangles. For other triangles you need the law of cosines, which reduces to Pythagoras when the angle is 90 degrees.

### How do I find a missing leg?

Subtract instead of adding: b = √(c² − a²). For c = 13 and a = 5, b = √(169 − 25) = 12.

### Why is a 55-inch TV not 55 inches wide?

Screen size is measured along the diagonal. For a 16:9 screen, the width is about 87 percent of the diagonal, so a 55-inch TV is roughly 47.9 inches wide.



---

# Quadratic equations explained: the formula, the discriminant and what the roots mean

> Every quadratic ax² + bx + c = 0 has the same solution formula. One number inside it, the discriminant, tells you in advance whether you will get two answers, one, or none you can plot. This guide solves four examples and ends with a ball thrown into the air.

Published 2026-10-09 by Toolcloak. 5 minute read. Original: https://toolcloak.pages.dev/articles/quadratic-equations-explained-the-formula-the-discriminant-and-what-the-roots-mean
Related free tool: [Quadratic Equation Solver](https://toolcloak.pages.dev/tools/quadratic-equation-solver)

A **quadratic equation** is any equation that can be written as **ax² + bx + c = 0**, with a ≠ 0. The x² term is what makes it quadratic. These equations describe thrown balls, profit curves, bridge cables and the area of fields, which is why every maths course teaches them.

> **In short**
> - The formula: **x = (−b ± √(b² − 4ac)) ÷ 2a**. The ± gives up to two answers, called roots.
> - The **discriminant D = b² − 4ac** decides everything: **D > 0** two real roots, **D = 0** one repeated root, **D < 0** no real roots (two complex ones).
> - The roots are where the parabola y = ax² + bx + c crosses the x-axis. Its lowest or highest point (the vertex) is at x = −b ÷ 2a.
> - Quick check: the roots **add up to −b/a** and **multiply to c/a**.

## The formula step by step

Start with x² − 5x + 6 = 0. Here a = 1, b = −5 and c = 6. Substitute into the formula and work from the inside out.

**Solving x² − 5x + 6 = 0**

1. **Identify a, b, c.** a = 1, b = −5, c = 6.
2. **Discriminant.** D = (−5)² − 4 × 1 × 6 = 25 − 24 = **1**.
3. **Square root.** √1 = 1.
4. **Two roots.** x = (5 + 1) ÷ 2 = **3** and x = (5 − 1) ÷ 2 = **2**.
5. **Check.** 3² − 15 + 6 = 0 and 2² − 10 + 6 = 0. Both work.

*Note that −b is −(−5) = +5. Dropping that sign is the most common slip.*

Try it: [Quadratic Equation Solver](https://toolcloak.pages.dev/tools/quadratic-equation-solver). Enter a, b and c to get the discriminant, both roots and the steps.

## The discriminant tells you what to expect

The part under the square root, D = b² − 4ac, is checked first. You cannot take the square root of a negative number and stay with real numbers, so the sign of D splits quadratics into three cases.

**Three outcomes, one number**

| Box | What it means |
| --- | --- |
| D > 0: two real roots | The parabola crosses the x-axis twice. Example: x² − 5x + 6, D = 1, roots 2 and 3. |
| D = 0: one repeated root | The parabola just touches the axis at its vertex. Example: x² − 6x + 9, D = 0, root 3. |
| D < 0: no real roots | The parabola floats above or below the axis. Example: x² + 2x + 5, D = −16. |
| D is a perfect square | The roots are rational and the equation factors neatly, as with 25 in the next example. |

*When D < 0 the two roots are complex numbers: x² + 2x + 5 = 0 gives x = −1 ± 2i.*

| Equation | a, b, c | D | Roots |
| --- | --- | --- | --- |
| x² − 5x + 6 = 0 | 1, −5, 6 | 1 | 2 and 3 |
| 2x² + 3x − 2 = 0 | 2, 3, −2 | 25 | 0.5 and −2 |
| x² − 6x + 9 = 0 | 1, −6, 9 | 0 | 3 (repeated) |
| x² + 2x + 5 = 0 | 1, 2, 5 | −16 | −1 + 2i and −1 − 2i |

*For 2x² + 3x − 2: x = (−3 ± 5) ÷ 4, so 2 ÷ 4 = 0.5 and −8 ÷ 4 = −2.*

## Seeing the roots on a graph

Plot y = x² − 5x + 6 and the curve is a U-shaped parabola. It crosses zero exactly at the two roots. The lowest point is halfway between them, at x = 5 ÷ 2 = 2.5, where y = −0.25.

| x | 0 | 1 | 2 | 2.5 | 3 | 4 | 5 |
| --- | --- | --- | --- | --- | --- | --- | --- |
| y = x² − 5x + 6 | 6 | 2 | 0 | −0.25 | 0 | 2 | 6 |

*The curve is symmetric about the vertex at x = 2.5. Zero appears at x = 2 and x = 3, the two roots, at equal distances either side of the lowest point.*

When a is positive the parabola opens upward and the vertex is a minimum. When a is negative it opens downward and the vertex is a maximum, as in a thrown ball.

## Three ways to solve one

| Method | Best when | Weakness |
| --- | --- | --- |
| Factoring | The numbers are small and the roots are whole or simple fractions | Hard to spot when roots are messy |
| Quadratic formula | Always works, for any a, b and c | More arithmetic, easy to slip on signs |
| Completing the square | You need the vertex form or want to see where the formula comes from | Slower for one-off answers |

*Factoring example: x² − 5x + 6 = (x − 2)(x − 3). Each bracket is zero at a root.*

**Completing the square** rewrites x² − 5x + 6 as (x − 2.5)² − 0.25. Setting that to zero gives (x − 2.5)² = 0.25, so x − 2.5 = ±0.5 and x = 2 or 3. The quadratic formula is exactly this process done once, in general.

## A quick check using sum and product

If the roots are r₁ and r₂, then r₁ + r₂ = −b ÷ a and r₁ × r₂ = c ÷ a. It is a fast way to confirm your answers.

**Checking x² − 5x + 6 (roots 2 and 3)**

|  | Value | Note |
| --- | --- | --- |
| Sum of roots | 5 | 2 + 3 = −(−5) ÷ 1 |
| Product of roots | 6 | 2 × 3 = 6 ÷ 1 |
| Midpoint = vertex x | 2.5 | (2 + 3) ÷ 2 |
| Vertex y | −0.25 | Lowest point of the curve |

*For 2x² + 3x − 2 the roots 0.5 and −2 sum to −1.5 = −3 ÷ 2 and multiply to −1 = −2 ÷ 2. Both checks pass.*

## A real case: a ball thrown upward

A ball is thrown straight up at 20 m/s from a height of 1.5 m. Ignoring air resistance, its height after t seconds is **h = −4.9t² + 20t + 1.5** metres. When does it hit the ground? Set h = 0 and solve.

**Solving −4.9t² + 20t + 1.5 = 0**

1. **Identify.** a = −4.9, b = 20, c = 1.5.
2. **Discriminant.** D = 20² − 4(−4.9)(1.5) = 400 + 29.4 = **429.4**, so √D ≈ 20.72.
3. **Two roots.** t = (−20 ± 20.72) ÷ (−9.8), giving **t ≈ 4.16 s** and t ≈ −0.07 s.
4. **Choose the sensible one.** Time cannot be negative here, so the ball lands after about **4.16 seconds**.

*The negative root is a real solution of the equation but not of the situation. Always ask which answers make sense.*

**The ball's flight**

|  | Value | Note |
| --- | --- | --- |
| Time in the air | 4.16 s | Positive root |
| Time to the top | 2.04 s | t = −b ÷ 2a = 20 ÷ 9.8 |
| Maximum height | 21.9 m | At the vertex |
| Discriminant | 429.4 | Positive, so it does reach the ground |

*The vertex gives the peak: at t = 2.04 s, h = 1.5 + 20² ÷ (4 × 4.9) ≈ 21.9 m.*

**Height of the ball over time (metres)**

| Seconds after throwing | Height |
| --- | --- |
| 0 | 1.5 |
| 1 | 16.6 |
| 2 | 21.9 |
| 3 | 17.4 |
| 4 | 3.1 |
| 4.16 | 0 |

*It rises for about 2 seconds, peaks near 21.9 m and then falls back to the ground. The path is a parabola.*

## Common mistakes

1. **Not rearranging to zero first.** The equation must be in the form ax² + bx + c = 0 before you read off a, b and c.
2. **Sign errors on b.** If b = −5, then −b = +5 and b² = 25.
3. **Dividing only part of the numerator by 2a.** The whole of (−b ± √D) is divided by 2a.
4. **Forgetting a = 0 case.** If a = 0 the equation is linear, not quadratic.
5. **Keeping answers that do not fit the problem,** such as negative times or lengths.

## Frequently asked questions

### What is the quadratic formula?

For ax² + bx + c = 0 the solutions are x = (−b ± √(b² − 4ac)) ÷ 2a. It works for every quadratic equation.

### What is the discriminant?

It is b² − 4ac, the expression under the square root. If it is positive there are two real roots, if zero there is one repeated root, and if negative there are no real roots.

### Can a quadratic equation have no solution?

It has no real solution when the discriminant is negative. It always has two solutions among the complex numbers, though.

### How do I find the vertex of a parabola?

The x-coordinate of the vertex is −b ÷ 2a. Substitute it back into the equation to find the y-coordinate.

### How do I check my roots?

Substitute each root into the original equation, or use the sum and product: the roots add to −b/a and multiply to c/a.

### When should I use factoring instead of the formula?

Factoring is quicker when the coefficients are small and the roots are whole numbers. The formula is the reliable choice when they are not.



---

# Regex from zero: ten patterns and the trap that freezes pages

> A regular expression looks like a cat walked across the keyboard. It is really a tiny language for describing text. Learn six symbols and you can check a PIN code, pull numbers out of a paragraph or tidy a messy list. Learn one warning and you will not freeze your own page.

Published 2026-10-09 by Toolcloak. 4 minute read. Original: https://toolcloak.pages.dev/articles/regex-from-zero-ten-patterns-and-the-trap-that-freezes-pages
Related free tool: [Regex Tester](https://toolcloak.pages.dev/tools/regex-tester)

A **regular expression**, or regex, is a pattern that describes a piece of text. Instead of asking 'is this exactly the word cat?', you ask 'does this look like six digits?' or 'is there a date somewhere in here?'. Almost every programming language and text editor understands the same core ideas.

> **In short**
> - Six building blocks cover most uses: `\d` (a digit), `\w` (a letter, digit or underscore), `.` (any character), `[ ]` (one of these), `+` / `*` / `?` / `{n}` (how many), and `^` / `$` (start and end).
> - Round brackets **capture** parts of a match so you can reuse them.
> - By default quantifiers are **greedy**: they take as much as they can. Add `?` after them to make them lazy.
> - Nested repeats like `(a+)+` can make a regex take forever on some inputs. Keep patterns simple.

## Anatomy of a pattern

Take the pattern `^[6-9]\d{9}$`, a simple check for a 10-digit Indian mobile number. It reads left to right, one piece at a time:

**Reading ^[6-9]\d{9}$ in plain words**

1. **^.** The match must start at the very beginning of the text.
2. **[6-9].** The first character is one of 6, 7, 8 or 9.
3. **\d{9}.** Followed by exactly nine more digits.
4. **$.** And then the text must end. Nothing extra is allowed.

*Without ^ and $, the pattern would also match inside a longer string such as 1234567890123. Anchors turn 'contains' into 'is exactly'.*

Try it: [Regex Tester](https://toolcloak.pages.dev/tools/regex-tester). Type a pattern and some text. Matches are highlighted live, with each capture group listed.

## The symbols you will use daily

| Symbol | Meaning | Example |
| --- | --- | --- |
| `\d` | One digit, 0 to 9 | `\d\d` matches 42 |
| `\w` | A letter, digit or underscore | `\w+` matches user_1 |
| `\s` | A space, tab or line break | `\s+` matches a gap of spaces |
| `.` | Any one character except a new line | `c.t` matches cat and cot |
| `[abc]` | One of the listed characters | `gr[ae]y` matches gray and grey |
| `[^abc]` | Any character not listed | `[^0-9]` matches a non-digit |
| `+` | One or more | `a+` matches a, aa, aaa |
| `*` | Zero or more | `ab*` matches a, ab, abb |
| `?` | Zero or one (optional) | `colou?r` matches color and colour |
| `{n,m}` | Between n and m times | `\d{2,4}` matches 12, 123 and 1234 |
| `( )` | A group, captured for reuse | `(\d+)-(\d+)` captures both numbers |
| `a|b` | Either a or b | `cat|dog` matches either word |

*In most languages a backslash inside a normal string has to be doubled, so \d is written \\d in code. In the tester you type it once.*

## Ten patterns worth keeping

| Goal | Pattern |
| --- | --- |
| A whole number | `^\d+$` |
| Indian PIN code (6 digits, not starting with 0) | `^[1-9][0-9]{5}$` |
| Indian mobile number | `^[6-9]\d{9}$` |
| A date in the shape 2026-10-09 | `^\d{4}-\d{2}-\d{2}$` |
| A hex colour like #1a2b3c or #abc | `^#?([0-9a-fA-F]{3}|[0-9a-fA-F]{6})$` |
| A rough email check | `^[^\s@]+@[^\s@]+\.[^\s@]+$` |
| Extra spaces at either end | `^\s+|\s+$` |
| Two or more spaces in a row | `\s{2,}` |
| Every number inside a paragraph | `\d+(\.\d+)?` |
| A repeated word such as 'the the' | `\b(\w+)\s+\1\b` |

*The date pattern checks the shape only: it accepts 2026-99-99. The email pattern is a quick sanity check, not full validation. For real validation, confirm by sending a message.*

## Capture groups: keep the parts you care about

Round brackets mark parts of the match you want to keep. Try `(\d{4})-(\d{2})-(\d{2})` on the text 2026-10-09. You get the whole match, plus three groups: **2026**, **10** and **09**. In JavaScript you can name them with `(?<year>\d{4})` and read them back by name, which makes longer patterns far easier to follow.

**One match, four pieces**

|  | Value | Note |
| --- | --- | --- |
| Whole match | 2026-10-09 | The full text found |
| Group 1 | 2026 | The year |
| Group 2 | 10 | The month |
| Group 3 | 09 | The day |

*Groups are numbered by their opening bracket, from left to right.*

## Greedy or lazy?

Quantifiers like `+` and `*` are **greedy**: they grab as much text as they can and only give some back if the rest of the pattern needs it. That is often not what you want. Take the text `<b>one</b> and <b>two</b>`:

**Same text, two behaviours**

| Greedy: <b>.*</b> | Lazy: <b>.*?</b> |
| --- | --- |
| 1 match | 2 matches |
| Runs to the last </b> | Stops at the first </b> |
| Swallows everything in between | Finds each bold word separately |
| Result: <b>one</b> and <b>two</b> | Results: <b>one</b> and <b>two</b> |

*Adding ? after a quantifier makes it lazy. Remember this one: it is behind a huge share of 'my regex matched too much' questions.*

## Flags change how the whole pattern behaves

**The four flags in the tester**

| Box | What it means |
| --- | --- |
| g (global) | Find every match, not just the first one. |
| i (ignore case) | Treat A and a as the same letter. |
| m (multiline) | Make ^ and $ match at the start and end of each line. |
| s (dot all) | Let the dot match line breaks as well. |

*Most 'it works on one line but not on my whole text' problems are solved by the m flag.*

## The trap: catastrophic backtracking

When a regex cannot match, the engine does not give up at once. It tries other ways of splitting the text among the quantifiers. With a pattern such as `(a+)+b` run against a long line of a's with no b at the end, the number of ways grows by about **double with every extra character**. A web page that runs such a pattern on user input can freeze its tab, or a server can spend its whole time on one request.

**Rough number of steps for (a+)+b on a string of n a's with no b**

|  | Steps |
| --- | --- |
| n = 10 | about 1 thousand |
| n = 20 | about 1 million |
| n = 25 | about 34 million |
| n = 30 | about 1 billion |

*The count is roughly 2 to the power n. Real engines differ in details and some have protections, but you should never depend on that.*

> **How to stay safe:** Avoid a repeat inside a repeat, such as `(a+)+`, `(.*)*` or `(\w+\s?)*`. Make alternatives clearly different from each other. Limit lengths with `{1,50}` where you can. And if the pattern runs on text a stranger sent you, cap the input length first.

## A good way to learn

1. **Start tiny.** Match one letter, then add one idea at a time.
2. **Test on good and bad text.** Always include at least one example that should fail.
3. **Anchor when you validate.** Use ^ and $ to check a whole value.
4. **Prefer simple patterns.** If it takes you a minute to read, it will take your teammate longer. Add a comment where you use it.
5. **Know when not to use regex.** For full HTML, JSON or email parsing, use a proper parser.

## Frequently asked questions

### What is a regular expression?

A regular expression, or regex, is a pattern that describes a set of strings. You use it to search, check or replace text, for example finding every number in a paragraph or checking that a value looks like a PIN code.

### What does \d mean in regex?

\d matches one digit from 0 to 9. \d+ matches one or more digits and \d{6} matches exactly six digits.

### What is the difference between greedy and lazy matching?

A greedy quantifier such as .* matches as much as it can. A lazy one such as .*? matches as little as it can. On <b>one</b> and <b>two</b>, <b>.*</b> finds one long match and <b>.*?</b> finds two short ones.

### What do ^ and $ do?

^ matches the start of the text and $ matches the end. Using both around a pattern means the whole text must match, not just a part of it.

### What is catastrophic backtracking?

It happens when a pattern with nested repeats, such as (a+)+b, fails on a long input and the engine tries an exponential number of ways to split the text. It can make a page or server hang.



---

# Rental yield calculator: gross vs net yield, payback years and whether a rental flat is worth it

> Rent looks like free money until you subtract tax, repairs and empty months. This guide shows how to work out gross and net rental yield, what payback years really mean, how yield compares with a loan rate, and why rent is only half of a property's return.

Published 2026-10-09 by Toolcloak. 3 minute read. Original: https://toolcloak.pages.dev/articles/rental-yield-calculator-gross-vs-net-yield-payback-years-and-whether-a-rental-flat-is-worth-it
Related free tool: [Rental Yield Calculator](https://toolcloak.pages.dev/tools/rental-yield-calculator)

**Rental yield** is the yearly rent as a percentage of what the property costs. It is the quickest way to compare a flat that earns rent with a fixed deposit, a fund or another flat. There are two versions: **gross** yield uses the full rent, and **net** yield first subtracts the costs of owning the property.

> **In short**
> - **Gross yield = yearly rent ÷ price × 100.** ₹20,000 a month on a ₹60 lakh flat is ₹2,40,000 a year, or **4.00%**.
> - **Net yield = (yearly rent − yearly costs) ÷ price × 100.** With ₹30,000 of costs it is **3.50%**.
> - **Payback years = price ÷ net yearly income** = **28.6 years** of rent to earn back the price.
> - Yield is only the rent part of the return. Price changes, loan interest and your own time matter too.

## The formulas, step by step

**Example: a ₹60 lakh flat rented at ₹20,000 a month**

1. **Work out the yearly rent.** ₹20,000 × 12 = **₹2,40,000**.
2. **Gross yield.** ₹2,40,000 ÷ ₹60,00,000 × 100 = **4.00%**.
3. **Subtract the yearly costs.** ₹2,40,000 − ₹30,000 = **₹2,10,000** net income.
4. **Net yield and payback.** ₹2,10,000 ÷ ₹60,00,000 = **3.50%**, and ₹60,00,000 ÷ ₹2,10,000 = **28.6 years**.

*The tool returns four numbers: net yield, gross yield, net yearly income and payback years.*

Try it: [Rental Yield Calculator](https://toolcloak.pages.dev/tools/rental-yield-calculator). Enter the price, the monthly rent and the yearly costs to see the gross and net yield.

## What counts as a cost

- **Property tax and society charges,** paid by the owner and not by the tenant.
- **Repairs and painting** between tenants. Setting aside a small share of the rent each month smooths the bumps.
- **Vacancy,** the months between tenants when you earn nothing.
- **Agent or brokerage fees** for finding a new tenant.
- **Income tax on the rent,** if you want a true after-tax yield.

> **One empty month is bigger than it looks:** A month without a tenant costs ₹20,000, which is 8.3% of the year's rent. Two empty months wipe out about 17% of it. Many first-time owners forget to count vacancy at all.

## How rent changes the yield

| Monthly rent | Yearly rent | Net income | Gross yield | Net yield |
| --- | --- | --- | --- | --- |
| ₹15,000 | ₹1,80,000 | ₹1,50,000 | 3.00% | 2.50% |
| ₹20,000 | ₹2,40,000 | ₹2,10,000 | 4.00% | 3.50% |
| ₹25,000 | ₹3,00,000 | ₹2,70,000 | 5.00% | 4.50% |
| ₹30,000 | ₹3,60,000 | ₹3,30,000 | 6.00% | 5.50% |
| ₹35,000 | ₹4,20,000 | ₹3,90,000 | 7.00% | 6.50% |

*All rows use a ₹60 lakh price and ₹30,000 of yearly costs. Each ₹5,000 of extra monthly rent adds 1 percentage point to the net yield.*

**Net yield for different monthly rents (%)**

|  | Value |
| --- | --- |
| ₹15k | 2.5% |
| ₹20k | 3.5% |
| ₹25k | 4.5% |
| ₹30k | 5.5% |
| ₹35k | 6.5% |

*A rent increase flows straight into net yield because the costs stay the same. A flat at ₹35,000 a month would give a net yield of 6.5% here.*

## Payback years in plain words

Payback years tells you how long the rent alone would take to repay the price: ₹60,00,000 ÷ ₹2,10,000 is about 28.6 years. It ignores price changes, so it is not a forecast. It does help you see when a high price compared with the rent makes a flat a long-term bet rather than an income source.

**The example flat in four numbers**

|  | Value | Note |
| --- | --- | --- |
| Gross yield | 4.00% | Before costs |
| Net yield | 3.50% | After yearly costs |
| Net income a year | ₹2,10,000 | ₹2,40,000 − ₹30,000 |
| Payback | 28.6 years | Price ÷ net income |

*A yield around 3% to 4% is common for residential flats in many Indian cities, but it varies a lot by city, locality and building type.*

## Yield against your loan rate

If you borrow to buy, compare the net yield with the loan rate. Suppose you finance 80% of the flat, which is ₹48,00,000, at 8.75% a year. The interest alone is about ₹4,20,000 a year, while the net rent is ₹2,10,000.

**Rent against loan interest in year one**

| Net rent | Interest on the loan |
| --- | --- |
| ₹2,10,000 a year | ₹4,20,000 a year |
| Gross rent ₹2,40,000 | 80% of ₹60 lakh = ₹48,00,000 |
| Less costs ₹30,000 | 8.75% a year, interest part only |
| Before income tax | Principal repaid is extra |

*The rent covers about half of the interest. You top up the rest from your salary and hope the price rises. This is called negative carry.*

> **Compare with a fixed deposit:** If a fixed deposit pays 7% a year, a flat yielding 3.5% needs about 3.5% a year of price growth just to match it. Use the [FD Calculator](https://toolcloak.pages.dev/tools/fd-calculator) to see what a deposit earns.

## Rent is only half the return

The **total return** on a property is the yield plus the change in price. If a flat yields 3.5% net and its price rises 5% a year, the combined return is roughly 8.5% before tax and loan costs. If the price falls, it can drag the total below zero even when the rent keeps coming. Always look at both parts.

## Checklist before you buy to rent

1. **Ask local brokers for real rents** on similar flats, not just the asking price.
2. **List every yearly cost,** including tax, repairs and at least one empty month.
3. **Calculate gross and net yield** with the calculator.
4. **Compare the net yield** with your loan rate and with a deposit.
5. **Test a worse case:** run the numbers with 10% lower rent and two empty months.

## Frequently asked questions

### How do I calculate rental yield?

Divide the yearly rent by the property price and multiply by 100. For rent of 20,000 a month on a 60 lakh flat, the yearly rent is 2,40,000, so the gross yield is 4%.

### What is the difference between gross and net rental yield?

Gross yield uses the full rent. Net yield subtracts yearly costs such as tax, repairs and vacancy first. Net yield is the more realistic figure.

### What is a good rental yield?

It depends on the city and the type of property. Many residential flats in Indian cities yield about 2% to 4% gross, while commercial property can yield more. Compare it with the loan rate and with deposit rates.

### What does payback years mean?

It is the price divided by the net yearly rental income, and shows how many years of rent would be needed to earn back the purchase price. It ignores price changes and tax.

### Does rental yield include price appreciation?

No. Yield is only the income from rent. The total return is the yield plus the change in the property's value, minus costs such as loan interest.

### Can I use the calculator for a commercial property?

Yes. Enter the price, the monthly rent and your yearly costs. The [Rental Yield Calculator](https://toolcloak.pages.dev/tools/rental-yield-calculator) works for any property that earns rent.



---

# robots.txt explained: Allow, Disallow, wildcards and the mistakes that block your site

> One small text file can tell search engines to crawl your whole site or none of it. This guide covers the four directives that matter, how crawlers pick between conflicting rules, why robots.txt is not a security tool, and the single-character slip that has taken sites out of search.

Published 2026-10-09 by Toolcloak. 4 minute read. Original: https://toolcloak.pages.dev/articles/robots-txt-explained-allow-disallow-wildcards-and-the-mistakes-that-block-your-site
Related free tool: [Robots.txt Generator](https://toolcloak.pages.dev/tools/robots-txt-generator)

Before a well-behaved crawler visits any page on your site, it first asks for one file: **/robots.txt**. The file lists which parts of the site it may crawl and which it should skip. It is simple, and that is the danger: a single wrong line can hide an entire website from search.

> **In short**
> - The file must live at the **root** of the host: `https://example.com/robots.txt`. A copy in a subfolder is ignored.
> - Four directives do almost everything: **User-agent**, **Disallow**, **Allow** and **Sitemap**.
> - `Disallow: /` blocks the **whole site**. `Disallow:` with nothing after it blocks nothing.
> - robots.txt controls **crawling**, not indexing, and it is **not security**. Anyone can read it, and a blocked URL can still appear in results if other sites link to it.

## The shape of the file

The file is plain text, built from groups. Each group starts with the crawler it addresses, followed by the rules for it.

| Directive | Meaning | Example |
| --- | --- | --- |
| `User-agent` | Which crawler the group applies to. `*` means all. | `User-agent: Googlebot` |
| `Disallow` | A path the crawler should not fetch | `Disallow: /admin/` |
| `Allow` | A path that is allowed, used to carve out exceptions | `Allow: /admin/help/` |
| `Sitemap` | Full address of your XML sitemap, outside any group | `Sitemap: https://example.com/sitemap.xml` |

*Paths are case-sensitive and start with a slash. `/Admin/` and `/admin/` are different.*

Try it: [Robots.txt Generator](https://toolcloak.pages.dev/tools/robots-txt-generator). Choose crawlers, add allow and disallow paths and your sitemap, then copy the finished file.

## A complete example

**A typical small-site file, read line by line**

1. **User-agent: *.** The rules below apply to every crawler that has no group of its own.
2. **Disallow: /admin/.** Do not crawl anything under /admin/.
3. **Disallow: /cart/.** Skip the shopping cart. It has no value in search.
4. **Allow: /admin/help/.** But the help pages inside /admin/ may be crawled.
5. **Sitemap: https://example.com/sitemap.xml.** Tell crawlers where the full list of pages is.

*A crawler follows the most specific matching group. A named group such as Googlebot replaces the * group for that crawler. It does not add to it.*

## How crawlers choose between rules

Rules can overlap. Google and most major crawlers pick the **longest matching path**, because it is the most specific. If an Allow and a Disallow match with equal length, the less restrictive one, Allow, wins.

| URL | Matching rules | Result |
| --- | --- | --- |
| `/admin/login` | `Disallow: /admin/` | Blocked |
| `/admin/help/faq` | `Disallow: /admin/` and `Allow: /admin/help/` (longer) | Allowed |
| `/cart/` | `Disallow: /cart/` | Blocked |
| `/blog/post-1` | No rule matches | Allowed (the default) |

*Anything not mentioned is allowed. You only list what you want kept out.*

## Wildcards and the end marker

Two special characters work in Google, Bing and most other major crawlers. Both can save you from writing hundreds of lines.

**Pattern characters**

| Box | What it means |
| --- | --- |
| * (star) | Matches any run of characters. `Disallow: /*?sort=` blocks any URL containing `?sort=`. |
| $ (end) | Matches the end of the URL. `Disallow: /*.pdf$` blocks addresses that end in .pdf. |
| Trailing slash | `/shop` also matches `/shop-sale`. `/shop/` matches only inside the folder. |
| No regex | Only these two characters are special. Full regular expressions do not work. |

*Test a new pattern before publishing. A star in the wrong place can match far more than you meant.*

## The mistakes that hurt most

**One character apart**

| Blocks nothing | Blocks everything |
| --- | --- |
| Disallow: | Disallow: / |
| An empty value means no restrictions | The slash means the whole site |
| Every page may be crawled | Crawlers stop fetching all pages |
| Often used as an explicit 'allow all' | Pages can drop out of search over time |

*This often happens when a staging file is copied to the live site. Check robots.txt after every launch.*

1. **Staging rules on the live site.** A leftover `Disallow: /` from a test server is the classic cause of a sudden loss of traffic.
2. **Blocking CSS and JavaScript.** If a crawler cannot load your styles and scripts, it cannot see the page as visitors do.
3. **Using robots.txt to hide pages.** The file is public. Listing `/secret-offers/` advertises exactly where it is.
4. **Wrong location.** It must be at the root of each host. `blog.example.com` needs its own file.
5. **Blocking a page you want removed.** If a page is blocked, the crawler never sees a noindex tag on it.

## robots.txt against noindex

These two tools solve different problems and are often confused.

| Goal | Use | Why |
| --- | --- | --- |
| Save crawl effort on low-value URLs | `Disallow` in robots.txt | The crawler does not fetch the page at all |
| Keep a page out of search results | `noindex` meta tag or header | The crawler must be able to fetch the page to see it |
| Protect private data | Passwords or access rules on the server | robots.txt is a request, not a lock |
| Remove an old page quickly | Removal tool in Search Console, then noindex or 404 | Blocking alone can leave the address listed |

*Never combine Disallow and noindex on the same page and expect noindex to work.*

> **It is not security:** Well-behaved crawlers obey robots.txt. Malicious bots ignore it and may use it as a map. Anything private belongs behind a login.

## Limits and details worth knowing

- **Size.** Google reads about the first 500 KiB of the file and ignores the rest.
- **Encoding.** Save as plain UTF-8 text with the name in lowercase: `robots.txt`.
- **Status code.** A 404 for the file means 'no restrictions'. A server error (5xx) can make Google pause crawling the site.
- **Crawl-delay.** Some crawlers honour it. Google ignores it.
- **Caching.** Crawlers cache the file, often for up to a day, so changes may take a while to be noticed.
- **AI crawlers.** Many publishers now add groups for specific AI user-agents. Each one needs its own `User-agent` line.

## Before you publish

1. **Open the file in a browser** at yoursite.com/robots.txt to confirm it loads.
2. **Search for `Disallow: /`** with nothing after the slash, and be sure it is meant.
3. **Test your important URLs** in the robots.txt report or tester in Search Console.
4. **Add the Sitemap line** with the full https address.
5. **Recheck after every redesign or move** to a new server.

## Frequently asked questions

### What is robots.txt?

It is a text file at the root of a website that tells crawlers which paths they may or may not fetch. It uses the directives User-agent, Disallow, Allow and Sitemap.

### Does Disallow remove a page from Google?

Not reliably. It stops crawling, but the address can still be listed if other pages link to it. To keep a page out of results, use a noindex tag on a page that is not blocked.

### What does Disallow: / do?

It blocks the entire site for the crawlers in that group. A Disallow line with no value blocks nothing.

### Where should robots.txt be placed?

At the root of the host, for example https://example.com/robots.txt. Files in subfolders are not read, and each subdomain needs its own.

### Can I use wildcards in robots.txt?

Yes. The star matches any sequence of characters and the dollar sign marks the end of a URL. For example, Disallow: /*.pdf$ blocks addresses ending in .pdf.

### Is robots.txt a security measure?

No. It is a public request that only cooperative crawlers follow. Protect private pages with a login or server access rules.



---

# Salary hike calculator: how to work out your new salary and what a raise is really worth

> Turn a hike percentage into a new yearly and monthly salary, see how hikes compound over a career, check what your raise is worth after inflation and find the percentage you need to ask for to reach a target.

Published 2026-10-09 by Toolcloak. 4 minute read. Original: https://toolcloak.pages.dev/articles/salary-hike-calculator-how-to-work-out-your-new-salary-and-what-a-raise-is-really-worth
Related free tool: [Salary Hike Calculator](https://toolcloak.pages.dev/tools/salary-hike-calculator)

A **salary hike** is a percentage increase on your current pay. The arithmetic is simple, but three questions usually follow: what is my new monthly salary, what will a string of hikes add up to, and is the raise actually beating the rise in prices? This guide answers all three with real numbers.

> **In short**
> - **New salary = current salary × (1 + hike% ÷ 100).** A 15% hike on ₹6,00,000 a year gives **₹6,90,000**.
> - That is **₹90,000** more a year and **₹7,500** more a month, with a new monthly figure of **₹57,500** (yearly ÷ 12).
> - Hikes **compound**. Ten years of 10% hikes turn ₹6 L into about **₹15.56 L**, not ₹12 L.
> - To judge a raise, subtract inflation. An 8% hike with 6% inflation is only about **1.9%** real growth.

## The formula

Convert the percentage to a multiplier by adding it to 1. A 15% hike is 1.15. Multiply your current yearly salary (or CTC) by that number to get the new figure. Divide by 12 for the monthly amount.

**Example: ₹6,00,000 a year with a 15% hike**

1. **Start with the yearly salary.** ₹6,00,000 a year.
2. **Find the multiplier.** 1 + 15 ÷ 100 = **1.15**.
3. **Multiply.** ₹6,00,000 × 1.15 = **₹6,90,000**.
4. **Work out the monthly figures.** ₹6,90,000 ÷ 12 = **₹57,500**, which is **₹7,500** more per month.

*The tool shows four numbers: new yearly salary, the increase per year, the new monthly figure and the increase per month.*

Try it: [Salary Hike Calculator](https://toolcloak.pages.dev/tools/salary-hike-calculator). Enter your current yearly salary and the hike percentage to see the new salary and the monthly difference.

## What different hikes look like

| Hike | New yearly salary | Extra per year | New monthly | Extra per month |
| --- | --- | --- | --- | --- |
| 5% | ₹6,30,000 | +₹30,000 | ₹52,500 | +₹2,500 |
| 8% | ₹6,48,000 | +₹48,000 | ₹54,000 | +₹4,000 |
| 10% | ₹6,60,000 | +₹60,000 | ₹55,000 | +₹5,000 |
| 12% | ₹6,72,000 | +₹72,000 | ₹56,000 | +₹6,000 |
| 15% | ₹6,90,000 | +₹90,000 | ₹57,500 | +₹7,500 |
| 20% | ₹7,20,000 | +₹1,20,000 | ₹60,000 | +₹10,000 |
| 25% | ₹7,50,000 | +₹1,50,000 | ₹62,500 | +₹12,500 |
| 30% | ₹7,80,000 | +₹1,80,000 | ₹65,000 | +₹15,000 |

*All rows start from ₹6,00,000 a year. Every extra 5 percentage points adds about ₹2,500 a month at this salary.*

**Extra monthly pay from a hike on ₹6,00,000 a year (₹ thousand)**

|  | Value |
| --- | --- |
| 5% | ₹2.5k |
| 10% | ₹5k |
| 15% | ₹7.5k |
| 20% | ₹10k |
| 30% | ₹15k |

*A hike of 15% adds ₹7,500 a month at this salary. The relationship is a straight line: double the percentage and the monthly gain doubles.*

## Hikes compound over the years

Each hike applies to the salary you already have, so the second hike is worth more than the first in rupees. Ten years of 10% is not 100% growth. It is 1.10 multiplied by itself ten times, which is about 2.59 times your starting salary.

**Yearly salary over 10 years starting at ₹6,00,000 (value in lakh rupees)**

| Years from now | 10% hike every year | 5% hike every year |
| --- | --- | --- |
| 0 | 6 | 6 |
| 1 | 6.6 | 6.3 |
| 2 | 7.26 | 6.62 |
| 3 | 7.99 | 6.95 |
| 4 | 8.78 | 7.29 |
| 5 | 9.66 | 7.66 |
| 6 | 10.63 | 8.04 |
| 7 | 11.69 | 8.44 |
| 8 | 12.86 | 8.86 |
| 9 | 14.15 | 9.31 |
| 10 | 15.56 | 9.77 |

*After ten years, 10% a year gives ₹15.56 L and 5% gives ₹9.77 L. The gap keeps widening because hikes build on earlier hikes.*

> **Why the first big hike matters most:** A higher starting salary lifts every later percentage hike. Moving from ₹6 L to ₹7.2 L with a job switch raises the base for all future 8% to 10% increments, which is why a strong first offer is worth more than it looks.

## Is your raise beating inflation?

A hike only helps if prices rise more slowly than your pay. The real growth rate is (1 + hike) ÷ (1 + inflation) − 1. If you got 8% and prices rose 6%, your buying power grew by about 1.9%, not 2%.

| Hike | If inflation is 4% | If inflation is 5% | If inflation is 6% | If inflation is 7% |
| --- | --- | --- | --- | --- |
| 6% | 1.92% | 0.95% | 0.00% | -0.93% |
| 8% | 3.85% | 2.86% | 1.89% | 0.93% |
| 10% | 5.77% | 4.76% | 3.77% | 2.80% |
| 12% | 7.69% | 6.67% | 5.66% | 4.67% |
| 15% | 10.58% | 9.52% | 8.49% | 7.48% |

*Real hike after inflation. A hike lower than inflation is a pay cut in what your money can buy.*

**Same 8% hike, two kinds of year**

| Inflation 4% | Inflation 8% |
| --- | --- |
| +3.8% real | 0% real |
| Pay grows faster than prices | Pay only keeps pace with prices |
| Savings goals get easier | Nothing extra to save |
| A hike you can feel | Feels like standing still |

*Use the [Inflation Calculator](https://toolcloak.pages.dev/tools/inflation-calculator) to see how much a rupee today is worth in the future.*

## Working backwards: what hike do you need?

If you know the salary you want, divide it by your current salary and subtract 1. This is useful in negotiations, because it tells you how big the ask really is in percentage terms.

| Target salary | Hike needed | Extra per year |
| --- | --- | --- |
| ₹7,00,000 | 16.7% | +₹1,00,000 |
| ₹8,00,000 | 33.3% | +₹2,00,000 |
| ₹9,00,000 | 50.0% | +₹3,00,000 |
| ₹10,00,000 | 66.7% | +₹4,00,000 |

*All targets start from ₹6,00,000. The percentage needed grows quickly: a target of 10 L is a 66.7% hike from this base.*

## A hike on CTC is not a hike in take-home pay

The hike percentage usually applies to your CTC (cost to company). Your monthly in-hand pay is lower, because PF contributions, professional tax, income tax and sometimes insurance come out first. A bigger CTC can also push part of your income into a higher tax slab, so the take-home increase can be a little smaller than the CTC increase. Use the [In-Hand Salary Calculator](https://toolcloak.pages.dev/tools/in-hand-salary-calculator) to see the actual monthly effect.

> **Check what the hike applies to:** Ask whether the percentage is on the fixed salary or on the total CTC including bonus and variable pay. A 10% hike on the fixed part alone is smaller than 10% on the whole package.

## Using the numbers well

1. **Calculate the new yearly and monthly figures** so you can compare offers on the same basis.
2. **Check the real growth** against current inflation before you call it a good hike.
3. **Plan the extra income.** Decide in advance how much of the monthly increase goes to savings or an SIP.
4. **Keep a record** of each hike and the date, so you can see your pay history over time.
5. **Compare job offers on total pay,** not just on the headline percentage.

## Frequently asked questions

### How do I calculate my new salary after a hike?

Multiply your current salary by 1 plus the hike percentage as a decimal. A 15% hike on 6,00,000 gives 6,00,000 × 1.15 = 6,90,000 a year.

### How do I find the monthly increase from a hike?

Subtract the old salary from the new one and divide by 12. A rise from 6,00,000 to 6,90,000 is 90,000 a year, or 7,500 a month.

### How do I calculate the hike percentage from two salaries?

Divide the new salary by the old one, subtract 1 and multiply by 100. Going from 6,00,000 to 7,50,000 is 7,50,000 ÷ 6,00,000 − 1 = 25%.

### Is a 10% hike every year the same as a 100% hike in ten years?

No. Hikes compound, so ten years at 10% a year gives about 2.59 times the starting salary, a 159% rise, not 100%.

### What is a good salary hike?

A good hike beats inflation by a clear margin. If prices rise 6% a year, a 6% hike only keeps your buying power the same. Anything above that is real growth. Typical hikes also vary by industry, role and company performance.

### Does the Salary Hike Calculator show in-hand salary?

No. The [Salary Hike Calculator](https://toolcloak.pages.dev/tools/salary-hike-calculator) works on your gross yearly salary or CTC. For take-home pay after deductions, use the in-hand salary calculator.



---

# Sample size: how many people do you need to survey?

> Why can a poll of 1,000 people describe a country of 100 million? This guide explains margin of error, confidence level and the 385 rule in plain words, with charts that show where adding more respondents stops being worth the cost.

Published 2026-10-09 by Toolcloak. 4 minute read. Original: https://toolcloak.pages.dev/articles/sample-size-how-many-people-to-survey-margin-of-error-and-confidence
Related free tool: [Sample Size Calculator](https://toolcloak.pages.dev/tools/sample-size-calculator)

You almost never need to ask everyone. A well-chosen **sample** of a few hundred or a few thousand people can describe a very large group with a known amount of error. The calculator turns three choices into one number: how precise you want to be, how sure you want to be and how varied you expect the answers to be.

> **In short**
> - For a large population, **385 responses** give a ±5% margin of error at 95% confidence, whether you survey a town or a country.
> - Precision is expensive: halving the margin of error needs **four times** as many responses.
> - Population size barely matters unless the group is small. For under a few thousand people, a correction lowers the number you need.
> - A big sample cannot fix a bad sample. If the wrong people answer, more of them just makes you confidently wrong.

## The three dials

**What each setting means**

| Box | What it means |
| --- | --- |
| Margin of error | How far the result may sit from the truth. ±5% means a measured 60% really lies between 55% and 65%. |
| Confidence level | How often the method gets it right. 95% means 19 out of 20 surveys would land within the margin. |
| Expected proportion | Your guess of the answer split. If unsure, use 50%, the case that needs the most people. |
| Population size | The whole group you care about. It matters only when the group is small. |

*The common formula is n = z² × p × (1 − p) ÷ e², where z is 1.96 for 95% confidence, p is the expected proportion and e is the margin of error.*

Try it: [Sample Size Calculator](https://toolcloak.pages.dev/tools/sample-size-calculator). Set the margin of error, confidence level and population to get the number of responses you need.

## Where 385 comes from

**Sample size for ±5% at 95% confidence, answers split 50-50**

1. **Pick the numbers.** z = 1.96 (95% confidence), p = 0.5, e = 0.05.
2. **Top of the formula.** 1.96² × 0.5 × 0.5 = 3.8416 × 0.25 = 0.9604.
3. **Bottom of the formula.** 0.05² = 0.0025.
4. **Divide and round up.** 0.9604 ÷ 0.0025 = 384.16, so you need **385** responses.

*Always round up. 384 would fall just short of the target.*

## Smaller margin, much bigger sample

Precision follows a square-root law. To cut the margin of error in half you need four times the responses. The curve below is why most surveys settle for ±3% to ±5%.

| Margin of error | Responses needed (95%) | Compared with ±5% |
| --- | --- | --- |
| ±10% | 97 | about a quarter |
| ±5% | **385** | baseline |
| ±3% | 1,068 | 2.8 times |
| ±2% | 2,401 | 6.2 times |
| ±1% | 9,604 | 25 times |

*Going from ±5% to ±1% means asking about 25 times as many people.*

**Responses needed for each margin of error (95% confidence)**

|  | Value |
| --- | --- |
| ±10% | 97 |
| ±5% | 385 |
| ±3% | 1068 |
| ±2% | 2401 |
| ±1% | 9604 |

*The last bar is so tall that the first two look tiny. That is the point: the last bit of precision is the costliest.*

**Margin of error as the sample grows (95%, answers split 50-50)**

| Responses | Margin of error (±%) |
| --- | --- |
| 50 | 13.9 |
| 100 | 9.8 |
| 250 | 6.2 |
| 400 | 4.9 |
| 1000 | 3.1 |
| 2500 | 2 |
| 10000 | 1 |

*The line drops fast at first and then flattens. The jump from 100 to 400 responses roughly halves the error; the jump from 2,500 to 10,000 only takes off one point.*

**Is 385 responses enough? Margin of error for your sample**

| Zone | Range |
| --- | --- |
| Sharp | under 3% |
| Usable | 3–6% |
| Rough | over 6% |

*A rough guide only. Close races and small differences need a sharper margin than a general 'how many people like this' question.*

## How sure do you want to be?

A higher confidence level means you accept fewer surprises, but it costs responses. 95% is the common default; 99% is used when the cost of being wrong is high.

**Responses for ±5% at different confidence levels**

|  | Responses |
| --- | --- |
| 90% (z 1.645) | 271 |
| 95% (z 1.96) | 385 |
| 99% (z 2.576) | 664 |

*Moving from 95% to 99% asks for about 70% more people for the same margin of error.*

## Does population size matter?

For large groups, no. Surveying a town of 100,000 needs about the same 383 responses as a country of 100 million. For small groups there is a **finite population correction**: n = n₀ ÷ (1 + (n₀ − 1) ÷ N), where n₀ is the basic answer (384.16) and N is the population.

**Responses needed for ±5% at 95% confidence, by population size**

|  | Value |
| --- | --- |
| 200 | 132 |
| 500 | 218 |
| 1,000 | 278 |
| 5,000 | 357 |
| 10,000 | 370 |
| 100,000 | 383 |

*For a class or a company of 200 people you need 132 answers (66% of them). For 100,000 people you need 383, which is less than half a percent.*

**A team of 200: how many must answer for ±5%?**

| Group | Count |
| --- | --- |
| Must respond | 132 |
| Can stay silent | 68 |

*Small groups need a high response share. In a tiny group it is often simpler to ask everyone.*

## When you already know the answer will be lopsided

The 50% assumption is the safest, because it gives the largest sample. If you are confident that only about 10% of people will say yes, you need fewer. Do not use this shortcut unless you really have evidence.

| Expected share | Responses for ±5% (95%) |
| --- | --- |
| 50% (unknown, safest) | 385 |
| 30% | 323 |
| 20% | 246 |
| 10% | 139 |
| 5% | 73 |

*The formula uses p × (1 − p), which is biggest at 50%.*

> **Sample size is not the same as sample quality:** The formula assumes everyone in the group had an equal chance of being asked. An online poll that only reaches people who are already fans, or a form that most people ignore, can have thousands of answers and still be badly wrong. A smaller random sample usually beats a bigger self-selected one.

> **Plan for people who do not reply:** If you expect only 25% of invited people to respond and you need 385 answers, invite about 1,540 (385 ÷ 0.25). Response rate is often the real limit, not the formula.

## A short checklist

1. **Decide the margin of error** you can live with, such as ±5%.
2. **Choose the confidence level.** Use 95% unless there is a reason not to.
3. **Use 50%** for the expected share if you have no good estimate.
4. **Enter your population,** especially if it is under a few thousand people.
5. **Divide by your expected response rate** to know how many invitations to send.

## Frequently asked questions

### What sample size do I need for a survey?

For a large population, 385 responses give a ±5% margin of error at 95% confidence. For ±3% you need about 1,068, and for ±1% about 9,604.

### Why is 385 so common?

It comes from the formula n = z² × p × (1 − p) ÷ e² with 95% confidence (z = 1.96), a 50% expected split and a 5% margin of error, which gives 384.16, rounded up to 385.

### Does population size matter for sample size?

Only for small groups. Beyond a few thousand people the required sample barely changes. For small populations a finite population correction reduces the number you need.

### What is margin of error?

It is the range around your result in which the true value is expected to lie at your chosen confidence level. A result of 60% with ±5% means the real figure is likely between 55% and 65%.

### What confidence level should I use?

95% is standard. Use 90% for quick, low-stakes checks and 99% when a wrong conclusion would be costly, remembering that higher confidence needs more responses.

### Can a large sample still be wrong?

Yes. If the people who answer are not representative of the group, for example only volunteers or only one region, a large sample gives a precise but biased result.



---

# sitemap.xml explained: what to include, the 50,000 URL limit and HTML sitemaps

> A sitemap is a list of pages you want search engines to find. It does not guarantee ranking, but it helps new and deep pages get discovered. This guide covers the tags that matter, the limits, the mistakes that waste crawl time and how an HTML sitemap helps real visitors.

Published 2026-10-09 by Toolcloak. 3 minute read. Original: https://toolcloak.pages.dev/articles/sitemap-xml-explained-what-to-include-the-50000-url-limit-and-html-sitemaps
Related free tool: [Sitemap XML Generator](https://toolcloak.pages.dev/tools/sitemap-xml-generator)

Search engines find pages by following links. That works well for pages linked from your home page, and poorly for new pages, deep pages or pages nobody has linked yet. A **sitemap.xml** file hands the crawler a clean list of the addresses you want found, along with when each last changed.

> **In short**
> - A sitemap helps **discovery**. It does not force indexing or improve rankings by itself.
> - One file can hold up to **50,000 URLs** or **50 MB** uncompressed. Bigger sites split into several files and list them in a **sitemap index**.
> - Only list pages you want in search: real, canonical, indexable addresses that return a normal 200 response.
> - Google reads `loc` and an accurate `lastmod`. It ignores `priority` and `changefreq`.
> - An **HTML sitemap** is a different thing: a page for people, with links to your main sections.

## How a sitemap fits into crawling

**From file to search results**

1. **You publish /sitemap.xml.** The file sits at the root of your site, for example example.com/sitemap.xml.
2. **You tell crawlers.** Add a Sitemap: line to robots.txt and submit the file in Google Search Console and Bing Webmaster Tools.
3. **The crawler reads it.** It queues the listed URLs and notes the lastmod dates.
4. **Pages are fetched.** Each page is checked for quality, redirects, blocks and duplicate content.
5. **Good pages are indexed.** Only some listed pages will be indexed. The sitemap is a request, not an order.

*Many sites have thousands of pages listed but far fewer indexed. The gap is a quality signal worth investigating.*

Try it: [Sitemap XML Generator](https://toolcloak.pages.dev/tools/sitemap-xml-generator). Paste your URLs, set dates and download a ready-to-upload sitemap.xml.

## What a sitemap file looks like

A sitemap is plain XML. It starts with a `urlset` wrapper, and each page sits in its own `url` block. Only `loc` is required.

| Tag | Required? | What it holds | Does Google use it? |
| --- | --- | --- | --- |
| `loc` | Yes | The full page address, including https:// | Yes |
| `lastmod` | No | The date the page last changed in a meaningful way, e.g. 2026-10-09 | Yes, when it proves accurate |
| `changefreq` | No | A hint such as daily or monthly | No, ignored |
| `priority` | No | A number from 0.0 to 1.0 | No, ignored |

*Do not set lastmod to today on every build. If the date changes without the content changing, search engines learn to ignore it.*

## The size limits

One sitemap file may contain at most **50,000 URLs** and be at most **50 MB** uncompressed. The URL count is usually the limit you hit first. Go beyond it and you split the list into several files, then list those files in a **sitemap index**.

**A 120,000-page site split into sitemap files**

|  | URLs in the file |
| --- | --- |
| sitemap-1.xml | 50,000 |
| sitemap-2.xml | 50,000 |
| sitemap-3.xml | 20,000 |

*A small sitemap-index.xml then lists the three files. Submit only the index.*

**How many sitemap files for different site sizes**

|  | Value |
| --- | --- |
| 800 pages | 1 |
| 50k pages | 1 |
| 120k pages | 3 |
| 500k pages | 10 |
| 2M pages | 40 |

*Files needed = pages ÷ 50,000, rounded up. Many large sites split by section (blog, products) rather than by count, which makes problems easier to trace.*

## What to include and what to leave out

**A simple rule: list only pages you would be happy to see in search**

| Box | What it means |
| --- | --- |
| Include | Canonical, indexable pages that return 200: articles, tools, category pages, key landing pages. |
| Leave out | Pages with noindex, redirects (3xx), errors (404, 500) and duplicate versions of the same page. |
| Also leave out | Login, cart, internal search results, filters and tracking-parameter URLs. |
| Keep consistent | Use the same version everywhere: https, one host name and either trailing slashes or none. |

*A sitemap full of redirected or blocked URLs wastes crawl attention and makes reports noisy.*

**Healthy sitemap against messy sitemap**

| Messy | Healthy |
| --- | --- |
| Hurts more than helps | Easy to trust |
| Includes 404 and redirected URLs | Only live, canonical pages |
| Mixes http and https, www and non-www | One consistent address format |
| lastmod changes on every deploy | lastmod changes only when content does |

*Search engines build trust in a sitemap over time. Accurate files get read more attentively.*

## Where to put it and how to announce it

1. **Save it at the site root** as `/sitemap.xml`. A sitemap can only list URLs at or below its own folder.
2. **Add one line to robots.txt:** `Sitemap: https://yourdomain.com/sitemap.xml` (use the full address).
3. **Submit it** in Google Search Console and Bing Webmaster Tools, under Sitemaps.
4. **Check the report.** Look for 'discovered' versus 'indexed' counts and any errors.
5. **Regenerate on change.** New pages should appear automatically; this is easy if your build creates the file.

## XML sitemap versus HTML sitemap

**Two things with the same name**

| XML sitemap | HTML sitemap |
| --- | --- |
| For crawlers | For visitors |
| File at /sitemap.xml | A normal page, such as /site-map |
| Lists every important URL with dates | Links to sections, tools and articles |
| Not meant to be read by people | Helps people who are lost and also gives crawlers more links |

*Use both. The XML file feeds search engines; the HTML page helps humans and adds internal links.*

A good HTML sitemap groups links by topic, stays readable (a few hundred links at most on one page) and is linked from the footer of every page. It also helps with accessibility, since it is a reliable second way to navigate.

> **Common mistakes:** Listing pages blocked in robots.txt, mixing http with https, forgetting to update after a domain change, including more than 50,000 URLs in one file, leaving test pages in the list and setting every lastmod to the build date.

## Quick checklist

1. Every URL is absolute, starts with https:// and uses the same host.
2. Each page returns 200, has no noindex and points to itself as canonical.
3. The file is UTF-8 and the special characters in addresses are properly escaped.
4. The file is under 50,000 URLs and 50 MB, or split with an index.
5. robots.txt points to the sitemap and the sitemap is submitted in Search Console.

## Frequently asked questions

### What is a sitemap.xml?

It is an XML file listing the URLs of a website that you want search engines to discover, optionally with the date each page last changed. It usually lives at /sitemap.xml.

### Do I need a sitemap for a small website?

Not strictly, since a well-linked small site is easily crawled. A sitemap still helps new pages appear sooner and gives you indexing reports, so it is worth having.

### How many URLs can a sitemap hold?

Up to 50,000 URLs and 50 MB uncompressed per file. Larger sites use several sitemap files listed in a sitemap index file.

### Does Google use priority and changefreq?

No. Google ignores both. It reads the loc tag and uses lastmod when the dates have proved reliable.

### What is the difference between an HTML and an XML sitemap?

An XML sitemap is a machine-readable file for search engines. An HTML sitemap is a normal web page with links for human visitors, and it also adds internal links for crawlers.

### How do I tell search engines about my sitemap?

Add a Sitemap line with the full address to robots.txt and submit the file in Google Search Console and Bing Webmaster Tools.



---

# Speed, distance and time calculator: how to find speed, distance or travel time, and why average speed is not a simple average

> Speed, distance and time are linked by one formula, and knowing any two gives the third. This guide shows how to solve each case, how to convert km/h to metres per second, and why a trip at two speeds does not average to the middle value.

Published 2026-10-09 by Toolcloak. 3 minute read. Original: https://toolcloak.pages.dev/articles/speed-distance-time-calculator-how-to-find-speed-distance-or-travel-time-and-why-average-speed-is-not-a-simple-average
Related free tool: [Speed, Distance & Time Calculator](https://toolcloak.pages.dev/tools/speed-distance-time-calculator)

**Speed**, **distance** and **time** are three views of the same thing: how far you go, how long it takes, and how fast you move. Know any two and you can find the third. The formulas are short, but the units and the idea of average speed trip people up, especially when a journey is made at different speeds.

> **In short**
> - **Speed = distance ÷ time.** **Distance = speed × time.** **Time = distance ÷ speed.**
> - A trip of **150 km** at **60 km/h** takes **2.5 hours**, which is **2 h 30 min**.
> - Speed in km/h ÷ 3.6 gives metres per second, so 60 km/h is **16.67 m/s**.
> - Going 60 km at 60 km/h and coming back at 40 km/h is an average of **48 km/h**, not 50.

## The three formulas

**Example: how long does 150 km take at 60 km/h?**

1. **Write down what you know.** Distance = 150 km, speed = 60 km/h.
2. **Choose the formula.** Time = distance ÷ speed.
3. **Divide.** 150 ÷ 60 = **2.5 hours**.
4. **Convert to hours and minutes.** 0.5 hour is 30 minutes, so the trip takes **2 h 30 min**.

*Choose what to find, fill in the other two values and press Solve. The tool shows time in hours and minutes as well as decimal hours.*

Try it: [Speed, Distance & Time Calculator](https://toolcloak.pages.dev/tools/speed-distance-time-calculator). Pick what you want to find, enter the other two values, and get speed, distance or time.

## Keep the units matching

The tool uses kilometres, hours and km/h. If your numbers are in other units, convert first. Minutes must become hours: 90 minutes is 1.5 hours, not 90. Metres must become kilometres: 2,500 m is 2.5 km. Mixing km with minutes is the most common reason an answer is off by a factor of 60.

- **km/h to m/s:** divide by 3.6. **m/s to km/h:** multiply by 3.6.
- **Minutes to hours:** divide by 60. **Hours to minutes:** multiply by 60.
- **Metres to km:** divide by 1,000.

## Travel time at different speeds

| Speed | Time for 150 km | Speed in metres per second |
| --- | --- | --- |
| 40 km/h | 3 h 45 min | 11.11 m/s |
| 50 km/h | 3 h 0 min | 13.89 m/s |
| 60 km/h | 2 h 30 min | 16.67 m/s |
| 75 km/h | 2 h 0 min | 20.83 m/s |
| 90 km/h | 1 h 40 min | 25.00 m/s |

*The same 150 km. Doubling the speed halves the time, but going from 40 to 60 km/h saves more minutes than going from 75 to 90.*

**Minutes needed for 150 km at different speeds**

|  | Minutes |
| --- | --- |
| 40 km/h | 225 min |
| 50 km/h | 180 min |
| 60 km/h | 150 min |
| 75 km/h | 120 min |
| 90 km/h | 100 min |

*Extra speed gives smaller and smaller savings. Going from 40 to 60 km/h saves 75 minutes here, while going from 75 to 90 km/h saves only 20.*

## Average speed is not a simple average

Suppose you drive 60 km to a town at 60 km/h and return at 40 km/h. The simple average of the two speeds is 50, but the real average speed is **total distance ÷ total time**. The way there takes 1 hour and the way back takes 1.5 hours. The total is 120 km in 2.5 hours, which is **48 km/h**. You spend longer at the slower speed, so it pulls the average down.

**Same trip, two ways of averaging**

| Simple average | Total distance ÷ total time |
| --- | --- |
| 50 km/h | 48 km/h |
| (60 + 40) ÷ 2 | 120 km ÷ 2.5 h |
| Ignores the time spent at each speed | Counts the longer time at 40 km/h |
| Would give 2 h 24 min for 120 km | Gives 2 h 30 min for 120 km |

*Always add the distances and add the times first, then divide.*

> **The tool does not add stops:** The result is time on the move. Fuel stops, traffic, tolls and rest breaks add to the real journey, so add a buffer when planning. A flat extra 10% is a common rough allowance on long drives.

## Solving each case

1. **Decide what is missing:** speed, distance or time.
2. **Check the units** and convert minutes, metres or m/s first.
3. **Use the matching formula** or choose that option in the calculator.
4. **For a journey in parts,** work out distance and time for each part, add them up, then find the average speed.
5. **Add a buffer** for stops and traffic when you plan the real trip.

## Frequently asked questions

### How do I calculate travel time?

Divide the distance by the speed. For 150 km at 60 km/h, the time is 150 ÷ 60 = 2.5 hours, which is 2 hours 30 minutes.

### How do I calculate speed from distance and time?

Divide the distance by the time in hours. A 90 km trip in 1.5 hours means a speed of 60 km/h.

### How do I convert km/h to m/s?

Divide by 3.6. Sixty km/h is 60 ÷ 3.6 = 16.67 metres per second. To go back, multiply metres per second by 3.6.

### What is average speed?

Average speed is total distance divided by total time. It is not the average of the speeds, unless you spend equal time at each speed.

### Why is the average of 60 and 40 km/h not 50?

Because over equal distances you spend more time at the slower speed. The trip takes 1 hour plus 1.5 hours, so 120 km in 2.5 hours is an average of 48 km/h.

### Does the calculator work for miles?

The fields are in km, km/h and hours. For miles and mph the formulas are the same, so enter the numbers in miles and mph and read the answer as miles and mph. The metres-per-second line would then not apply. The [Speed Converter](https://toolcloak.pages.dev/tools/speed-converter) changes between units.



---

# Subscription cost calculator: how small monthly payments add up, and how to audit them

> A few streaming, app and gym plans can quietly cost more than a weekend trip every year. This guide shows how to add them up per month, per year and over five years, how to compare monthly and annual plans, and a simple audit to decide what to keep.

Published 2026-10-09 by Toolcloak. 3 minute read. Original: https://toolcloak.pages.dev/articles/subscription-cost-calculator-how-small-monthly-payments-add-up-and-how-to-audit-them
Related free tool: [Subscription Cost Calculator](https://toolcloak.pages.dev/tools/subscription-cost-calculator)

A **subscription** is a payment that repeats until you stop it. Each one looks small, which is the point. Added together, they are a fixed cost that leaves your account every month whether you used the service or not. The first step to controlling them is to see the total.

> **In short**
> - Add every plan at its **monthly** price: Netflix ₹649, Spotify ₹119, Cloud storage ₹130 and Gym ₹1,500 come to **₹2,398 a month**.
> - That is **₹28,776 a year** and **₹1,43,880 over five years** at today's prices.
> - An annual plan that costs less than 12 monthly payments saves money **only if you will really use it** for the full year.
> - Review the list every six months. Cancel what you have not used in the last 30 days.

## How the numbers are worked out

Convert everything to one period, then multiply. If a price is yearly, divide it by 12 to get the monthly cost. Then multiply the monthly total by 12 for a year and by 60 for five years.

**Example: four subscriptions**

1. **List each price per month.** Netflix ₹649 · Spotify ₹119 · Cloud storage ₹130 · Gym ₹1,500
2. **Add the monthly prices.** ₹2,398 per month.
3. **Multiply by 12.** ₹2,398 × 12 = **₹28,776** per year.
4. **Multiply the yearly figure by 5.** ₹28,776 × 5 = **₹1,43,880** over five years.

*The calculator reads one subscription per line, with the price at the end of the line, and shows per month, per year, in 5 years and the number of items.*

Try it: [Subscription Cost Calculator](https://toolcloak.pages.dev/tools/subscription-cost-calculator). Type one subscription per line, choose whether the prices are monthly or yearly, and see the full cost.

## Where the money goes

**Yearly cost of each item in the example (₹)**

|  | Per year |
| --- | --- |
| Gym | ₹18,000 |
| Netflix | ₹7,788 |
| Cloud storage | ₹1,560 |
| Spotify | ₹1,428 |

*The gym is 63% of the monthly total. The three small digital plans together cost ₹898 a month, about 37%.*

The lesson from the chart is that the biggest bill is usually not the one you worry about. One large, rarely used plan can cost more than all your app subscriptions together.

## Five years is a big number

**The example list over time**

|  | Value | Note |
| --- | --- | --- |
| Per month | ₹2,398 | Four subscriptions |
| Per year | ₹28,776 | Monthly × 12 |
| In 5 years | ₹1,43,880 | At today's prices |
| If invested instead | ₹1.98 L | ₹2,398 a month for 5 years at 12%, an illustration |

*The last figure is a what-if. Investing ₹2,398 every month at an assumed 12% a year grows to about ₹1.98 L in five years. Returns are never guaranteed, and some subscriptions are worth their cost.*

> **This is not about cutting everything:** A gym you use five times a week or a work tool that saves hours is good value. The goal is to pay on purpose, not by default. Also remember prices rise: many services raise fees every year or two, so five years at today's prices is a floor.

## Monthly or annual: which plan is cheaper?

Many services offer a discount if you pay for a year in advance. A plan at ₹129 a month costs ₹1,548 over 12 months. If the annual plan is ₹1,299, you save ₹249, which is 16.1%. Divide the yearly price by 12 to compare: ₹7,188 a year equals ₹599 a month.

**One service, two ways to pay**

| Monthly plan | Annual plan |
| --- | --- |
| ₹1,548 | ₹1,299 |
| ₹129 × 12 months | Saves ₹249 (16.1%) |
| Costs ₹249 more over a full year | Paid in one go |
| Cancel any month, no lock-in | Usually non-refundable if you stop early |

*The annual plan wins only if you will use the service for the whole year. If you stop after five months, the monthly plan would have cost less.*

> **Watch the renewal date:** Annual plans renew automatically in most apps, and the charge arrives as one big amount. Put the renewal date in your calendar a week early, so you can decide before you are billed.

## Cost per use

A cheap-looking monthly fee can be expensive per use. A gym membership of ₹1,500 used 8 times a month costs ₹187.50 per visit. At 20 visits it is ₹75 per visit. The same plan is a bargain or a waste depending only on how often you go.

| Visits or uses per month | Cost per use of a ₹1,500 plan |
| --- | --- |
| 2 | ₹750 |
| 4 | ₹375 |
| 8 | ₹187.50 |
| 12 | ₹125 |
| 20 | ₹75 |

*Cost per use = monthly price ÷ number of uses. Halving your usage doubles what each visit costs.*

## A five-step subscription audit

1. **Find them all.** Check your bank and card statements, UPI history and app-store subscription pages for the last three months.
2. **Write one line for each,** with the name and the price, and put them in the calculator.
3. **Mark each one** as keep, review or cancel. Cancel what you have not used in 30 days.
4. **Switch to annual plans** only for services you are sure to use all year.
5. **Set a reminder** to repeat the audit every six months.

> **Free trials:** A free trial often turns into a paid plan without a reminder. Cancel on the day you sign up if you are unsure, because most services let you keep the trial until it ends.

## Frequently asked questions

### How do I calculate the total cost of my subscriptions?

Convert every price to a monthly amount, add them up, then multiply by 12 for a year. If you pay annually for something, divide that price by 12 first. The calculator does this when you choose whether the prices are monthly or yearly.

### How do I enter my subscriptions?

Type one subscription per line with the name first and the price at the end, such as Netflix 649. The calculator takes the number at the end of each line as the price.

### Is an annual subscription cheaper than a monthly one?

Often, but only if you use it for the whole year. Compare the annual price with 12 monthly payments. If you might stop early, the monthly plan may cost less overall.

### How much do subscriptions cost per year?

Multiply the total monthly cost by 12. A set of subscriptions costing 2,398 a month comes to 28,776 a year and 1,43,880 over five years at the same prices.

### How can I find subscriptions I forgot about?

Scan the last three months of bank, card and UPI statements for repeated charges of the same amount. App-store subscription pages on your phone also list every active plan.

### Does the calculator account for price rises?

No. It uses the prices you enter and assumes they stay the same. Many services raise their prices from time to time, so treat the five-year figure as a minimum. Use the [Subscription Cost Calculator](https://toolcloak.pages.dev/tools/subscription-cost-calculator) again whenever a price changes.



---

# How to calculate a tip and split a bill fairly, with 10%, 15% and 20% shortcuts

> Work out the tip in your head in two seconds, see what each person owes when the bill does not divide evenly, and learn when to check for a service charge first. Includes lookup tables, a worked example and a fair way to settle up at the table.

Published 2026-10-09 by Toolcloak. 5 minute read. Original: https://toolcloak.pages.dev/articles/how-to-calculate-a-tip-and-split-a-bill-fairly-with-10-15-and-20-percent-shortcuts
Related free tool: [Tip & Bill Split Calculator](https://toolcloak.pages.dev/tools/tip-calculator)

A **tip** is a percentage of the bill that you add as thanks for service. The arithmetic is one multiplication and one division, but doing it at a busy table with a calculator-less phone and four hungry friends is where mistakes creep in. This guide gives you the formula, the head-math shortcuts and a clear way to split the total.

> **In short**
> - **Tip = bill × tip% ÷ 100.** On ₹2,400 at 10% the tip is **₹240** and the total is **₹2,640**.
> - **Each person pays = total ÷ number of people.** Four people share ₹2,640 as **₹660** each.
> - Shortcuts: **10%** means move the decimal one place, **15%** is 10% plus half of it, **20%** is double the 10% figure.
> - Look at the bill first. Many restaurants already add a **service charge**, and paying both is usually not what you meant to do.

## The formula behind every tip

Three steps cover every case. Find the tip amount, add it to the bill to get the total, then divide the total by the number of people. The same three steps work for a coffee for two or a birthday dinner for twelve.

**Example: ₹2,400 bill, 10% tip, 4 people**

1. **Find the tip.** ₹2,400 × 10 ÷ 100 = **₹240**.
2. **Add it to the bill.** ₹2,400 + ₹240 = **₹2,640**.
3. **Split the total.** ₹2,640 ÷ 4 = **₹660** each.
4. **Check the tip share.** ₹240 ÷ 4 = **₹60** of each payment is the tip.

*The tool shows the same four numbers: what each person pays, the tip amount, the total bill and the tip per person.*

Try it: [Tip & Bill Split Calculator](https://toolcloak.pages.dev/tools/tip-calculator). Enter the bill, the tip percentage and the number of people to see each share instantly.

## Head-math shortcuts that work anywhere

1. **10%:** shift the decimal point one place to the left. A bill of 2,400 gives 240.
2. **5%:** take half of the 10% figure. Half of 240 is 120.
3. **15%:** add the 10% and 5% figures. 240 + 120 = 360.
4. **20%:** double the 10% figure. 240 × 2 = 480.
5. **12.5%:** one eighth of the bill. Halve three times: 2,400 → 1,200 → 600 → 300.

Round the bill to a friendly number first if you are in a hurry. A bill of 1,187 is close enough to 1,200 for a 15% estimate, and nobody will check your answer to the last rupee.

| Bill | 10% tip | 15% tip | 20% tip |
| --- | --- | --- | --- |
| ₹500 | ₹50 | ₹75 | ₹100 |
| ₹1,200 | ₹120 | ₹180 | ₹240 |
| ₹2,400 | ₹240 | ₹360 | ₹480 |
| ₹3,600 | ₹360 | ₹540 | ₹720 |

*Tips are shown before any rounding. Round up or down to whatever feels natural when you pay.*

**What each tip level adds to a ₹2,400 bill**

|  | Tip amount |
| --- | --- |
| 5% | ₹120 |
| 10% | ₹240 |
| 15% | ₹360 |
| 20% | ₹480 |
| 25% | ₹600 |

*Going from 10% to 20% doubles the tip but changes the total by less than 10%. The tip is a small slice of the total even at the high end.*

## Check the bill before you tip

Many restaurants add a **service charge** as a line on the bill, often between 5% and 12.5%. It is meant to replace the tip, and some places pass it on to staff while others do not. Paying a service charge and then adding a full tip on top means you have tipped twice.

> **Read the last lines of the bill:** Look for 'service charge', 'service fee' or 'gratuity included'. If it is there, you can ask the staff how it is shared and add a small extra only if you want to. A service charge is also not the same as tax, so check which lines were added to your total.

## Tipping habits differ by place

There is no single correct percentage. Customs vary by country, by type of venue and even by city, so treat the table below as a rough guide and confirm locally when you can.

| Where | Typical practice |
| --- | --- |
| United States and Canada | 15% to 20% is common at sit-down restaurants, usually added after the meal. |
| United Kingdom | 10% to 12.5% is common, and many restaurants add it as a service charge. |
| Much of continental Europe | Rounding up or leaving 5% to 10% for good service. |
| India | Often 5% to 10% where no service charge is added, or rounding up for small bills. |
| Japan and South Korea | Tipping is not expected and may confuse staff. |

*A rough guide only. Rules change, and some places have formal service-charge laws. Ask if you are unsure.*

## Splitting the bill: equal or itemised?

An equal split is fastest, but it is only fair when everyone ordered about the same. If one person had a starter, a main and dessert while another had a salad and water, an itemised split avoids quiet resentment. The tip can follow the same rule: each person adds the same percentage to their own items.

**The same ₹2,400 table, split two ways (10% tip)**

| Equal split | Itemised split |
| --- | --- |
| ₹660 each | ₹440 to ₹990 |
| One division, done in seconds | Asha ₹990, Ben ₹550 |
| Everyone pays the same | Chitra ₹660, Dev ₹440 |
| Fair when orders were similar | Fair when orders differ a lot |

*Items of 900, 500, 600 and 400 add up to 2,400. Each person pays their items plus 10%, and the four shares total 2,640, the same as the equal split.*

## When the numbers do not divide evenly

Real bills rarely divide neatly. Take a bill of ₹2,350 with a 15% tip shared by 3 people. The tip is ₹352.50, the total is ₹2,702.50 and each share is ₹900.83. Nobody can pay 83 paise in cash, so you round.

**A bill that does not divide evenly**

|  | Value | Note |
| --- | --- | --- |
| Bill | ₹2,350 | Before tip |
| Tip at 15% | ₹352.50 | ₹2,350 × 0.15 |
| Total | ₹2,702.50 | Bill plus tip |
| Each of 3 pays | ₹900.83 | Round to ₹901 |

*Rounding each share up to ₹901 collects ₹2,703, which is 50 paise more than needed. The extra goes to the server as a slightly larger tip.*

> **Round in the server's favour:** When a share ends in odd paise, round up. The difference is tiny for each person, and it keeps the final collection simple. If you want exact amounts, use a payment app that splits to the paisa.

## Tip on the pre-tax or post-tax amount?

Etiquette varies. Many diners tip on the pre-tax subtotal, because the tax is not a charge for service. Others tip on the full total because it is easier. The calculator tips on whichever amount you type in the bill field, so enter the figure you want the percentage to apply to. The difference is small, but it adds up on large group bills.

## A fair way to settle up at the table

1. **Read the bill** and look for service charge, tax and any discount.
2. **Decide the tip percentage** together before you start dividing.
3. **Enter the bill, the tip and the head count** in the calculator.
4. **Agree on equal or itemised** if orders were very different.
5. **Round up** each share a little rather than collecting exact paise.
6. **Pay in one go** if you can, so one person is not left out of pocket.

## Frequently asked questions

### How do I calculate a 15% tip quickly?

Find 10% by moving the decimal point one place, then add half of that. For a bill of 2,400 the 10% figure is 240, half is 120, so the tip is 360.

### How do I split a bill with a tip between several people?

Add the tip to the bill to get the total, then divide by the number of people. For a bill of 2,400 with a 10% tip shared by 4, the total is 2,640 and each person pays 660.

### Should I tip before or after tax?

There is no single rule. Many people tip on the pre-tax subtotal because tax is not a charge for service. Others tip on the total for simplicity. The calculator uses whichever amount you enter.

### What if a service charge is already on the bill?

A service charge is meant to replace the tip. You can ask how it is shared with staff and add a small extra only if you wish. Paying both in full usually means you have tipped twice.

### How much should I tip in a restaurant?

It depends on the country and venue. 15% to 20% is common in the United States, 10% to 12.5% in the United Kingdom, and 5% to 10% in many other places. Where tipping is not customary, it is not expected.

### Can I calculate a tip for one person?

Yes. Set the number of people to 1. The tool then shows the total and the tip for a single payer. The [Tip & Bill Split Calculator](https://toolcloak.pages.dev/tools/tip-calculator) accepts any number of people from 1 upwards.



---

# Unit price calculator: how to compare pack sizes and find the real best buy per gram or per litre

> Two packs, two prices and two sizes make it hard to see the cheaper one. Dividing price by quantity gives a unit price you can compare directly. This guide covers the method, the unit mix-ups to avoid, why the bigger pack is not always cheaper and how waste changes the answer.

Published 2026-10-09 by Toolcloak. 3 minute read. Original: https://toolcloak.pages.dev/articles/unit-price-calculator-how-to-compare-pack-sizes-and-find-the-real-best-buy-per-gram-or-per-litre
Related free tool: [Unit Price Comparison Calculator](https://toolcloak.pages.dev/tools/unit-price-calculator)

A **unit price** is the price of one unit of a product, such as one gram, one millilitre or one piece. It is the only fair way to compare two packs of different sizes. The calculation is one division, but shop labels, multi-buy offers and different units make it easy to pick the wrong pack.

> **In short**
> - **Unit price = price ÷ quantity.** Compare the unit prices, not the pack prices.
> - Pack A costs ₹120 for 500 g, which is **₹0.24 per g**. Pack B costs ₹210 for 1,000 g, which is **₹0.21 per g**.
> - Pack B is cheaper by **12.5%** per gram, even though it costs more on the shelf.
> - Both packs must use the **same unit**. Convert kilograms to grams or litres to millilitres first.

## The method

**Example: two packs of the same product**

1. **Pick one unit.** Use grams for both packs.
2. **Divide each price by its quantity.** A: ₹120 ÷ 500 = **₹0.24**. B: ₹210 ÷ 1,000 = **₹0.21**.
3. **Compare.** ₹0.21 is lower than ₹0.24, so **Pack B** is the better deal.
4. **Find the saving.** (0.24 − 0.21) ÷ 0.24 = **12.5%** cheaper per gram.

*The tool takes two prices, two quantities and a unit name, and shows the better deal, both unit prices and the percentage saved.*

Try it: [Unit Price Comparison Calculator](https://toolcloak.pages.dev/tools/unit-price-calculator). Enter the price and quantity of both packs, plus the unit, to see which one is cheaper per unit.

## Make the unit easier to read

A unit price of ₹0.24 per gram is correct, but it is hard to feel. Scale it to a size you recognise, such as 100 g or 1 kg, or use the same scale on every product you compare.

| Pack | Per gram | Per 100 g | Per kg |
| --- | --- | --- | --- |
| A: ₹120 for 500 g | ₹0.24 per g | ₹24 per 100 g | ₹240 per kg |
| B: ₹210 for 1,000 g | ₹0.21 per g | ₹21 per 100 g | ₹210 per kg |
| C: ₹250 for 1,000 g | ₹0.25 per g | ₹25 per 100 g | ₹250 per kg |

*Pack C is the same size as B but costs more, so it is the worst buy per gram.*

**Price per kg of the three packs (₹)**

|  | Per kg |
| --- | --- |
| A (500 g) | ₹240 |
| B (1 kg) | ₹210 |
| C (1 kg) | ₹250 |

*A bigger pack is not always cheaper. C is double the size of A and costs more per kilogram.*

## Mistakes that give the wrong answer

- **Mixed units.** Typing 500 for one pack and 1 (for 1 kg) for the other makes the 1 kg pack look hundreds of times dearer. Use grams for both, or kilograms for both.
- **Pack count against weight.** Six small tubs of 90 g are not the same as one 500 g tub. Compare weight, not the number on the front.
- **Servings instead of weight.** Serving sizes differ between brands, so compare the total weight or volume.
- **Offers that change the quantity.** 'Buy 2 get 1 free' means you pay for 2 and get 3 packs, so divide the price by 3 packs.
- **Shrinking packs.** The same price with a smaller pack is a hidden price rise, and a unit price makes it visible.

**A multi-buy offer**

| Single pack | Buy 2, get 1 free |
| --- | --- |
| ₹0.24 per g | ₹0.16 per g |
| ₹120 for 500 g | Pay ₹240 for 1,500 g |
| No offer | ₹240 ÷ 1,500 g |
| Pay the full unit price | A third off the unit price |

*Always divide by everything you actually receive. An offer is a good one only if you will use all of it.*

## When the bigger pack costs more in the end

Unit price assumes you use every gram. If food spoils or you never finish it, the effective price is higher. If a quarter of the 1 kg pack B is wasted, you pay ₹210 for 750 g that you use, which is ₹280 per kg. That is more than the ₹240 per kg of pack A, which you would use up completely.

**Effective cost per kg if some of it is wasted**

|  | Value | Note |
| --- | --- | --- |
| Pack A, all used | ₹240 | 500 g pack |
| Pack B, all used | ₹210 | 1 kg pack |
| Pack B, 25% wasted | ₹280 | Only 750 g used |
| Saving if all of B is used | 12.5% | Against pack A |

*Bulk buying pays off for things you use steadily and that keep well, such as rice, oil or detergent, and less for fresh or short-life items.*

> **The break-even waste:** Pack B stays cheaper than pack A as long as you waste less than about 12.5% of it. Beyond that, the smaller pack is the better buy.

## Use it for more than groceries

- **Household items** such as detergent, tissue paper and cooking oil.
- **Fuel and liquids,** where you compare price per litre.
- **Subscriptions,** where you can compare price per month across plans.
- **Printing and stationery,** comparing price per sheet or per piece.

## Frequently asked questions

### How do I calculate the unit price?

Divide the price by the quantity. For a pack of 500 grams at 120 rupees, the unit price is 120 ÷ 500 = 0.24 rupees per gram.

### How do I compare two products of different sizes?

Work out the unit price of each, in the same unit, and compare. The one with the lower unit price is cheaper per gram, litre or piece.

### Is the bigger pack always cheaper?

No. Some brands price larger packs higher per unit than the smaller ones. Always check the unit price. Also consider whether you will use the whole quantity.

### What unit should I use?

Any unit works as long as both products use the same one. Convert kilograms to grams, or litres to millilitres, before you enter the quantities.

### How do I work out a buy-two-get-one-free offer?

Pay for two, receive three. Divide the total price you pay by the total quantity you receive, which is three packs, to get the unit price.

### Can I compare more than two products?

The [Unit Price Calculator](https://toolcloak.pages.dev/tools/unit-price-calculator) compares two at a time. For three or more, work out the unit price of each pair and keep the cheapest.



---

# UTM parameters explained: source, medium, campaign and naming rules that keep reports clean

> UTM tags are short labels added to a link so your analytics can tell which post, email or ad sent each visitor. They work only if you name things the same way every time. This guide covers the five parameters, a naming system, the mistakes that split your data, and when not to use them.

Published 2026-10-09 by Toolcloak. 3 minute read. Original: https://toolcloak.pages.dev/articles/utm-parameters-explained-source-medium-campaign-and-naming-rules-that-keep-reports-clean
Related free tool: [UTM Link Builder](https://toolcloak.pages.dev/tools/utm-link-builder)

Someone clicks a link to your site and lands on a page. Analytics can tell they arrived from, say, a social network, but not whether it was your Monday post, your newsletter or a paid ad. **UTM parameters** are labels added to the end of a link that answer that question when the visitor arrives.

> **In short**
> - Five parameters exist: **utm_source**, **utm_medium**, **utm_campaign** (the core three) plus optional **utm_term** and **utm_content**.
> - They are **case-sensitive**. `Facebook` and `facebook` become two separate rows in your report.
> - Use a fixed **naming convention**: lowercase, hyphens for spaces, no random variations.
> - Never put UTM tags on **links inside your own site**. They can overwrite the real source of a session.

## What a tagged link looks like

The tags are added after a question mark, joined by ampersands. Everything before the question mark is your normal page address.

**https://example.com/offer?utm_source=newsletter&utm_medium=email&utm_campaign=diwali-sale**

1. **https://example.com/offer.** The page the visitor lands on. It is unchanged.
2. **utm_source=newsletter.** Where the click came from: the specific sender, site or list.
3. **utm_medium=email.** The type of channel: email, cpc, social, referral.
4. **utm_campaign=diwali-sale.** The named campaign this link belongs to.

*The visitor sees the same page. The tags only travel to analytics. Tags go before any #fragment in the address.*

Try it: [UTM Link Builder](https://toolcloak.pages.dev/tools/utm-link-builder). Fill in the page address and labels to get a clean, encoded tagged link.

## The five parameters

| Parameter | Answers | Example | Required? |
| --- | --- | --- | --- |
| `utm_source` | Who sent the visitor? | `newsletter`, `instagram`, `google` | Yes |
| `utm_medium` | What kind of channel? | `email`, `cpc`, `social`, `referral` | Yes in practice |
| `utm_campaign` | Which campaign? | `diwali-sale`, `launch-2026` | Yes in practice |
| `utm_term` | Which paid keyword? | `running-shoes` | Optional |
| `utm_content` | Which version of the link or ad? | `header-button`, `blue-banner` | Optional |

*Source and medium together are the backbone of most reports. Campaign groups them. Content is for A/B tests.*

**Think of them as four questions**

| Box | What it means |
| --- | --- |
| Where from? | utm_source. The brand, site or list: `instagram`, `partner-blog`, `weekly-digest`. |
| What kind? | utm_medium. The channel type, from a short fixed list. |
| Why? | utm_campaign. The promotion, launch or theme this link serves. |
| Which version? | utm_content. Which button, image or link within the same campaign. |

*If you can answer these four in a spreadsheet before creating the link, the tagging will be consistent.*

## Why naming matters so much

Analytics treats every different spelling as a different thing. If four people in your team tag the same channel four ways, one source becomes four rows, each with a fraction of the visits. Nobody sees the true total.

**One real source, four spellings (illustrative example, 700 visits)**

|  | facebook | Facebook | fb | FB |
| --- | --- | --- | --- | --- |
| What the report shows | facebook 420 | Facebook 135 | fb 90 | FB 55 |
| What you wanted to see | facebook 700 |

*No single row looks important, yet together they are the biggest source. Good naming keeps all 700 in one place.*

## A simple naming convention

| Rule | Do | Don't |
| --- | --- | --- |
| Case | `newsletter` | `Newsletter`, `NEWSLETTER` |
| Spaces | `diwali-sale` | `diwali sale`, `Diwali_Sale` |
| Medium | A short fixed list: `email`, `cpc`, `social`, `referral` | `e-mail`, `Email Blast`, `paid` |
| Dates | `launch-2026-10` | `Oct 9th launch!!` |
| Special characters | Letters, numbers and hyphens | `&`, `#`, `%`, accents, emoji |

*Write the rules in a shared sheet and tick each link off against them. Ten minutes of agreement saves hours of cleaning.*

> **Mediums with a built-in meaning:** Google Analytics groups traffic into default channels using the medium and source. Values such as `email`, `cpc` and `social` are recognised, while made-up ones like `blast` fall into 'Unassigned' or 'Other'. When in doubt, use the standard word.

## A worked example: one campaign across channels

Suppose you run a festive sale and promote it in four places, all to the same landing page. Same campaign name, different source and medium.

| Channel | utm_source | utm_medium | utm_campaign | utm_content |
| --- | --- | --- | --- | --- |
| Newsletter | `newsletter` | `email` | `diwali-sale` | `header-button` |
| Newsletter, second link | `newsletter` | `email` | `diwali-sale` | `footer-text` |
| Instagram bio | `instagram` | `social` | `diwali-sale` | `bio-link` |
| Google ad | `google` | `cpc` | `diwali-sale` | `ad-1` |
| Partner blog | `partner-blog` | `referral` | `diwali-sale` | `review-post` |

*Filter the campaign report by `diwali-sale` and every channel appears together. Split by content to see which button works.*

**Visits by channel in the example campaign (illustrative, 1,000 visits)**

|  | Share |
| --- | --- |
| Newsletter | 420 |
| Google ad | 280 |
| Instagram | 180 |
| Partner blog | 120 |

*These numbers are made up to show how a clean report reads. Your data will differ.*

## Mistakes that damage your data

1. **Tagging internal links.** A UTM link from your home page to another page of your site starts a new session and hides the real source.
2. **Inconsistent spelling.** Capital letters, underscores against hyphens and abbreviations all create new rows.
3. **Putting personal data in a tag.** Never include names, emails or phone numbers. The tag shows up in analytics and in URLs people share.
4. **Tagging links where parameters are stripped.** Some platforms rewrite or remove parameters. Test with a real click.
5. **Using spaces or symbols without encoding.** A space should be a hyphen or `%20`. A builder encodes it for you.
6. **No record.** Keep a sheet of every link, so you can explain a strange row later.

**Where to use UTM tags**

| Good places | Avoid |
| --- | --- |
| External links you control | Internal navigation |
| Email newsletters and campaigns | Menu and button links on your own site |
| Social bios and posts | Links between your own subdomains |
| Paid ads and sponsored posts | Organic search results (you cannot tag them) |
| QR codes on print | Links where you cannot control the page |

*For clicks inside your own site, use events or link IDs in your analytics tool instead.*

## Long links and sharing

A tagged link can run to 150 characters or more. Link shorteners keep it neat in print and posts, but keep the original in your sheet. QR codes for posters work well when the code carries a UTM link, so you can see how many scans came from each placement.

> **Respect privacy and consent:** UTM tags are labels about the campaign, not about the person. Keep them free of anything that identifies an individual, and follow the consent rules that apply to your analytics setup.

## Frequently asked questions

### What are UTM parameters?

They are labels added to a web address, such as utm_source and utm_campaign, that tell analytics tools which campaign, channel and link sent a visitor.

### Which UTM parameters are required?

Source, medium and campaign are the three you should always set. Term and content are optional and used for paid keywords and A/B tests.

### Are UTM parameters case-sensitive?

Yes. Facebook and facebook appear as two different sources in your report, so pick lowercase and stay with it.

### Can I use UTM tags on internal links?

No. Tagging links within your own website can start a new session and overwrite the real source, so your reports become inaccurate.

### What is the difference between utm_source and utm_medium?

Source is the specific origin, like newsletter or instagram. Medium is the type of channel, like email, social or cpc.

### Do UTM tags affect SEO?

They do not change rankings directly, but tagged URLs can create duplicate addresses. A canonical tag pointing to the clean address avoids confusion.



---

# How to add and remove VAT from a price, and why subtracting the percentage is wrong

> Adding VAT is a simple multiplication, but taking it back out trips up almost everyone. This guide shows both directions with worked examples, a rate table, the shortcut for finding the VAT inside a price, and how VAT is collected at every stage of a supply chain.

Published 2026-10-09 by Toolcloak. 4 minute read. Original: https://toolcloak.pages.dev/articles/how-to-add-and-remove-vat-from-a-price-and-why-subtracting-the-percentage-is-wrong
Related free tool: [VAT Calculator](https://toolcloak.pages.dev/tools/vat-calculator)

**VAT** (value added tax) is a percentage added to the selling price of goods and services in many countries. A price **without** VAT is called the **net** price, and a price **with** VAT is called the **gross** price. Going from net to gross is easy. Going back from gross to net is where most mistakes happen.

> **In short**
> - **Add VAT:** gross = net × (1 + rate ÷ 100). At 20%, a net price of 1,000 becomes **1,200**, with **200** of VAT.
> - **Remove VAT:** net = gross ÷ (1 + rate ÷ 100). From 1,200 at 20% you get back **1,000**.
> - Do **not** take 20% off 1,200. That gives 960, which is wrong, because the VAT was worked out on the smaller net price.
> - At 20%, the VAT inside a gross price is **one sixth** of it (16.67%), not one fifth.

## Adding VAT to a net price

Multiply the net price by 1 plus the rate written as a decimal. A rate of 20% becomes 1.20. The VAT amount is the gross price minus the net price.

**Example: net price 1,000 at 20% VAT**

1. **Start with the net price.** 1,000 before VAT.
2. **Work out the multiplier.** 1 + 20 ÷ 100 = **1.20**.
3. **Multiply.** 1,000 × 1.20 = **1,200** gross.
4. **Find the VAT.** 1,200 − 1,000 = **200** of VAT.

*The same steps work for any rate. Swap 1.20 for 1.05, 1.19 or 1.25 as needed.*

Try it: [VAT Calculator](https://toolcloak.pages.dev/tools/vat-calculator). Choose add or remove, enter the amount and the VAT rate, and see the net, VAT and gross figures.

## Removing VAT from a gross price

Divide the gross price by the same multiplier. This undoes the multiplication exactly. It is tempting to take 20% off, but 20% of the gross price is more than the VAT that was added, because the gross price is already bigger than the net one.

**Taking VAT out of 1,200 at 20%**

| Subtract 20% | Divide by 1.20 |
| --- | --- |
| 960 | 1,000 |
| 1,200 × 0.80 = 960 | 1,200 ÷ 1.20 = 1,000 |
| Removes too much (240 instead of 200) | Removes exactly 200 of VAT |
| The net price comes out 40 too low | Matches the original net price |

*The gross price includes VAT on the net price. Dividing reverses that step. Subtracting a percentage of the gross does not.*

> **The most common VAT mistake:** If a supplier quotes a VAT-inclusive price, never multiply by the rate and subtract. Always divide by 1 plus the rate. On large invoices the error is real money, and it also shows up in tax returns.

## How much VAT is hidden inside a gross price?

The VAT share of a gross price is rate ÷ (100 + rate). At 20% that is 20 ÷ 120, or one sixth. At 25% it is one fifth, and at 10% it is one eleventh. This is handy when you want to guess the VAT on a receipt in your head: divide the total by 6 for a 20% rate.

| VAT rate | Add VAT | Remove VAT | VAT share of gross | VAT inside a gross price of 1,000 |
| --- | --- | --- | --- | --- |
| 5% | × 1.05 | ÷ 1.05 | 4.76% | 47.62 |
| 10% | × 1.10 | ÷ 1.10 | 9.09% | 90.91 |
| 15% | × 1.15 | ÷ 1.15 | 13.04% | 130.43 |
| 19% | × 1.19 | ÷ 1.19 | 15.97% | 159.66 |
| 20% | × 1.20 | ÷ 1.20 | 16.67% | 166.67 |
| 21% | × 1.21 | ÷ 1.21 | 17.36% | 173.55 |
| 23% | × 1.23 | ÷ 1.23 | 18.70% | 186.99 |
| 25% | × 1.25 | ÷ 1.25 | 20.00% | 200 |

*The share of gross is always smaller than the rate itself. A 25% VAT means one fifth of the final price, not one quarter.*

**VAT contained in a gross price of 1,000, by rate**

|  | VAT amount |
| --- | --- |
| 5% | 47.62 |
| 10% | 90.91 |
| 20% | 166.67 |
| 25% | 200 |

*Doubling the rate from 10% to 20% does not double the VAT inside a gross price: 90.91 becomes 166.67, which is about 1.8 times as much.*

## Where the VAT goes at each stage

VAT is charged at every step in a supply chain, but each business gets back the VAT it paid on its own purchases. Only the value it adds is taxed. In the end the total VAT collected equals the rate applied to the final price, and the consumer is the one who really bears it.

**A 20% VAT chain from raw wood to a table in a shop**

1. **Sawmill sells wood for 100 net.** Charges **20** VAT and pays the whole 20 to the tax office.
2. **Carpenter sells a table for 300 net.** Charges **60** VAT, claims back the 20 paid, and pays **40**.
3. **Shop sells it for 500 net.** Charges **100** VAT, claims back the 60 paid, and pays **40**.
4. **The customer pays 600.** 500 plus **100** VAT. The tax office has received 20 + 40 + 40 = **100** in total.

*The total VAT collected, 100, is exactly 20% of the final net price of 500. Businesses only pay the tax on the extra value they add.*

## Standard rates in a few countries

Rates change, and most countries also have reduced rates for some goods, so always check the current figure with your tax authority. These standard rates give a feel for the range.

| Country | Standard rate (approximate) |
| --- | --- |
| United Kingdom | 20% |
| Germany | 19% |
| France | 20% |
| Spain | 21% |
| Sweden | 25% |

*Standard rates only. Reduced and zero rates exist for food, books, medicine and other categories in many countries. Confirm the current rate before you file a return.*

> **VAT, GST and sales tax:** Many countries use GST or a sales tax instead of VAT. GST works much like VAT, with credit for tax paid on purchases. A US-style sales tax is usually added at the till and is not shown inside the shelf price. If you want the Indian version, see the [GST Calculator](https://toolcloak.pages.dev/tools/gst-calculator).

## Rounding on invoices

Round the VAT to the smallest currency unit, usually two decimals. On an invoice with many lines you can either calculate VAT per line and add them up, or add the net amounts first and calculate VAT once. The totals may differ by a few cents. Follow the rule of your tax authority or accounting software and use the same method every time.

## A checklist for VAT calculations

1. **Decide the direction.** Do you have a net price (add VAT) or a gross price (remove VAT)?
2. **Use the right rate.** Look up the rate for that item and date, not just the standard rate.
3. **Add by multiplying, remove by dividing.** Never subtract a percentage of a gross price.
4. **Keep the net and VAT separate** on invoices and records.
5. **Round at the end,** following your local rule, and keep the same method for every invoice.

## Frequently asked questions

### How do I add VAT to a price?

Multiply the net price by 1 plus the rate as a decimal. For 20% VAT, multiply by 1.20. A net price of 1,000 becomes 1,200 and the VAT is 200.

### How do I remove VAT from a price?

Divide the gross price by 1 plus the rate as a decimal. For 20% VAT, divide by 1.20. A gross price of 1,200 gives a net price of 1,000 and VAT of 200.

### Why can I not just subtract 20% from a VAT-inclusive price?

Because the VAT was added to the net price, so it is 20% of the smaller number. Twenty percent of the gross price is larger than the VAT that was added, and the result comes out too low.

### How do I find the VAT inside a gross price quickly?

Use rate ÷ (100 + rate). At 20% that is one sixth of the gross price. At 10% it is one eleventh, and at 25% it is one fifth.

### What is the difference between net price and gross price?

The net price is the price before VAT. The gross price is the price including VAT. Invoices usually show the net amount, the VAT and the gross total on separate lines.

### Does the VAT Calculator work for any country or currency?

Yes. The [VAT Calculator](https://toolcloak.pages.dev/tools/vat-calculator) works with plain numbers, so you can use it for any currency and any VAT, GST or similar rate. Enter the rate that applies to your item.



---

# Waist-to-hip ratio: how to measure it and what the 0.85 and 0.90 lines mean

> BMI says nothing about where your body carries weight. Waist-to-hip ratio does. This guide shows how to measure correctly, how to read the number, why the same waist can give different results and what the ratio can and cannot tell you.

Published 2026-10-09 by Toolcloak. 3 minute read. Original: https://toolcloak.pages.dev/articles/waist-to-hip-ratio-how-to-measure-it-what-the-0-85-and-0-90-lines-mean
Related free tool: [Waist to Hip Ratio Calculator](https://toolcloak.pages.dev/tools/waist-to-hip-ratio-calculator)

Two people can weigh the same and have the same BMI, yet carry fat in different places. Fat stored around the middle is linked with higher health risk than fat stored on the hips and thighs. **Waist-to-hip ratio (WHR)** is a simple tape-measure check for that difference: waist circumference divided by hip circumference.

> **In short**
> - **WHR = waist ÷ hip**, both in the same unit. A 82 cm waist and 98 cm hip give 0.84.
> - The World Health Organization uses **0.90 for men and 0.85 for women** as the point where abdominal obesity is flagged.
> - Measure carefully: small tape mistakes change the result a lot.
> - WHR is a screening clue, not a diagnosis. Talk to a doctor about your own health.

## How to measure properly

**Five steps for a reliable measurement**

1. **Stand relaxed.** Feet together, arms by your sides, breathe out normally. Do not pull your stomach in.
2. **Waist.** Wrap the tape around the narrowest part of your torso, usually just above the belly button. If unclear, use the midpoint between the lowest rib and the top of the hip bone.
3. **Hip.** Measure around the widest part of your buttocks, with the tape level all the way round.
4. **Keep the tape snug.** Firm against the skin but not squeezing. Measure over light clothing or bare skin.
5. **Divide.** Waist ÷ hip, in the same unit (cm or inches). Repeat twice and average.

*Measure at the same time of day, ideally in the morning before eating. Bloating after a meal can add a few centimetres.*

Try it: [Waist to Hip Ratio Calculator](https://toolcloak.pages.dev/tools/waist-to-hip-ratio-calculator). Enter your waist and hip to get the ratio and where it sits against the WHO lines.

## A worked example

**Waist 82 cm, hip 98 cm**

|  | Value | Note |
| --- | --- | --- |
| Waist | 82 cm | About 32.3 inches |
| Hip | 98 cm | About 38.6 inches |
| Ratio | 0.84 | 82 ÷ 98 = 0.837 |
| WHO line for women | Below 0.85 | Under the flag |

*The unit cancels out, so centimetres and inches give the same ratio. Just do not mix them.*

## Where the lines are

The WHO report on waist circumference and waist-hip ratio (2008) describes abdominal obesity as a ratio of **0.90 or more in men** and **0.85 or more in women**. Other bodies use slightly different bands, and some risk tables are age-specific, so treat the lines as a guide.

**Women: WHR and the 0.85 line**

| Zone | From | To |
| --- | --- | --- |
| Below the line | 0.65 | 0.85 |
| At or above | 0.85 | 1.05 |

*The marker shows the 82 cm / 98 cm example. At or above 0.85, the WHO flags abdominal obesity.*

**Men: WHR and the 0.90 line**

| Zone | From | To |
| --- | --- | --- |
| Below the line | 0.7 | 0.9 |
| At or above | 0.9 | 1.1 |

*The marker shows a man with a 96 cm waist and 102 cm hip: 96 ÷ 102 = 0.94, above the 0.90 line.*

## Same waist, different hips

WHR changes when either measurement changes. Take a waist of 85 cm. With narrower hips the ratio climbs, with wider hips it falls. That is why WHR describes **shape**, not just size.

**Waist 85 cm: how the ratio changes with hip size**

|  | Value |
| --- | --- |
| Hip 90 | 0.944 |
| Hip 95 | 0.895 |
| Hip 100 | 0.85 |
| Hip 105 | 0.81 |
| Hip 110 | 0.773 |

*With hip 100 cm the ratio is exactly 0.85. Wider hips lower the ratio, but only the waist side of the story is linked to the extra risk.*

## Different bodies, different results

| Waist (cm) | Hip (cm) | Ratio | Vs WHO lines |
| --- | --- | --- | --- |
| 70 | 96 | 0.73 | Under both lines |
| 78 | 100 | 0.78 | Under both lines |
| 82 | 98 | 0.84 | Just under 0.85 |
| 90 | 100 | 0.90 | At the men's line, over the women's |
| 96 | 102 | 0.94 | Over both lines |
| 100 | 104 | 0.96 | Over both lines |

*The same ratio can sit on different sides of the line depending on sex. That is why the calculator asks which line to use.*

## WHR, waist-to-height and BMI compared

**What each measure notices**

| BMI | WHR / waist-to-height |
| --- | --- |
| Weight for height | Where you carry it |
| Cannot tell fat from muscle | Picks up belly fat |
| Ignores where fat sits | Needs only a tape |
| Easy to calculate | Not designed for children, pregnancy or athletes |

*None of them is perfect alone. Doctors often use several measures together.*

Another simple rule is the **waist-to-height ratio**: keep your waist below half your height. For 170 cm that means under 85 cm. It uses one measurement and is easy to remember.

| Height | Half of height (waist guide) |
| --- | --- |
| 150 cm | 75 cm |
| 160 cm | 80 cm |
| 170 cm | 85 cm |
| 180 cm | 90 cm |

*A rule of thumb, not a medical limit. Age, ethnicity and health history can change what is appropriate.*

> **What this number cannot do:** WHR cannot diagnose any condition, and it is not meant for children, pregnant women or people with certain medical conditions. Muscle build, body frame and ethnicity can shift what a healthy ratio looks like. If your number is above the line, a doctor can check blood pressure, blood sugar and cholesterol. Do not use it to judge yourself harshly.

## Common mistakes

1. **Measuring over thick clothes,** which adds centimetres to both numbers.
2. **Pulling in the stomach,** which makes the waist look smaller than it is.
3. **Tape not level.** A tilted tape reads longer.
4. **Mixing units,** such as a waist in inches and a hip in centimetres.
5. **Using different spots each time,** so changes you see are just noise.

## Frequently asked questions

### What is waist-to-hip ratio?

It is your waist circumference divided by your hip circumference. It describes how your body fat is distributed, with a higher ratio meaning more weight carried around the middle.

### What is a healthy waist-to-hip ratio?

The WHO flags abdominal obesity at 0.90 or above for men and 0.85 or above for women. Below those numbers is generally considered lower risk, though it is only a screening guide.

### Where exactly should I measure my waist?

At the narrowest part of the torso, usually just above the belly button, or halfway between the lowest rib and the top of the hip bone. Stand relaxed and breathe out normally.

### Is waist-to-hip ratio better than BMI?

It answers a different question. BMI compares weight with height, while WHR shows where fat is stored. Many researchers find central fat a useful risk signal, so the two are often used together.

### Do I use centimetres or inches?

Either one, as long as both measurements use the same unit. The unit cancels out when you divide.

### Can I use WHR if I am an athlete or pregnant?

It is not designed for those cases. Muscle build can change hip and waist size, and pregnancy changes the waist for obvious reasons. Ask a health professional instead.



---

# Weighted average calculator: how to find an average when some values count more than others

> A plain average treats every number the same. A weighted average lets bigger or more important values count for more. This guide shows the formula, a worked example with course marks, how it differs from the simple average, and uses in grades, prices and portfolios.

Published 2026-10-09 by Toolcloak. 3 minute read. Original: https://toolcloak.pages.dev/articles/weighted-average-calculator-how-to-find-an-average-when-some-values-count-more-than-others
Related free tool: [Weighted Average Calculator](https://toolcloak.pages.dev/tools/weighted-average-calculator)

A **weighted average** is an average in which each value has its own weight, showing how much it counts. A final exam worth 50% should move your course mark more than a quiz worth 10%. A purchase of 500 units should move your average price more than a purchase of 20. The plain average cannot do that, and the weighted one can.

> **In short**
> - **Weighted average = sum of (value × weight) ÷ sum of weights.**
> - Marks of 85, 92 and 78 with weights 30, 50 and 20 give **87.10**.
> - The plain average of the same three marks is **85.00**. The weights pulled the result up, because the highest mark has the biggest weight.
> - Weights do not have to add up to 100. The formula divides by their total, so any positive numbers work.

## The formula, step by step

**Example: three marks with different weights**

1. **Multiply each value by its weight.** 85 × 30 = **2,550** · 92 × 50 = **4,600** · 78 × 20 = **1,560**
2. **Add the products.** 2,550 + 4,600 + 1,560 = **8,710**.
3. **Add the weights.** 30 + 50 + 20 = **100**.
4. **Divide.** 8,710 ÷ 100 = **87.10**.

*Type one item per line as the value and then the weight, such as 85 30. The tool shows the weighted average, the total weight and the number of items.*

Try it: [Weighted Average Calculator](https://toolcloak.pages.dev/tools/weighted-average-calculator). Enter one value and weight on each line to get the weighted average.

## How each value contributes

| Mark | Weight | Contribution to the average | Share of the weight |
| --- | --- | --- | --- |
| 85 | 30 | 25.5 | 30% |
| 92 | 50 | 46.0 | 50% |
| 78 | 20 | 15.6 | 20% |

*The contributions add up to 87.10. The 92 carries half the weight, so it contributes about half of the result.*

**Weighted against plain average**

| Weighted average | Plain average |
| --- | --- |
| 87.10 | 85.00 |
| Weights 30, 50 and 20 | (85 + 92 + 78) ÷ 3 |
| The 92 counts for half | Every mark counts the same |
| Matches how a course is graded | Ignores how much each is worth |

*The difference is 2.10 points here. It can be much larger when the weights are very uneven.*

## When all weights are equal

If every weight is the same, the weighted average equals the plain average. Entering 85 1, 92 1 and 78 1 gives 85.00. So a plain average is just a weighted average with equal weights, and the weighted version is the more general tool.

## Uses beyond marks

- **Average buying price.** If you buy 50 units at ₹100, 30 at ₹120 and 20 at ₹90, the average price is (5,000 + 3,600 + 1,800) ÷ 100 = ₹104 per unit, not the plain ₹103.33 of the three prices.
- **Course and exam marks,** where each part has a stated weight.
- **Portfolio returns,** where each holding's return is weighted by how much is invested in it.
- **Survey scores,** where groups of different sizes are combined.
- **Ratings,** where a product with 500 reviews should count for more than one with 5.

**Average buying price, the same idea**

|  | Value | Note |
| --- | --- | --- |
| Weighted average price | ₹104.00 | Weights are the units bought |
| Plain average of prices | ₹103.33 | Ignores the units |
| Total units | 100 | The sum of the weights |
| Total spent | ₹10,400 | Sum of price × units |

*For prices and units, the weighted average is the price per unit you actually paid. The [Stock Average Calculator](https://toolcloak.pages.dev/tools/stock-average-calculator) uses the same idea.*

> **Check what the weights mean:** In a course, weights are shares of the final grade. In a purchase, weights are units. Do not mix them in one list. If a weight is zero, that value does not count at all, and negative weights are rejected by the tool.

## Using it well

1. **List each value** and write its weight next to it.
2. **Make sure the weights match one idea:** all shares, or all units.
3. **Run the calculator** and compare it with the plain average to see the effect of the weights.
4. **Check a result** by seeing if it falls between the lowest and the highest value. A weighted average always does.
5. **Use the weights to plan:** to lift the result, improve the values that carry the most weight.

## Frequently asked questions

### How do I calculate a weighted average?

Multiply each value by its weight, add the products, and divide by the sum of the weights. For 85, 92 and 78 with weights 30, 50 and 20, the result is 8,710 ÷ 100 = 87.10.

### Do the weights have to add up to 100?

No. The formula divides by the total of the weights, so they can add up to any positive number, such as 10, 100 or 1,000 units.

### What is the difference between a weighted and a simple average?

A simple average gives every value the same importance. A weighted average gives each value a weight, so some values count more.

### Can the weighted average be outside the range of the values?

No. With non-negative weights it always lies between the smallest and the largest value.

### How do I use it for course grades?

Enter each mark with the percentage the assessment is worth. If the weights do not add to 100 because some work is still to come, use the [Final Grade Calculator](https://toolcloak.pages.dev/tools/final-grade-calculator) to find the score you need.

### Is it the same as the average calculator?

The [Average Calculator](https://toolcloak.pages.dev/tools/average-calculator) treats every number equally. Use the [Weighted Average Calculator](https://toolcloak.pages.dev/tools/weighted-average-calculator) when the numbers carry different weights.



---

# Working days calculator: how to count business days between two dates and why holidays change the answer

> Calendar days and working days are not the same. This guide shows how to count business days between two dates, what changes when the end date is included or the week has a different day off, and how to subtract public holidays by hand.

Published 2026-10-09 by Toolcloak. 3 minute read. Original: https://toolcloak.pages.dev/articles/working-days-calculator-how-to-count-business-days-between-two-dates-and-why-holidays-change-the-answer
Related free tool: [Working Days Calculator](https://toolcloak.pages.dev/tools/working-days-calculator)

A deadline in **calendar days** and a deadline in **working days** can be weeks apart. Delivery promises, notice periods, payment terms and project plans all depend on which one they mean. Counting working days is simple once you decide three things: which days are off each week, whether the end date counts, and which public holidays to remove.

> **In short**
> - From **1 November 2026** to **30 November 2026**, counting both dates, there are **30 calendar days** and **21 working days** with a Saturday and Sunday off.
> - With only Sunday off the same month has **25 working days**. With no weekly off it has **30**.
> - If the end date is left out, the count becomes **20 working days** over 29 days.
> - The tool does **not** remove public holidays. Subtract the ones that fall on a working day yourself.

## How the count works

**Example: 1 to 30 November 2026, Saturday and Sunday off**

1. **Count every day in the range.** 1 November to 30 November, both included, is **30 days**.
2. **Find the weekly off days.** November 2026 has 4 Saturdays and 5 Sundays, so **9 days** are off.
3. **Subtract.** 30 − 9 = **21 working days**.
4. **Convert to work weeks.** 21 ÷ 5 = **4.2 work weeks**.

*The tool also shows total days counted, weekly off days and work weeks.*

Try it: [Working Days Calculator](https://toolcloak.pages.dev/tools/working-days-calculator). Pick a start date, an end date and the weekly off to count the working days between them.

## The weekly off changes the result

| Weekly off | Working days in November 2026 | Work week used |
| --- | --- | --- |
| Saturday and Sunday | 21 | 5 days |
| Sunday only | 25 | 6 days |
| No weekly off | 30 | 7 days |

*Same dates, 1 to 30 November 2026, both days counted. Many offices work a Monday to Friday week, while shops, factories and some services work six days.*

**Working days in November 2026 by weekly off**

|  | Value |
| --- | --- |
| Sat + Sun off | 21 |
| Sun off | 25 |
| No off | 30 |

*A six-day week gives about four more working days in this month than a five-day week.*

## Is the end date included?

The tool lets you choose. If you include the end date, a range from Monday 12 October to Friday 23 October 2026 has **10 working days**. If you leave the end date out, it has 9. Use the first when both days are worked days, such as a project that starts and finishes on the dates given. Use the second when the end date is the day something is due and the work should be finished before it.

> **Read the wording:** 'Within 10 working days' and 'by the 10th working day' can mean the same date or a different one, depending on whether the first day counts. Check how your contract or notice describes it.

## Public holidays are not removed

Public holidays differ by country, state and employer, so the tool counts only the weekly off. Look up the holidays in your range, keep the ones that fall on a working day, and subtract them. A holiday that lands on a Saturday or Sunday has already been taken out and must not be subtracted twice.

**A quarter, 1 October to 31 December 2026**

|  | Value | Note |
| --- | --- | --- |
| Calendar days | 92 | Both dates counted |
| Working days | 66 | Saturday and Sunday off |
| Work weeks | 13.2 | 66 ÷ 5 |
| Days off | 26 | Weekends only, no holidays |

*Subtract public holidays and your own leave from the working days to see the time you really have.*

## Where working-day counts are used

- **Delivery and shipping,** where a promise of 5 to 7 working days skips weekends.
- **Notice periods and payment terms,** which may be written in working days or in calendar days.
- **Project plans,** where tasks are estimated in working days and then laid on the calendar.
- **Pay and leave,** where a month's pay or a leave balance is worked out per working day.

## A simple method for your own count

1. **Count the calendar days** between the two dates, and decide if the end date counts.
2. **Choose the weekly off** that applies to you: two days, one day or none.
3. **Run the calculator** for the working days.
4. **List the public holidays** in the range and remove the ones on working days.
5. **Remove your own leave** to get the days you can actually work.

## Frequently asked questions

### How do I count working days between two dates?

Count all the days in the range, then remove the weekly off days. With Saturday and Sunday off, 1 to 30 November 2026 has 30 days and 21 working days.

### Does the calculator exclude public holidays?

No. It removes only the weekly off. Subtract the public holidays that fall on a working day yourself.

### Is the start date counted?

Yes. The count starts on the start date. You can choose whether the end date is included.

### What if my week has only Sunday off?

Choose Sunday only as the weekly off. The same November 2026 range then has 25 working days.

### How many working days are in a month?

Most months with a Saturday and Sunday off have 20 to 23 working days. It depends on the month's length and the day of the week it starts on.

### How is this different from counting calendar days?

Calendar days count every date. For a plain count of days between two dates, use the [Days Until Date Calculator](https://toolcloak.pages.dev/tools/days-until-date-calculator) or the [Date Difference Calculator](https://toolcloak.pages.dev/tools/date-difference-calculator).



---

# Z-score explained: compare anything on one scale and spot outliers

> A score of 78 in maths and 70 in science: which is the better result? A z-score answers that by counting how many standard deviations a value sits from the average. This guide covers the formula, the percentile table, the 68-95-99.7 rule, and how rare 3, 4 and 5 sigma events really are.

Published 2026-10-09 by Toolcloak. 4 minute read. Original: https://toolcloak.pages.dev/articles/z-score-explained-compare-anything-on-one-scale-and-spot-outliers
Related free tool: [Z-Score Calculator](https://toolcloak.pages.dev/tools/z-score-calculator)

Raw numbers hide their context. Is 78 good? It depends on how the rest of the class did. A **z-score** (also called a standard score) removes the units and the scale: it says how many standard deviations a value lies above or below the mean. Once everything is in z-scores, you can compare marks, heights, speeds or prices on one common ruler.

> **In short**
> - **z = (value − mean) ÷ standard deviation.** Positive means above average, negative means below, and 0 is exactly average.
> - In a roughly bell-shaped set, about **68%** of values fall within z = ±1, **95%** within ±2 and **99.7%** within ±3.
> - A z of 2 is about the **97.7th percentile**. A z of −1 is about the 16th.
> - Z-scores let you compare different tests fairly, and flag outliers beyond ±3. They need a sensible mean and spread, so they work best on bell-shaped data.

## The formula in plain words

Take the value, subtract the average, and divide by the typical spread. The subtraction tells you how far from average you are. The division tells you how big that distance is compared with normal variation.

**Example: a student scores 78, class mean 62, standard deviation 8**

1. **Distance from the mean.** 78 − 62 = **16** marks above average.
2. **Divide by the standard deviation.** 16 ÷ 8 = **2.0**.
3. **Read the result.** The score is 2 standard deviations above the mean, around the 97.7th percentile if marks are bell-shaped.

*Going the other way is just as easy: value = mean + z × standard deviation. A z of −1.5 here would be 62 − 1.5 × 8 = 50.*

Try it: [Z-Score Calculator](https://toolcloak.pages.dev/tools/z-score-calculator). Enter a value, the mean and the standard deviation to get the z-score and its percentile.

## The z-to-percentile table

For bell-shaped data, each z-score maps to a share of the group that scores lower. The mapping comes from the normal distribution, so a calculator or table gives it instantly.

| z-score | Percentile (below) | Share within ±z of the mean |
| --- | --- | --- |
| 0 | 50.00% | 0% |
| 0.5 | 69.15% | 38.29% |
| 1.0 | 84.13% | 68.27% |
| 1.5 | 93.32% | 86.64% |
| 1.96 | 97.50% | 95.00% |
| 2.0 | 97.72% | 95.45% |
| 2.5 | 99.38% | 98.76% |
| 3.0 | 99.87% | 99.73% |

*The row for 1.96 is the famous one: it marks off the middle 95%, which is why 1.96 appears in so many confidence intervals.*

**How to read a z-score**

| Zone | From | To |
| --- | --- | --- |
| Very low | -4 | -2 |
| Below avg | -2 | 0 |
| Above avg | 0 | 2 |
| Very high | 2 | 4 |

*Beyond ±2 you are in the outer 5% of a bell curve. Beyond ±3 you are in the outer 0.3%.*

## The 68-95-99.7 rule

Because the bell curve is symmetrical, the share of values in each band is fixed. This is the quick mental version of the table above.

**Share of values in each z-score band (%)**

|  | Value |
| --- | --- |
| −3 to −2 | 2.14% |
| −2 to −1 | 13.59% |
| −1 to 0 | 34.13% |
| 0 to 1 | 34.13% |
| 1 to 2 | 13.59% |
| 2 to 3 | 2.14% |

*The extra 0.13% beyond −3 and 0.13% beyond +3 are not shown. Adding all eight pieces gives 100%.*

**Inside versus outside**

|  | Value | Note |
| --- | --- | --- |
| Within ±1 SD | 68.3% | About 2 in 3 |
| Within ±2 SD | 95.4% | About 19 in 20 |
| Within ±3 SD | 99.7% | About 997 in 1,000 |
| Beyond ±3 SD | 0.27% | About 1 in 370 |

*These are for an exactly normal distribution. Real data is only roughly bell-shaped, so treat them as a guide.*

## Comparing two different tests fairly

This is the classic use. Maria scored 78 in maths and 70 in science. The maths mean was 65 with a standard deviation of 8. The science mean was 55 with a standard deviation of 6.

| Subject | Score | Mean | SD | z-score | Percentile |
| --- | --- | --- | --- | --- | --- |
| Maths | 78 | 65 | 8 | 1.63 | 94.8% |
| Science | 70 | 55 | 6 | 2.50 | 99.4% |

*The science mark is lower in raw terms, but much further above its class average.*

**Maria's two subjects, by z-score**

|  | z-score |
| --- | --- |
| Maths (78) | 1.63 |
| Science (70) | 2.50 |

*Her science result is the stronger one, even though 78 looks bigger than 70.*

## How rare is rare?

Outside a given number of standard deviations, the share of a normal distribution gets tiny very quickly. This table counts both tails together.

| Distance from mean | Share outside | About 1 in |
| --- | --- | --- |
| ±1 SD | 31.73% | 3 |
| ±2 SD | 4.55% | 22 |
| ±3 SD | 0.27% | 370 |
| ±4 SD | 0.0063% | 15,787 |
| ±5 SD | 0.000057% | 1.74 million |

*Physicists quote 'five sigma' for discoveries, but they use one tail only, which is about 1 in 3.5 million.*

**Out of 1,000,000 values in a normal distribution**

| Step | Value |
| --- | --- |
| Within ±1 SD | 682,689 |
| Within ±2 SD | 954,500 |
| Within ±3 SD | 997,300 |
| Within ±4 SD | 999,937 |

*Each wider band adds fewer and fewer extra values. That is why a 4 sigma result is a strong signal that something unusual happened.*

## Finding outliers with z-scores

A common rule flags values with z above 3 or below −3. Some analysts use 2.5, or 2 for small sets. It is a screen, not a verdict: an outlier might be a typing mistake, a faulty sensor, or a genuine rare event worth keeping.

1. **Compute the mean and standard deviation** of the whole set.
2. **Calculate a z-score for each value.**
3. **Flag the ones beyond your cut-off** (3 is common).
4. **Look at each flagged value.** Check the source before deleting anything.
5. **Recalculate** if you remove a value, because the mean and standard deviation will change.

> **One big outlier can hide itself:** An extreme value pulls the mean towards it and inflates the standard deviation, which shrinks its own z-score. In small samples the biggest possible z-score is limited, so a wild value may not even reach 3. Check a chart as well as the numbers.

## Z-scores for sample means

When you test an average rather than a single value, divide by the **standard error**, σ ÷ √n, instead of σ. Example: a population has mean 70 and standard deviation 10. A sample of 36 people averages 73. The standard error is 10 ÷ 6 = 1.67, so z = (73 − 70) ÷ 1.67 = **1.8**.

**Same data, two different questions**

| One person scores 73 | A group of 36 averages 73 |
| --- | --- |
| z = 0.3 | z = 1.8 |
| Divide by the standard deviation, 10 | Divide by the standard error, 1.67 |
| (73 − 70) ÷ 10 = 0.3 | (73 − 70) ÷ 1.67 = 1.8 |
| Perfectly ordinary | Only about 3.6% of group averages would be this high |

*Averages vary far less than individuals, so the same gap counts for much more.*

## When z-scores mislead

- **Skewed data.** Incomes, house prices and website visits have long right tails. Percentiles from the normal table will be off.
- **Tiny samples.** With fewer than about 30 values, the standard deviation itself is uncertain. A t-score is more honest.
- **Mixed groups.** If you blend two different populations, the single mean and spread describe neither.
- **Ordinal scales.** Ratings from 1 to 5 are not truly continuous, so z-scores are only a rough guide.

## Frequently asked questions

### What is a z-score?

A z-score is the number of standard deviations a value lies from the mean. It is calculated as (value − mean) ÷ standard deviation.

### What is a good z-score?

It depends on the context. A z-score above 0 is above average. Above 2 puts a value in roughly the top 2.3% of a normal distribution, and below −2 in the bottom 2.3%.

### Can a z-score be negative?

Yes. A negative z-score means the value is below the mean. For example, z = −1 is one standard deviation below average.

### How do I convert a z-score to a percentile?

Use a standard normal table or calculator. For instance, z = 1 is about the 84th percentile, z = 1.96 is the 97.5th and z = 3 is the 99.87th.

### What z-score counts as an outlier?

Many analysts flag values beyond ±3. Others use ±2.5 or ±2, especially for small data sets. Always inspect flagged values before removing them.

### Is a z-score the same as a t-score?

No. A z-score uses the known population standard deviation. A t-score uses the sample's own estimate and is wider for small samples.



---

# 16:9, 4:3, 1:1: aspect ratios without cropping surprises

> You resize a photo and the face goes oval. You upload a video and it gets black bars. Both come from one idea, the shape of a frame, and a bit of arithmetic that takes the guesswork out.

Published 2026-10-08 by Toolcloak. 3 minute read. Original: https://toolcloak.pages.dev/articles/16-9-4-3-1-1-aspect-ratios-without-cropping-surprises
Related free tool: [Aspect Ratio Calculator](https://toolcloak.pages.dev/tools/aspect-ratio-calculator)

A photo is 4000 wide and 3000 tall. You need it 1200 wide for a website. What height keeps it undistorted? If you guessed 900, you already understand aspect ratios. If you typed 1000 and the people turned thin, this article is for you.

> **In short**
> - An aspect ratio is a shape, not a size: 1920 × 1080 and 1280 × 720 are both 16:9.
> - New height = new width × (old height ÷ old width). 4000 × 3000 at 1200 wide is 900 tall.
> - Crop, pad or stretch. Stretching distorts, and cropping 16:9 to 9:16 keeps only about 32% of the width.

An aspect ratio is only the **shape** of a frame: how wide it is compared with how tall. It says nothing about the number of pixels. A phone screenshot and a poster can share a ratio even though one is tiny and one is enormous.

## What 16:9 actually means

A ratio of 16:9 means that for every 16 units of width there are 9 units of height. Shrink or grow the frame, and as long as that relationship holds, the shape is the same. 1920 × 1080 is 16:9. So is 1280 × 720, and so is 3840 × 2160, the usual “4K” screen.

**Common shapes, drawn to scale**

| Shape | Used for |
| --- | --- |
| 16:9 | TV, YouTube, most screens |
| 4:3 | Older TVs, tablets, slides |
| 3:2 | Many cameras and prints |
| 1:1 | Square posts, avatars |
| 4:5 | Portrait feed posts |
| 9:16 | Stories, Reels, Shorts |

*Every frame is fitted into the same square so you can compare the shapes. The numbers say nothing about size, only about proportion.*

## The one calculation you need

> **new height = new width × (old height ÷ old width):** For 4000 × 3000 shrunk to 1200 wide: 1200 × (3000 ÷ 4000) = 900. Going the other way, new width = new height × (old width ÷ old height). Do it once and round to the nearest whole pixel.

Try it: [Aspect Ratio Calculator](https://toolcloak.pages.dev/tools/aspect-ratio-calculator). Enter a width and height, then change either one. The other updates so the shape stays the same.

This is what the little padlock between the width and height boxes does in most editors. With it locked, the program does exactly this sum for you. With it unlocked, you decide both numbers, and the picture stretches to fit.

## Three things you can do when shapes do not match

Sooner or later you need a picture to fit a frame of a different shape. There are only three honest options.

1. **Crop.** Cut away the parts that do not fit. The picture stays undistorted, but you lose content.
2. **Pad.** Keep the whole picture and fill the leftover space with bars or a background. Nothing is lost, but the picture looks smaller.
3. **Stretch.** Squash or pull the picture to fill the frame. It fits perfectly, and everything looks wrong. Almost never the right choice.

**Putting a wide picture into a tall frame**

| Stretching | Cropping or padding |
| --- | --- |
| Distorted | Honest |
| Circles turn into ovals | Shapes stay correct |
| Faces look thin | You choose what to keep |
| People notice immediately | Crop loses edges, padding adds bars |

*If you have to choose, crop for impact and pad when the whole picture matters.*

## How much you lose when you crop

Turning a wide picture into a tall one is the harshest crop. Take a 1920 × 1080 video frame, which is 16:9, and cut it to 9:16 for a vertical format. The new frame must be 1080 tall and only about 608 wide. Almost seven tenths of the picture disappears.

**How much of a 16:9 frame survives each crop (full height kept)**

|  | Kept |
| --- | --- |
| To 16:9 | 100% |
| To 4:3 | 75% |
| To 1:1 | 56% |
| To 4:5 | 45% |
| To 9:16 | 32% |

*Calculated from the width that remains when the height is kept: 1440, 1080, 864 and 608 pixels out of 1920.*

Cameras have the same problem. A 6000 × 4000 photo is 3:2. To make it 16:9 you keep the full width and cut the height to 3375, losing 625 rows, about 16 percent. A 4:3 phone photo at 4000 × 3000 becomes 4000 × 2250 in 16:9, losing a quarter of the frame.

> **Plan the crop when you take the picture:** If you know the photo will end up in a tall or square frame, leave breathing space around the subject. It is much easier to crop in than to invent pixels that were never captured.

## Sizes people look for most

| Where it is used | Ratio | Common size in pixels |
| --- | --- | --- |
| Full HD video, YouTube | 16:9 | 1920 × 1080 |
| HD video | 16:9 | 1280 × 720 |
| Square post | 1:1 | 1080 × 1080 |
| Portrait feed post | 4:5 | 1080 × 1350 |
| Stories, Reels, Shorts | 9:16 | 1080 × 1920 |
| Classic monitor or slide | 4:3 | 1024 × 768 |

*These are the commonly recommended sizes. Platforms change their rules from time to time, so check the current guidance before you export something important.*

## Why 4K is not 4000

The shape matters more than the number of pixels. A 3840 × 2160 screen has four times the pixels of a 1920 × 1080 one, yet exactly the same 16:9 proportions. Doubling both sides quadruples the pixel count. This is why scaling a picture up by 2× makes the file roughly four times larger.

## Frequently asked questions

### What is an aspect ratio?

The proportion between a frame’s width and its height, written like 16:9. It describes shape, not size.

### How do I resize an image without distorting it?

Keep the aspect ratio locked. New height = new width × (old height ÷ old width). For 4000 × 3000 at 1200 wide that gives 900.

### What is the aspect ratio of 1920 × 1080?

16:9. Divide both numbers by 120 to get 16 and 9.

### How do I crop a 16:9 video to 9:16?

Keep the full height and cut the width to height × 9 ÷ 16. For 1080 tall that is about 608 pixels wide, so you keep roughly 32 percent of the original width.

### Is cropping or padding better?

Crop when the edges are not important and you want the subject to fill the frame. Pad when every part of the picture matters, such as a logo or a chart.



---

# What your BMI can tell you, and what it cannot

> It is one division and one number, used by doctors for over fifty years. It is also easy to misread. Here is where BMI is a handy first glance and where it quietly fails.

Published 2026-10-08 by Toolcloak. 3 minute read. Original: https://toolcloak.pages.dev/articles/what-your-bmi-can-tell-you-and-what-it-cannot
Related free tool: [BMI Calculator](https://toolcloak.pages.dev/tools/bmi-calculator)

Someone is 1.70 metres tall and weighs 70 kilograms. Their BMI is **24.2**. Is that good? The honest answer is: it is a starting point for a conversation, not a verdict on anyone’s health, and the bands attached to it are rougher than most charts make them look.

> **In short**
> - BMI = weight in kg ÷ height in metres squared. It is a screening number, not a diagnosis.
> - Many Asian guidelines flag risk from 23, lower than the WHO standard of 25.
> - Muscle, age and where weight sits all fool it. Pair it with a waist measurement.

BMI stands for body mass index. It is a screening tool: cheap, quick and useful across large groups of people. Used on one individual it can mislead. This article is general information, not medical advice. For anything about your own health, a doctor looking at the whole picture is worth far more than any single number.

## The formula is one line

> **BMI = weight in kg ÷ (height in metres)²:** For 70 kg and 1.70 m: 1.70 × 1.70 = 2.89, and 70 ÷ 2.89 = 24.2. For the same 70 kg at 1.80 m the height squared is 3.24, so BMI is 21.6. Height is squared because the weight of a body tends to grow faster than its height.

Try it: [BMI Calculator](https://toolcloak.pages.dev/tools/bmi-calculator). Enter height and weight in metric or imperial. It shows your BMI and which band it falls in.

## The usual bands

The World Health Organization’s standard categories are the ones used on most charts and apps. They were set from studies of populations in which these ranges tracked health risk reasonably well on average.

**WHO adult BMI bands**

| Zone | From | To |
| --- | --- | --- |
| Under | 15 | 18.5 |
| Normal | 18.5 | 25 |
| Over | 25 | 30 |
| Obese | 30 | 40 |

*Under 18.5 is underweight, 18.5 to 24.9 normal, 25 to 29.9 overweight, 30 and above obese. The marker is the 70 kg, 1.70 m example.*

Notice how arbitrary the edges are. A BMI of 24.9 is “normal” and 25.0 is “overweight”, but nobody’s body changes at that line. These are bands drawn on a smooth curve.

## For many people of South Asian origin, the lines move

Research found that at the same BMI, many Asian populations tend to carry more body fat and a higher risk of diabetes and heart conditions than the standard bands imply. Because of that, many Asian and Indian guidelines use **lower cut-offs**: commonly 23 for overweight and 27.5 (or 25 in some Indian consensus guidance) for obesity. Which set your doctor uses can differ, so ask rather than assume.

**Same BMI, different reading depending on the guideline**

| WHO standard | Many Asian guidelines |
| --- | --- |
| BMI 24 | BMI 24 |
| Falls in the normal band | Already above the 23 line |
| Overweight starts at 25 | Gets flagged as raised risk earlier |

*Neither label is a diagnosis. The point is that the line depends on who it was drawn for, which is worth knowing before a chart worries or reassures you.*

## Where BMI quietly gets it wrong

- **It cannot tell muscle from fat.** A strength athlete can land in the “overweight” band while being lean, because muscle is dense and weighs a lot.
- **It ignores where the weight sits.** Fat around the middle is linked to more risk than the same amount on hips and thighs, and BMI cannot see the difference.
- **It changes with age.** Older adults lose muscle, so the same BMI can hide more fat. Children and teenagers use age-specific charts instead of these adult bands.
- **Pregnancy and some conditions** change weight in ways BMI is not designed for.
- **Height rounding matters.** Being off by two centimetres can move you a full point on the scale.

## A tape measure fills in the gaps

Many clinicians now pair BMI with waist size, or with **waist-to-height ratio**: your waist divided by your height, in the same units. A commonly used rule of thumb is to keep your waist under half your height, which for 170 cm means under 85 cm. It is not a substitute for medical advice, but it captures what BMI misses: where the weight is carried.

**Quick checks worth knowing**

|  | Value | Note |
| --- | --- | --- |
| The BMI formula | kg ÷ m² | Weight over height squared |
| Waist-to-height rule of thumb | 0.5 | Waist under half of height |
| Time to measure both | 5 min | Morning, same time each week |

*Trends matter more than a single reading. Measure the same way, at the same time of day, and look at the direction over weeks.*

## How to use the number sensibly

1. Treat it as a **first glance**, like a thermometer reading rather than a diagnosis.
2. If you are well outside the normal band, or the number changed a lot without reason, mention it to a doctor who can look at blood tests, family history and the rest.
3. If you train regularly and feel strong and well, a high BMI on its own is not a reason to panic. Waist size and fitness tell you more.
4. Do not chase a number. If a chart is making you anxious about food or your body, talk to a professional rather than adjusting on your own.

## Frequently asked questions

### How is BMI calculated?

Divide your weight in kilograms by your height in metres squared. 70 kg at 1.70 m gives 70 ÷ 2.89, which is 24.2.

### What is a normal BMI?

The WHO standard normal range for adults is 18.5 to 24.9. Many Asian guidelines flag raised risk from 23, so the cut-offs depend on the guideline your doctor uses.

### Is BMI accurate for athletes?

Often not. Muscle is dense, so a muscular person can have a high BMI without carrying much fat. BMI cannot separate muscle from fat.

### Can children use the adult BMI chart?

No. Children and teenagers are compared against age and sex specific charts, because their bodies change as they grow.

### What is better than BMI?

Nothing replaces a clinical check, but waist size or waist-to-height ratio add useful information about where weight is carried, and blood tests show what is happening inside.



---

# BMR and TDEE: how many calories your body burns before you lift a finger

> Most of the energy you burn each day goes on staying alive, not on workouts. This guide explains BMR, activity multipliers and TDEE with a worked example, shows how the common formulas compare, and what a calorie target can and cannot tell you.

Published 2026-10-08 by Toolcloak. 4 minute read. Original: https://toolcloak.pages.dev/articles/bmr-and-tdee-how-many-calories-your-body-burns-before-you-lift-a-finger
Related free tool: [BMR & Calorie Calculator](https://toolcloak.pages.dev/tools/bmr-calorie-calculator)

Even if you spent the entire day in bed, your body would burn roughly a thousand to two thousand calories keeping your heart beating, lungs working and temperature steady. That resting cost is your **basal metabolic rate (BMR)**. Add movement and digestion and you get your **total daily energy expenditure (TDEE)**, the figure most calorie plans are built on.

> **In short**
> - **BMR** is the energy needed at complete rest. **TDEE = BMR × an activity factor** (1.2 for a desk job up to about 1.9 for very hard training).
> - The Mifflin-St Jeor formula for a 30-year-old man of 75 kg and 175 cm gives a BMR of **1,699 kcal**, and about **2,633 kcal** with moderate exercise.
> - Every formula is an **estimate, often within ±10%**. Treat it as a starting point and adjust from real results over 2 to 3 weeks.
> - Roughly **60–70% of daily energy** goes on BMR, about 10% on digesting food and the rest on movement.

## Where your energy actually goes

People often assume exercise is the main part of their daily burn. For most people it is not. Staying alive costs the most, followed by day-to-day movement such as walking, standing and fidgeting, then deliberate exercise, and finally the cost of digesting food itself.

**A typical split of daily energy use**

|  | Share |
| --- | --- |
| Basal metabolism (BMR) | ≈ 60–70% |
| Daily movement and exercise | ≈ 20–30% |
| Digesting food | ≈ 10% |

*These shares are typical ranges, not exact. A very active person has a bigger movement slice; an older or smaller person has a smaller BMR slice.*

Try it: [BMR & Calorie Calculator](https://toolcloak.pages.dev/tools/bmr-calorie-calculator). Enter age, height, weight and activity level and get BMR, TDEE and example targets.

## The formula, with real numbers

The most widely used equation today is Mifflin-St Jeor (1990). It uses weight in kilograms, height in centimetres and age in years.

> **Mifflin-St Jeor:** Men: BMR = 10 × weight + 6.25 × height − 5 × age + 5. Women: BMR = 10 × weight + 6.25 × height − 5 × age − 161. The only difference between the two is the constant at the end: +5 versus −161.

**Building the BMR of a 30-year-old man (75 kg, 175 cm)**

| Step | Amount |
| --- | --- |
| Weight term (10 × 75) | 750 |
| Height term (6.25 × 175) | +1,094 |
| Age term (5 × 30) | −150 |
| Constant | +5 |
| BMR | 1,699 |

*750 + 1,093.75 − 150 + 5 = 1,699 kcal a day at rest. For a 30-year-old woman of 62 kg and 162 cm the same steps give about 1,322 kcal.*

**Two formulas, same man**

|  | kcal per day |
| --- | --- |
| Mifflin-St Jeor | 1,699 |
| Harris-Benedict (revised) | 1,763 |

*The older Harris-Benedict equation (revised in 1984) gives about 64 kcal more, a gap of roughly 4%. Real people differ from both by a similar margin.*

## From BMR to TDEE

To include daily movement, multiply BMR by an activity factor. The factors are rough averages, and most people choose a level that is too high.

**TDEE for the same man at each activity level (kcal)**

|  | TDEE |
| --- | --- |
| Desk job, little exercise (×1.2) | 2,039 |
| Light exercise 1–3 days (×1.375) | 2,336 |
| Moderate exercise 3–5 days (×1.55) | 2,633 |
| Hard exercise 6–7 days (×1.725) | 2,930 |
| Physical job plus training (×1.9) | 3,228 |

*Same body, different lives: the difference between a desk job and a physical job with training is about 1,200 kcal a day.*

> **Pick the lower level when unsure:** Workouts last an hour, but the other 23 hours matter just as much. If you sit most of the day, “light activity” is usually more honest than “moderate”. Choose the lower level and check against real results.

## Turning TDEE into a plan

Eating about as much as your TDEE keeps weight roughly stable. Eating more leads to gradual gain; eating less to gradual loss. A commonly quoted rule of thumb is that about 7,700 kcal is one kilogram of body weight, so a daily gap of 500 kcal is a bit under half a kilogram a week. Real bodies adapt, so the figure is a guide, not a promise.

**A moderate gap relative to maintenance (the 75 kg man at TDEE ≈ 2,633)**

| Zone | From | To |
| --- | --- | --- |
| Large cut | 70 | 80 |
| Moderate cut | 80 | 95 |
| Maintain | 95 | 105 |
| Gain | 105 | 130 |

*Values are the percentage of TDEE eaten. A cut of 10 to 20% is generally considered moderate. Very large cuts are harder to keep up and are best planned with a doctor or dietitian.*

**A sensible way to use the estimate**

1. **Estimate TDEE.** Use the calculator with an honest activity level.
2. **Track for 2–3 weeks.** Eat roughly that amount and weigh yourself at the same time each week.
3. **Adjust by 5–10%.** If weight moved the wrong way, change intake slightly and repeat. The trend is the real answer.

> **An estimate is not a prescription:** These formulas ignore muscle mass, illness, medication, pregnancy and many other things. They are unsuitable for children and for people with medical conditions or a history of disordered eating. If any of those apply, plan with a doctor or registered dietitian rather than a calculator.

## Why two people of the same weight can differ

**What moves your BMR up or down**

| Box | What it means |
| --- | --- |
| Muscle mass | Muscle burns more at rest than fat. More lean mass raises BMR. |
| Age | The formula subtracts about 5 kcal for each year, partly from lost muscle. |
| Body size | Bigger bodies burn more simply because they have more tissue to run. |
| Sleep and health | Illness and fever raise it; long-term severe under-eating can lower it. |

*This is why a formula can be 10% off for a particular person even when the inputs are right.*

## Quick practice

1. BMR for a 40-year-old man, 80 kg, 180 cm? 800 + 1,125 − 200 + 5 = **1,730 kcal**.
2. TDEE with light exercise (×1.375)? 1,730 × 1.375 ≈ **2,379 kcal**.
3. If someone eats 2,000 kcal against a TDEE of 2,400, what is the daily gap? **400 kcal**, or about 17%.
4. Why can two people with the same BMR gain and lose weight differently? **Different movement, digestion, sleep and measurement error.**

Pair the calorie estimate with other numbers: [BMI](https://toolcloak.pages.dev/articles/what-your-bmi-can-tell-you-and-what-it-cannot) for a rough weight check, [heart rate zones](https://toolcloak.pages.dev/articles/heart-rate-zones-explained-what-each-of-the-five-zones-is-for) for training, and [sleep cycles](https://toolcloak.pages.dev/articles/sleep-in-90-minute-cycles-when-to-go-to-bed) for recovery.

## Frequently asked questions

### What is BMR?

Basal metabolic rate is the energy your body needs at complete rest to keep you alive, mostly breathing, circulation, temperature control and cell repair.

### What is the difference between BMR and TDEE?

BMR is energy at rest. TDEE adds daily movement, exercise and digesting food, so it is always higher.

### Which BMR formula is the best?

Mifflin-St Jeor is widely recommended for the general population. The Katch-McArdle formula can be better when body fat is known. None is accurate for everyone.

### How many calories should I eat to lose weight?

A moderate gap of 10 to 20% below your TDEE is common. The best number depends on health, age and goals, so check with a professional for anything larger.

### Does muscle really raise BMR?

Yes, but modestly. A kilogram of muscle burns only a few more calories at rest than a kilogram of fat, so the effect is real but small.

### Are calorie calculators accurate?

They are estimates, usually within about 10%. Use the result as a starting point and adjust from how your weight changes over a few weeks.



---

# Break-even point: how many sales before you make a profit

> Break-even is the number of units you must sell before income covers all your costs. With ₹60,000 of fixed costs, a ₹500 price and ₹300 of variable cost per unit, you break even at exactly 300 units, which is ₹1,50,000 of sales. This guide shows the formula, the graph and five ways to move the point.

Published 2026-10-08 by Toolcloak. 3 minute read. Original: https://toolcloak.pages.dev/articles/break-even-point-how-many-sales-before-you-make-a-profit
Related free tool: [Break-Even Calculator](https://toolcloak.pages.dev/tools/break-even-calculator)

Every business starts the month in the red. Rent, salaries and software are due whether you sell one item or a thousand. Break-even tells you exactly how far into the month the red turns to black, and it needs only three numbers.

> **In short**
> - **Break-even units = fixed costs ÷ (price − variable cost per unit).** The bracket is called the contribution per unit.
> - Example: ₹60,000 ÷ (₹500 − ₹300) = **300 units**, or **₹1,50,000** of sales.
> - Below 300 units you lose money. Above it, each extra unit adds **₹200** of profit.
> - A small price change moves break-even a lot: **₹550 instead of ₹500 cuts it from 300 to 240 units**.

## Two kinds of cost

**Fixed costs** do not change with sales in the short run: rent, salaries, loan interest, subscriptions. **Variable costs** rise with every unit: raw material, packaging, delivery, payment fees. Break-even depends on keeping the two lists apart.

**Our example business, per month**

|  | Value | Note |
| --- | --- | --- |
| Fixed costs | ₹60,000 | Rent, wages, tools |
| Price per unit | ₹500 |  |
| Variable cost per unit | ₹300 | Materials, packing |

## What each sale contributes

Each unit you sell brings in ₹500, but ₹300 of it is used up making and delivering that unit. The ₹200 left is the **contribution**. It goes first towards the fixed costs and, once those are paid, becomes profit.

**One unit, from price to contribution (₹)**

| Step | Amount |
| --- | --- |
| Price | ₹500 |
| Variable cost | −₹300 |
| Contribution | ₹200 |

*The contribution margin is ₹200 ÷ ₹500 = 40%. Forty paise of every sales rupee is available to pay fixed costs.*

> **Break-even units = fixed costs ÷ contribution per unit:** ₹60,000 ÷ ₹200 = 300 units. In rupees: ₹60,000 ÷ 0.40 = ₹1,50,000 of sales. Both give the same answer, one counted in units and the other in sales.

**Revenue against total cost as units grow**

| Units sold | Revenue | Total cost | Fixed cost only |
| --- | --- | --- | --- |
| 0 | 0 | 60000 | 60000 |
| 100 | 50000 | 90000 | 60000 |
| 200 | 100000 | 120000 | 60000 |
| 300 | 150000 | 150000 | 60000 |
| 400 | 200000 | 180000 | 60000 |
| 500 | 250000 | 210000 | 60000 |
| 600 | 300000 | 240000 | 60000 |

*The two main lines cross at 300 units, the break-even point. To the left the cost line is higher, so you lose. To the right the revenue line is higher, so you profit.*

Try it: [Break-Even Calculator](https://toolcloak.pages.dev/tools/break-even-calculator). Enter your fixed cost, price and variable cost to find the break-even units and sales.

## Profit and loss at different volumes

**Contribution against fixed costs at different volumes**

|  | Contribution earned (units × ₹200) | Still short of ₹60,000 |
| --- | --- | --- |
| 200 units | ₹40k | ₹20k short |
| 300 units | ₹60k covers it |
| 450 units | ₹60k fixed costs | ₹30k profit |

*At 200 units the contribution is ₹40,000, so ₹20,000 of the fixed costs is left unpaid. At 450 units the first ₹60,000 pays the fixed costs and the next ₹30,000 is profit.*

## How far can sales fall? The margin of safety

If you expect to sell 450 units, your break-even is 300, so you are 150 units above it. Divided by 450, that is a margin of safety of 33%. Sales could fall by a third before you start losing money.

**Margin of safety at 450 units**

| Zone | Range |
| --- | --- |
| Fragile | under 10% |
| Workable | 10–25% |
| Comfortable | over 25% |

*The bands are a rough rule of thumb, not an official standard. Seasonal businesses usually want more cushion.*

## Five levers that move the break-even point

**Break-even units under different prices (fixed cost ₹60,000, variable cost ₹300)**

|  | Units |
| --- | --- |
| Price ₹450 | 400 units |
| Price ₹500 | 300 units |
| Price ₹550 | 240 units |
| Price ₹600 | 200 units |

*A 10% higher price (₹550) cuts the target by 20%. But a higher price may also cut how many units you sell, so test it carefully.*

1. **Raise the price.** Every rupee goes straight into the contribution. See [markup vs margin](https://toolcloak.pages.dev/articles/markup-vs-margin-why-a-50-percent-markup-is-only-a-33-percent-margin) before you pick a number.
2. **Cut variable costs.** Cheaper packaging, better supplier terms or fewer returns raise the contribution per unit.
3. **Cut fixed costs.** A ₹20,000 lower rent moves break-even from 300 units to 200 units.
4. **Change the mix.** If you sell several products, push the ones with the highest contribution.
5. **Avoid deep discounts.** A 10% discount on a ₹500 item cuts contribution from ₹200 to ₹150 and pushes break-even from 300 to 400 units. The [stacked discount guide](https://toolcloak.pages.dev/articles/50-percent-off-plus-20-percent-off-is-not-70-percent-off) shows how fast it adds up.

**Break-even units for different fixed costs and prices (variable cost ₹300)**

| Fixed cost | ₹450 | ₹500 | ₹550 |
| --- | --- | --- | --- |
| ₹40,000 | 267 | 200 | 160 |
| ₹60,000 | 400 | 300 | 240 |
| ₹80,000 | 533 | 400 | 320 |

## Setting a profit target

To earn ₹30,000 a month, add it to the fixed costs: (₹60,000 + ₹30,000) ÷ ₹200 = **450 units**. The same formula works for any target.

## Limits of the simple model

- It assumes the price and the variable cost stay the same at every volume. Bulk discounts and overtime break that.
- Some costs are in between, such as a salary that steps up when you hire. Treat them as fixed for each range of volume.
- It ignores timing. You may need cash for stock long before the sales come in, so a business that breaks even on paper can still run short.
- Taxes and loan repayments are not in the simple formula. Add loan interest to fixed costs and plan repayments separately.

> **Use it before you spend:** Before signing a lease or hiring, add the new cost to the fixed costs and recompute. If the new break-even is higher than you have ever sold in a month, wait or find a smaller step.

## Frequently asked questions

### What is the break-even formula?

Break-even units = fixed costs ÷ (selling price per unit − variable cost per unit). Break-even sales = fixed costs ÷ contribution margin ratio.

### What is contribution margin?

It is the price minus the variable cost per unit. It is the amount each unit contributes towards fixed costs and then profit.

### Is rent a fixed or variable cost?

Rent is normally a fixed cost because it does not change with the number of units you sell.

### How can I lower my break-even point?

Raise the price, cut variable costs, cut fixed costs, or shift sales to higher-contribution products.

### What does margin of safety mean?

It is how far sales can fall below your expected level before you reach break-even. (Expected sales − break-even sales) ÷ expected sales.



---

# Caesar cipher: why only 25 keys means anyone can crack it

> The Caesar cipher shifts every letter by a fixed number. With a shift of 3, HELLO becomes KHOOR. It is fun and historic, but there are only 25 useful shifts, so a computer (or a patient person) can try them all in seconds. Here is how it works and why it fails.

Published 2026-10-08 by Toolcloak. 3 minute read. Original: https://toolcloak.pages.dev/articles/caesar-cipher-why-only-25-keys-means-anyone-can-crack-it
Related free tool: [Caesar Cipher & ROT13](https://toolcloak.pages.dev/tools/caesar-cipher)

Julius Caesar is said to have used a simple trick to hide military messages: shift every letter a few places along the alphabet. The method is easy to learn and a good first step into cryptography. It is also easy to break, and seeing why teaches the key idea behind every secure system.

> **In short**
> - A Caesar cipher **shifts each letter by a fixed number**. Shift 3: A→D, B→E, **HELLO→KHOOR**.
> - There are only **25 useful shifts**, so trying them all (brute force) is instant.
> - A general substitution cipher has about **4.03 × 10²⁶** keys, but it still falls to **letter-frequency analysis**.
> - None of these is secure. Use it for puzzles, not for secrets.

## How it works

Write the alphabet twice, one row shifted. To encrypt, find each letter in the top row and replace it with the letter below it. To decrypt, go back the same number of places. The shift amount is the key.

| Plain | H | E | L | L | O |
| --- | --- | --- | --- | --- | --- |
| Shift 3 | K | H | O | O | R |
| Shift 13 (ROT13) | U | R | Y | Y | B |

*Shift 13 is special: applying it twice returns the original, because 13 + 13 = 26 letters.*

> **Encrypt: (letter + key) mod 26     Decrypt: (letter − key) mod 26:** Number the letters A=0 to Z=25. “mod 26” means wrap around after Z. H is 7, add 3 gives 10, which is K.

Try it: [Caesar Cipher & ROT13](https://toolcloak.pages.dev/tools/caesar-cipher). Type a message and a shift to encode or decode it.

## Why it is so weak

A shift of 0 does nothing and a shift of 26 returns to the start, so only shifts 1 to 25 change the text. An attacker does not need to guess the key. They try all 25 and read the one that makes sense.

**The whole key space: 25 useful shifts**

| Group | Count |
| --- | --- |
| Shifts to try | 25 |

**Breaking a Caesar cipher by brute force**

1. **Take the ciphertext.** For example KHOOR.
2. **Try shift 1 to 25.** Decode with each one. It takes seconds by hand and microseconds by computer.
3. **Read the results.** Only one will be a real word. Shift 3 gives HELLO.

## A bigger key space still is not enough

What if each letter could be swapped for any other, not just by a fixed shift? That is a general substitution cipher, with 26! (26 factorial) possible keys, about 4.03 × 10²⁶. Brute force is hopeless. But the cipher still leaks, because the language does.

**How often each letter appears in typical English text (%)**

|  | Value |
| --- | --- |
| E | 12.7% |
| T | 9.1% |
| A | 8.2% |
| O | 7.5% |
| I | 7% |
| N | 6.7% |
| S | 6.3% |

*These are approximate averages for English. In a long ciphertext, the most common symbol is likely E, the next T, and so on. That is enough to start cracking the cipher.*

Count the letters in the ciphertext and match the most common to E, T and A. Then use common words like THE and AND to refine. A few paragraphs of text usually give it away. This is frequency analysis, known for over a thousand years.

## What modern encryption does differently

**Caesar versus modern encryption**

| Caesar cipher | Modern ciphers like AES-256 |
| --- | --- |
| 25 keys | 2²⁵⁶ keys |
| Same shift for every letter | Scrambles bits, not letters |
| Pattern of the language shows through | Output looks random |
| Cracked by hand in minutes | Brute force is infeasible |

*AES-256 has about 1.2 × 10⁷⁷ keys. And the security does not rest on keeping the method secret, only the key.*

## Where the Caesar cipher is still used

- **Puzzles and escape rooms.** Easy to make and to solve.
- **ROT13 on forums.** Used to hide spoilers so you must choose to read them. It is not security.
- **Teaching.** It shows keys, encryption and brute force in one small example.
- **Obfuscation.** It sometimes hides text from a casual look. Do not use it for passwords. See [password entropy](https://toolcloak.pages.dev/articles/password-entropy-bits-guesses-and-why-length-beats-symbols) for what real strength looks like.

> **Never protect real secrets with it:** Base64, ROT13 and Caesar are encodings and toys, not encryption. [Base64 is not encryption](https://toolcloak.pages.dev/articles/base64-is-not-encryption-and-it-makes-files-a-third-bigger) explains the same mistake with another format.

## A quick challenge

Decode WKLV LV IXQ with a shift of 3. Go back three letters for each: W→T, K→H, L→I, V→S. The message is THIS IS FUN. Try the tool above with your own message and share the shift with a friend.

## Frequently asked questions

### How does the Caesar cipher work?

Each letter is replaced by the letter a fixed number of places further along the alphabet, wrapping around after Z.

### How many keys does a Caesar cipher have?

25 useful shifts (1 to 25). Shift 0 and 26 leave the text unchanged.

### How do you break a Caesar cipher?

Try all 25 shifts and read the one that gives real words, or use letter frequency to guess the shift.

### What is ROT13?

A Caesar cipher with a shift of 13. Applying it twice returns the original text.

### Is the Caesar cipher secure?

No. It is easy to crack and should only be used for fun or learning.



---

# Average return vs CAGR: why “20 percent a year” is not what you earned

> A fund up 50 percent, down 20 percent, then up 30 percent averages 20 percent. Your money grew about 16 percent a year. The gap is real, it has a name, and it is the number advertisements hope you will not compute.

Published 2026-10-08 by Toolcloak. 3 minute read. Original: https://toolcloak.pages.dev/articles/average-return-vs-cagr-why-20-percent-is-not-what-you-earned
Related free tool: [CAGR Calculator](https://toolcloak.pages.dev/tools/cagr-calculator)

Three years of returns: +50 percent, −20 percent, +30 percent. Add them up and divide by three, and you get **20 percent a year**. It sounds great. Put ₹1 lakh in at the start and you finish with ₹1.56 lakh, not ₹1.73 lakh. What you actually earned, each year, was about **16 percent**.

> **In short**
> - CAGR is the one steady yearly rate that links your start and end value.
> - Returns of +50%, −20% and +30% average 20% but compound to only about 16% a year.
> - Use CAGR for lump sums. Use XIRR when you add money over time, like a SIP.

That 16 is the CAGR, the compound annual growth rate. It is the single yearly rate that, applied every year, would have taken you from your starting amount to your ending amount. Once you know the difference between it and a simple average, a lot of return claims become easier to read.

## Walk through the three years

**₹1 lakh through +50%, −20%, +30% (value in rupees)**

|  | Value |
| --- | --- |
| Start | 1,00,000 |
| Year 1 (+50%) | 1,50,000 |
| Year 2 (−20%) | 1,20,000 |
| Year 3 (+30%) | 1,56,000 |

*Each year’s percentage is taken from the previous year’s balance, not from the original ₹1 lakh. The −20 percent year takes away more rupees than the +30 percent year gives back.*

Step by step: ₹1,00,000 × 1.5 = ₹1,50,000. Then × 0.8 = ₹1,20,000. Then × 1.3 = ₹1,56,000. If the true average was 20 percent every year, you would have ₹1,00,000 × 1.2 × 1.2 × 1.2 = ₹1,72,800.

## The CAGR formula

> **CAGR = (end ÷ start)^(1 ÷ years) − 1:** End ₹1,56,000, start ₹1,00,000, 3 years. 1.56 ÷ 1 = 1.56. The cube root of 1.56 is about 1.16. Subtract 1 and you get 0.16, which is 16 percent a year. The formula only cares where you started, where you ended and how long it took. The bumps in the middle do not appear in it.

Try it: [CAGR Calculator](https://toolcloak.pages.dev/tools/cagr-calculator). Enter the starting value, the ending value and the number of years to get the yearly growth rate.

## Why the simple average is always too flattering

Whenever returns jump around, the simple average is higher than the CAGR. The rougher the ride, the bigger the gap. This is sometimes called **volatility drag**. A loss hurts more than an equal gain helps, because you are then investing a smaller base. The earlier article on percentages shows why: a 50 percent fall needs a 100 percent rise to repair.

**Two paths to the same average of 20 percent (value in lakh rupees)**

|  | Bumpy: +50%, −20%, +30% | Smooth: 20% every year | Your original ₹1 lakh |
| --- | --- | --- | --- |
| Start | 1 | 1 | 1 |
| Year 1 | 1.5 | 1.2 | 1 |
| Year 2 | 1.2 | 1.44 | 1 |
| Year 3 | 1.56 | 1.728 | 1 |

*Both paths have an average return of exactly 20 percent. The smooth one ends with ₹1.73 lakh and the bumpy one with ₹1.56 lakh. The average hid ₹17,000.*

## Read return claims with this in mind

**Two ways to describe the same three years**

| Average annual return | CAGR |
| --- | --- |
| 20% | 16% |
| Add the yearly figures, divide by three | The one steady rate that gives the same result |
| Ignores the order and size of swings | Uses only start, end and years |
| Always flatters a bumpy path | What your bank balance actually did |

*If you only get one number, ask for the CAGR, or the actual start and end values so you can work it out.*

- **“Averaged 18 percent a year”** with no start and end figures is worth a second look. Ask what the CAGR was.
- **Different time windows change the story.** A fund’s CAGR over five years can differ a lot from its CAGR over ten. Picking a flattering start date is common.
- **Check fees and tax.** CAGR on the numbers the fund reports may be before costs. What lands in your account is what matters.

## Where CAGR is the right tool, and where it is not

Use CAGR when you put in **one lump sum** and took it out at the end: a house, a single investment, a company’s revenue over years. It is also good for comparing two investments over the same period, because it reduces each to one steady rate.

Do not use it directly for a monthly SIP or any plan where you add money over time. Each deposit has been invested for a different length, so a single start value is misleading. For those, a return method that accounts for the timing of every deposit, like XIRR, is the right one.

> **CAGR is smooth by construction:** A CAGR of 16 percent does not mean you earned 16 percent in any given year. It says nothing about the worst year you lived through. Always look at the range of yearly results next to it.

## Frequently asked questions

### What is CAGR?

Compound annual growth rate: the single steady yearly rate that would take an amount from its starting value to its ending value over a given number of years.

### How do I calculate CAGR?

Divide the ending value by the starting value, raise the result to the power of one over the number of years, and subtract 1. Multiply by 100 for a percentage.

### Why is average return higher than CAGR?

Because losses shrink the base you earn on next. When yearly returns vary, the simple average overstates growth, and the more they vary, the bigger the gap.

### Is CAGR the same as the return I got each year?

No. It is a smoothed figure. Your actual yearly returns may have been far higher or lower in individual years.

### Can I use CAGR for SIP investments?

Not directly, because each monthly deposit is invested for a different length of time. Use a measure like XIRR that takes the date of every deposit into account.



---

# HEX, RGB and HSL: three ways to write the same colour, and which to use

> The indigo #6366f1 is also rgb(99, 102, 241) and hsl(239, 84%, 67%). This guide explains how each format works, how to convert between them by hand, and why HSL is the friendliest one when you want a lighter, darker or matching shade.

Published 2026-10-08 by Toolcloak. 4 minute read. Original: https://toolcloak.pages.dev/articles/hex-rgb-and-hsl-three-ways-to-write-the-same-colour-and-which-to-use
Related free tool: [Color Converter](https://toolcloak.pages.dev/tools/color-converter)

A colour on a screen is three small lights, red, green and blue, each shining at some strength. HEX, RGB and HSL are different ways to write down the same three strengths. Knowing which to pick makes everyday design jobs much easier, such as making a button hover colour or a matching border.

> **In short**
> - **RGB** gives each light a strength from 0 to 255. **HEX** writes the same three numbers in base 16: `99, 102, 241` is `#6366f1`.
> - **HSL** describes a colour the way people think: **hue** (0–360° around the wheel), **saturation** (grey to vivid) and **lightness** (black to white).
> - All three can write the same **16,777,216** colours (256 × 256 × 256).
> - To make a lighter or darker shade, **change only the L in HSL**. Doing it in HEX or RGB means changing all three numbers at once.

## Same colour, three spellings

**One indigo, written three ways**

|  | Value | Note |
| --- | --- | --- |
| HEX | #6366f1 | Compact, the usual choice in design files and CSS |
| RGB | rgb(99, 102, 241) | Red, green and blue strengths from 0 to 255 |
| HSL | hsl(239, 84%, 67%) | Hue, saturation, lightness |

*These all produce the exact same pixels on screen. Pick the format that makes your next change easy.*

Try it: [Color Converter](https://toolcloak.pages.dev/tools/color-converter). Paste a colour in any format and get the others instantly, ready to copy.

## RGB: mixing three lights

Each of the three channels runs from 0 (off) to 255 (full). Full red, no green, no blue is pure red. All three at 255 makes white, and all at 0 makes black. When the three values are equal, you get a grey. Our indigo has little red (99), a bit more green (102) and a lot of blue (241), so it leans strongly blue.

**Channel strengths of #6366f1 (out of 255)**

|  | Value |
| --- | --- |
| Red | 99 |
| Green | 102 |
| Blue | 241 |

*Blue dominates at 241, so the colour reads as blue-violet. Red and green are close to each other and far lower.*

| Colour | RGB | HEX | HSL |
| --- | --- | --- | --- |
| Red | 255, 0, 0 | #ff0000 | 0°, 100%, 50% |
| Orange | 255, 128, 0 | #ff8000 | 30°, 100%, 50% |
| Green | 0, 255, 0 | #00ff00 | 120°, 100%, 50% |
| Blue | 0, 0, 255 | #0000ff | 240°, 100%, 50% |
| White | 255, 255, 255 | #ffffff | 0°, 0%, 100% |
| Black | 0, 0, 0 | #000000 | 0°, 0%, 0% |

*Pure white and black have equal channels, so they have no hue, and their saturation is 0 by convention. In HSL, white is always L = 100% and black L = 0%.*

## HEX: the same numbers in base 16

A hex code is just RGB with each number written as two characters in base 16, where the digits go 0–9 then A–F. 99 becomes 63, 102 becomes 66 and 241 becomes F1. Put them together after a hash: #6366f1.

**RGB 99 to HEX 63**

1. **Divide by 16.** 99 ÷ 16 = 6 remainder 3.
2. **Write both digits.** The quotient is the first digit (6), the remainder the second (3).
3. **Repeat for each channel.** 102 → 66, and 241 → F1 (15 × 16 + 1). Join: 63 66 F1.

> **Shorthand and transparency:** When each pair repeats, you can write three digits: `#f60` means `#ff6600`. Adding two more characters gives transparency: `#6366f180` is the same indigo at about 50% opacity (`80` hex = 128 out of 255). This is a detail worth remembering when you use [colour contrast](https://toolcloak.pages.dev/articles/is-your-text-readable-the-contrast-ratio-explained) checks on transparent layers.

To learn more about counting in base 16, see [binary, hex and decimal](https://toolcloak.pages.dev/articles/binary-hex-and-decimal-counting-in-bases-2-10-and-16).

## HSL: describing colour like a person

Ask someone to describe a colour and they say “a bright blue” or “a pale pink”. HSL matches that language. **Hue** is the position on the colour wheel in degrees (red at 0°, green at 120°, blue at 240°). **Saturation** is how vivid it is, from grey (0%) to full colour (100%). **Lightness** goes from black (0%) through the pure colour (50%) to white (100%).

**The hue wheel, laid flat (degrees)**

| Zone | From | To |
| --- | --- | --- |
| Red | 0 | 30 |
| Orange | 30 | 60 |
| Yellow | 60 | 90 |
| Green | 90 | 150 |
| Cyan | 150 | 210 |
| Blue | 210 | 270 |
| Purple | 270 | 330 |
| Pink | 330 | 360 |

*Our indigo sits at 239°, in the blue band, close to violet. Turn the dial and you move around the rainbow without changing how vivid or light it is.*

**Saturation of #6366f1: 84%**

| Zone | Range |
| --- | --- |
| Grey | 0–20% |
| Muted | 20–60% |
| Vivid | 60–100% |

*The bands are only a rough guide. At 84%, this indigo is quite vivid, which is why it is popular for buttons and links.*

**Three dials, three questions**

| Box | What it means |
| --- | --- |
| Hue (H) | Which colour? 0° red, 60° yellow, 120° green, 180° cyan, 240° blue, 300° magenta. |
| Saturation (S) | How strong? 0% is grey, 100% is the purest version of the hue. |
| Lightness (L) | How bright? 0% black, 50% pure colour, 100% white. |
| Alpha (A) | How opaque? 1 is solid and 0 is invisible. Written hsla or as a fourth value. |

*Changing one dial leaves the other two alone, which is exactly why HSL is easier to edit by hand.*

## Why HSL makes shades easy

Keep the hue (239°) and saturation (84%) the same and slide only the lightness. You get a full family of shades that obviously belong together, which is the idea behind most design systems.

| Lightness | HSL | HEX | Typical use |
| --- | --- | --- | --- |
| 95% | hsl(239, 84%, 95%) | #e8e8fd | Backgrounds, tints |
| 85% | hsl(239, 84%, 85%) | #b9baf9 | Backgrounds, tints |
| 75% | hsl(239, 84%, 75%) | #8a8bf5 | Borders, icons |
| 65% | hsl(239, 84%, 65%) | #5b5df1 | Borders, icons |
| 55% | hsl(239, 84%, 55%) | #2c2fed | Borders, icons |
| 45% | hsl(239, 84%, 45%) | #1216d3 | Buttons, links |
| 35% | hsl(239, 84%, 35%) | #0e11a4 | Hover and dark text |
| 25% | hsl(239, 84%, 25%) | #0a0c75 | Hover and dark text |

*Only the L value changes from row to row. Trying to do this by editing HEX digits is much harder.*

**Making a hover colour 10% darker**

| In HEX or RGB | In HSL |
| --- | --- |
| 3 numbers | 1 number |
| Change red, green and blue together | Lower L from 67% to 57% |
| Easy to shift the hue by accident | Hue and saturation stay put |
| Needs a tool or a lot of trial | Result is the same family of colour |

*Both end as the same kind of pixels, but HSL gets you there with one edit.*

For a matching accent, add 180° to the hue to get the opposite colour on the wheel. For our indigo that is hsl(59, 84%, 67%), or #f2ef64, a bright yellow that contrasts strongly.

> **Equal lightness is not equal brightness:** A yellow and a blue with the same L look very different in brightness, because our eyes are more sensitive to green and yellow light. If you need guaranteed readability, check the actual contrast ratio of the text and its background.

## How big is the colour space?

**Counting colours**

|  | Value | Note |
| --- | --- | --- |
| Levels per channel | 256 | 0 to 255, which is 8 bits |
| Colours in total | 16.7 million | 256 × 256 × 256 = 16,777,216 |
| Storage per pixel | 24 bits | Three bytes, one per channel |

*More than enough colours for ordinary screens, which is why 8 bits per channel became the standard.*

## Quick practice

1. What is `#ff0000` in RGB? **rgb(255, 0, 0)**, pure red.
2. What is `#808080`? **rgb(128, 128, 128)**, a mid grey with saturation 0%.
3. What is `#f60` written in full? **#ff6600**, which is rgb(255, 102, 0).
4. You want a lighter version of hsl(239, 84%, 67%). **Raise L**, for example to 77% for a softer tint.

Use HEX to copy and paste, RGB when working with code that needs numbers, and HSL when you want to adjust a colour by hand. A converter lets you switch whenever it helps.

## Frequently asked questions

### How do you convert HEX to RGB?

Split the code into three pairs and convert each from base 16 to decimal. For #6366f1: 63 → 99, 66 → 102, f1 → 241, so rgb(99, 102, 241).

### What is the difference between RGB and HSL?

They describe the same colours. RGB mixes red, green and blue light. HSL uses hue, saturation and lightness, which is closer to how people think about colour.

### What does #fff mean?

It is shorthand for #ffffff, which is white. Each digit is doubled, so #f60 is #ff6600.

### How do I make a colour lighter?

In HSL, raise the lightness value. In RGB or HEX you would need to move all three channels closer to 255.

### How many colours can HEX codes represent?

16,777,216 (256³), since each of the three channels has 256 levels.

### How do I add transparency to a HEX colour?

Add two more hex digits for alpha: 00 is invisible and ff is solid. For example, #6366f180 is about 50% opaque.



---

# Start at 25 or 35? What waiting ten years really costs

> Ravi invests ₹5,000 a month from age 25 to 35, then stops. Meera starts at 35 and keeps going until 60. At 12% a year, Ravi ends with ₹2.30 crore and Meera with ₹95 lakh, even though Meera put in 2.5 times more money. Here is the whole picture, year by year.

Published 2026-10-08 by Toolcloak. 5 minute read. Original: https://toolcloak.pages.dev/articles/start-at-25-or-35-what-waiting-ten-years-really-costs
Related free tool: [Investment Growth Calculator](https://toolcloak.pages.dev/tools/compound-growth-calculator)

Most advice about investing says “start early” and stops there. It never shows you how much early is worth. The answer is bigger than most people guess, and it comes from one idea: money that is invested earlier has more years to earn on its own earnings.

> **In short**
> - Ravi invests ₹5,000 a month for **10 years (₹6 lakh in total)** and then stops. By 60 he has about **₹2.30 crore**.
> - Meera invests ₹5,000 a month for **25 years (₹15 lakh in total)**, starting at 35. By 60 she has about **₹95 lakh**.
> - The early starter wins by **₹1.35 crore** while putting in **₹9 lakh less**. All numbers assume 12% a year, which is an illustration and not a promise.
> - To catch up from 35, Meera would need to invest about **₹12,100 a month**, more than **2.4 times** as much.

## The two people in this story

Both invest the same amount each month, ₹5,000, and both earn the same 12% a year, compounded monthly. The only difference is the calendar. Ravi starts at 25 and stops at 35. Meera starts at 35 and carries on until 60. Nobody withdraws anything before 60.

Ravi’s total outlay is ₹6 lakh. Meera’s is ₹15 lakh. If money were a simple pile, Meera would be far ahead. Look at what happens instead.

**Value of each portfolio from age 25 to 60 (lakh rupees)**

| Age | Ravi: ₹5,000 a month from 25 to 35, then stops | Meera: ₹5,000 a month from 35 to 60 |
| --- | --- | --- |
| 25 | 0 | 0 |
| 30 | 4.1 | 0 |
| 35 | 11.6 | 0 |
| 40 | 21.1 | 4.1 |
| 45 | 38.3 | 11.6 |
| 50 | 69.7 | 25.2 |
| 55 | 126.5 | 50 |
| 60 | 229.9 | 94.9 |

*Ravi stops adding money at 35 and his line keeps climbing. Meera’s line only catches Ravi’s starting value around age 45.*

## Why the early money does so much work

Compounding means your returns earn returns. A rupee invested at 25 has 35 years to do that. A rupee invested at 35 has 25. The extra ten years do not add ten years’ worth of growth. They multiply the growth, because every year builds on the one before.

> **Future value of a monthly investment:** FV = P × [((1 + i)ⁿ − 1) ÷ i] × (1 + i), where P is the monthly amount, i is the monthly rate (12% ÷ 12 = 1%) and n is the number of months. The 1.01 rises to the power 300 for Meera and the growth afterwards uses the same idea for Ravi’s 25 idle years.

Ravi’s 120 deposits turn into about ₹11.6 lakh by age 35. Then, with no new money, that ₹11.6 lakh grows for 25 more years. At 12% a year it multiplies by roughly 19.8 times, because (1.01)³⁰⁰ is about 19.8. That single step is what produces ₹2.30 crore.

**The headline numbers**

|  | Value | Note |
| --- | --- | --- |
| Ravi at 60 | ₹2.30 Cr | Invested only ₹6 lakh |
| Meera at 60 | ₹95 L | Invested ₹15 lakh |
| Ravi’s lead | 2.4× | With 40% of the money |

**What each portfolio is made of at 60**

|  | Money you put in | Growth on top |
| --- | --- | --- |
| Ravi | ₹6 L | ₹2.24 Cr growth |
| Meera | ₹15 L | ₹80 L growth |

*Of Ravi’s final amount, about 97% is growth. For Meera it is about 84%. Time did the heavy lifting in both cases.*

Try it: [Investment Growth Calculator](https://toolcloak.pages.dev/tools/compound-growth-calculator). Enter an amount, a rate and a number of years to see how your own money compounds.

## What a delay of one year costs

The loss is not even from year to year. Each extra year of waiting at the start removes more rupees than the year before it did, because those early years sit under the biggest pile of later growth.

**Final amount at 60 for ₹5,000 a month, by the age you start (lakh rupees)**

|  | Value |
| --- | --- |
| Start 25 | ₹3.25 Cr |
| Start 27 | ₹2.55 Cr |
| Start 30 | ₹1.77 Cr |
| Start 35 | ₹95 L |
| Start 40 | ₹50 L |
| Start 45 | ₹25 L |

*Starting at 30 instead of 25 costs about ₹1.48 crore. Starting at 35 instead of 30 costs about ₹82 lakh. Every delay hurts, and the early ones hurt most.*

## Start age and monthly amount together

If you cannot invest a large amount, starting earlier makes up for a lot. The table shades the result at age 60 so you can see how the two levers trade off.

**Corpus at 60 at 12% a year (lakh rupees)**

| Start age | ₹3,000/mo | ₹5,000/mo | ₹10,000/mo |
| --- | --- | --- | --- |
| 25 | ₹1.95 Cr | ₹3.25 Cr | ₹6.50 Cr |
| 30 | ₹1.06 Cr | ₹1.76 Cr | ₹3.53 Cr |
| 35 | ₹57 L | ₹95 L | ₹1.90 Cr |
| 40 | ₹30 L | ₹50 L | ₹1 Cr |
| 45 | ₹15 L | ₹25 L | ₹50 L |

## The price of catching up

Say Meera realises at 35 what Ravi did. She wants to reach ₹2.30 crore by 60. The monthly amount she needs goes up fast the later she begins.

**Monthly investment needed to reach ₹2.30 crore at 60**

|  | Per month |
| --- | --- |
| Start at 25 (10 years) | ₹5,000 (then stop) |
| Start at 35 | ₹12,114 |
| Start at 40 | ₹23,008 |
| Start at 45 | ₹45,559 |

*Each five years of delay roughly doubles the monthly amount needed. This is why “I will start when I earn more” is expensive.*

## What if returns are lower?

12% is a generous illustration, and no investment gives a smooth return every year. The point is not the exact rupee figure, it is the shape. Here is the same Ravi against Meera at three different rates.

| Rate a year | Ravi at 60 | Meera at 60 | Ravi ÷ Meera |
| --- | --- | --- | --- |
| 8% | ₹67.6 lakh | ₹47.9 lakh | 1.4× |
| 10% | ₹1.25 crore | ₹66.9 lakh | 1.9× |
| 12% | ₹2.30 crore | ₹94.9 lakh | 2.4× |

*Even at 8%, the early starter ends ahead with 40% of the money. A higher rate widens the gap because there is more growth to compound.*

## What this does not say

- **Ravi should stop at 35.** He should not. The story shows what the first ten years are worth, not that stopping is wise. Ravi who keeps going would end with about ₹3.25 crore.
- **Returns are guaranteed.** They are not. Market-linked investments rise and fall, and a bad year early or late changes the result. Use the figures to understand the shape, not to promise yourself a number.
- **You should drain your emergency money to start early.** Build a safety cushion first. Our [emergency fund guide](https://toolcloak.pages.dev/articles/how-big-should-an-emergency-fund-be) shows how big it should be.
- **Inflation is ignored.** ₹2.30 crore in 35 years buys much less than ₹2.30 crore today. See [what inflation does to your savings](https://toolcloak.pages.dev/articles/what-will-one-lakh-buy-in-2046-inflation-in-pictures) for the other side of the story.

> **A simple way to use this:** If you can only afford a small amount today, start with it anyway and raise it every year. A [step-up SIP](https://toolcloak.pages.dev/articles/step-up-sip-raise-your-sip-10-percent-a-year) combines both levers: an early start and a growing amount.

**Ravi’s money, decade by decade**

| When | What |
| --- | --- |
| Age 25 | First ₹5,000 goes in. Nothing to show yet. The first year ends with about ₹64,000. |
| Age 35 | Last deposit, about ₹11.6 lakh. He has put in ₹6 lakh. Growth is already about ₹5.6 lakh. |
| Age 45 | About ₹38.3 lakh with no new money. The pile is more than six times what he put in. |
| Age 55 | About ₹1.27 crore. The last ten years add more than all previous years combined. |
| Age 60 | About ₹2.30 crore. Final 5 years alone add over ₹1 crore. |

The practical lesson is short. Start with whatever you can, raise the amount when you can, and leave it alone. The calculator on the tool page lets you replace 25, 35 and ₹5,000 with your own numbers.

## Frequently asked questions

### How much does starting 10 years earlier matter?

At ₹5,000 a month and 12% a year, starting at 25 instead of 35 gives about ₹3.25 crore at 60 instead of about ₹95 lakh. That is roughly 3.4 times more from the same monthly amount.

### Is it too late to start at 35 or 40?

No. A later start gives a smaller result, but investing for 20 or 25 years still builds a large amount. Raise the monthly amount to make up for lost time, for example with a step-up SIP.

### Why did Ravi end up with more while investing less?

His ₹11.6 lakh at age 35 compounded for 25 more years. Early money gets the longest period to earn returns on returns, so a smaller early amount can beat a larger late one.

### Is 12% a realistic return?

It is an illustration. Actual returns vary every year and are not guaranteed. The comparison between early and late holds at lower rates too, for example 8% or 10%, just with smaller gaps.

### What is the formula for monthly compounding?

FV = P × [((1 + i)ⁿ − 1) ÷ i] × (1 + i), where P is the monthly deposit, i is the monthly rate and n is the number of months. Money left to grow without new deposits is multiplied by (1 + i)ⁿ.



---

# The minimum payment trap: what a credit card balance really costs

> Pay only the small amount the statement asks for and a ₹50,000 balance can take almost three years and cost ₹37,000 extra. Here is the arithmetic, drawn out, and the one number that shows whether you are getting anywhere.

Published 2026-10-08 by Toolcloak. 3 minute read. Original: https://toolcloak.pages.dev/articles/the-minimum-payment-trap-what-a-credit-card-balance-really-costs
Related free tool: [Credit Card Payoff Calculator (Interest & Months)](https://toolcloak.pages.dev/tools/credit-card-interest-calculator)

Your statement says total due ₹50,000 and minimum due ₹2,500. That small second number feels like a kindness. It is also the most expensive thing on the page, because it quietly turns a one-time purchase into a loan with a very high rate.

> **In short**
> - Paying only ₹2,500 a month on ₹50,000 (3.5% a month) takes 35 months and costs ₹37,494 in interest.
> - Doubling the payment cuts the interest to about a third.
> - If your payment is close to balance × monthly rate, the debt barely moves.

This article uses one made-up but realistic balance and follows it month by month. All figures below come from a simple model, so treat them as an illustration of the shape, not a quote from any bank. Your own card has its own rate, fees and rules, and your statement is the final word.

## The rate looks small, the year does not

Many cards in India charge something like 3 to 3.9 percent **per month** on an unpaid balance. Say 3.5. A monthly figure sounds harmless next to an annual one, so cards usually print both. Compounded over twelve months, 3.5 percent a month is about **51 percent a year**, because each month’s interest is charged on the previous month’s interest as well.

**The example used in this article**

|  | Value | Note |
| --- | --- | --- |
| Starting balance | ₹50,000 | No new purchases after this |
| Interest per month | 3.5% | About 51% over a full year |
| Minimum-style payment | ₹2,500 | 5% of the starting balance, paid fixed |

*Assumptions: interest is added each month on the whole balance, then your payment is subtracted. Fees and taxes on the interest are ignored, so real life is a little worse.*

## What ₹2,500 a month really does

In the first month, 3.5 percent of ₹50,000 is ₹1,750 of interest. So of your ₹2,500, only ₹750 reduces the debt. The other ₹1,750 simply feeds the bank. Slowly the interest part shrinks as the balance falls, but that is a long time coming.

**How the ₹50,000 balance falls (value in thousand rupees)**

| Month | Paying ₹2,500 a month | Paying ₹5,000 a month | Paying ₹1,750 (only the interest) |
| --- | --- | --- | --- |
| 0 | 50000 | 50000 | 50000 |
| 6 | 45087 | 28712 | 50000 |
| 12 | 39049 | 2544 | 50000 |
| 18 | 31625 | 0 | 50000 |
| 24 | 22500 | 0 | 50000 |
| 30 | 11283 | 0 | 50000 |
| 36 | 0 | 0 | 50000 |

*Slide along the lines. After a full year of ₹2,500 payments you still owe about ₹39,000 of the original ₹50,000. The grey line is the cruellest: a payment equal to the monthly interest never touches the balance at all.*

Try it: [Credit Card Payoff Calculator (Interest & Months)](https://toolcloak.pages.dev/tools/credit-card-interest-calculator). Enter your balance, the monthly rate on your statement and what you can pay. It shows how long it takes and what it costs.

## The same debt, three different payments

| Monthly payment | Months to clear | Total paid | Interest paid |
| --- | --- | --- | --- |
| ₹2,500 | 35 | ₹87,494 | ₹37,494 |
| ₹5,000 | 13 | ₹62,633 | ₹12,633 |
| ₹10,000 | 6 | ₹55,961 | ₹5,961 |
| ₹1,750 | Never | Grows without end | Everything you pay |

*Same card, same starting balance, same rate. Only the monthly payment changes. Doubling the payment cuts the interest by about two thirds.*

**Total interest on the ₹50,000 balance**

|  | Interest |
| --- | --- |
| Paying ₹2,500 | ₹37,494 |
| Paying ₹5,000 | ₹12,633 |
| Paying ₹10,000 | ₹5,961 |

*Paying double the minimum does not halve the cost. It cuts it to about a third. Paying four times as much cuts it to about a sixth.*

**Where your money goes when you pay ₹2,500 a month**

|  | Share |
| --- | --- |
| The original purchases | ₹50,000 |
| Interest to the bank | ₹37,494 |

*For every ₹100 of things you actually bought, you hand over about ₹175 in total.*

## The one number that tells you if you are winning

> **Monthly interest = balance × monthly rate:** On ₹50,000 at 3.5 percent that is ₹1,750. If your payment is **close to that number**, you are paying for the privilege of owing, not repaying. Any payment has to be clearly above it before the balance starts to move.

This is also why the first year feels so unrewarding. When you pay ₹2,500, only ₹750 goes against the balance. By month 12 the interest portion has dropped, and a bigger share of each payment does real work. The early months are the expensive ones.

## What actually helps

1. **Pay the full statement when you can.** If you clear the whole balance every cycle, most cards charge no interest on purchases at all. The trap is in carrying even a small amount, because it can end that interest-free period for new purchases too. Check your card’s terms.
2. **Pick a fixed monthly amount well above the minimum.** The minimum shrinks as the balance shrinks, which keeps you in debt longer. A fixed number does not.
3. **Aim your extra money at the card, not at savings, while the rate is this high.** Nothing safe earns 51 percent a year. Paying it off is a guaranteed return of the card’s rate.
4. **Look for a cheaper loan.** A personal loan or a card balance transfer offer may charge far less than the card’s own rate. Compare the real cost including fees, not just the headline.

> **A word on fees and taxes:** Late fees, GST on interest and fees, and other charges are usually added on top. They make the real figures higher than the model here. Read the charges section of your own statement.

## Frequently asked questions

### Why is the minimum amount due so dangerous?

It is set low so that most of it is interest. A ₹50,000 balance at 3.5 percent a month costs ₹1,750 in interest in the first month, so a ₹2,500 payment barely dents the principal.

### How is credit card interest calculated?

Cards apply a monthly rate to the unpaid balance, often daily from the purchase date. The interest is added to the balance, so you pay interest on previous interest. Your statement shows the exact method and rate.

### Is 3.5 percent a month the same as 42 percent a year?

Not quite. Simple multiplication gives 42 percent, but monthly compounding makes it about 51 percent over a full year.

### How can I pay off a credit card faster?

Pay a fixed amount well above the minimum, stop adding new purchases, and consider a lower-rate personal loan or balance transfer if the total cost, including fees, is genuinely lower.

### What happens if I only pay the interest each month?

The balance does not fall. On ₹50,000 at 3.5 percent you would pay ₹1,750 every month forever and still owe ₹50,000.



---

# Reading a cron expression: five fields and six gotchas

> “0 9 * * 1-5” looks like noise. It means 9:00 every weekday. Once you can read the five fields left to right, you can schedule a backup, a report or a reminder in a single line.

Published 2026-10-08 by Toolcloak. 3 minute read. Original: https://toolcloak.pages.dev/articles/reading-a-cron-expression-five-fields-and-six-gotchas
Related free tool: [Cron Expression Explainer](https://toolcloak.pages.dev/tools/cron-expression-explainer)

Cron is how computers have run things on a schedule since the 1970s: backups, emails, cleanups, reports. The schedule is one line of five fields. It is compact, it is exact, and it is easy to get subtly wrong.

> **In short**
> - Five fields, in this order: **minute, hour, day of month, month, day of week**.
> - `*` means “every”, `*/15` means “every 15”, `1-5` is a range, `1,15` is a list.
> - Cron runs in the **server's time zone**, which is often not yours.
> - If you set both day-of-month and day-of-week, most crons run when **either** matches.

## The five fields

**Read it left to right**

1. **Minute.** 0 to 59
2. **Hour.** 0 to 23, in 24-hour time
3. **Day of month.** 1 to 31
4. **Month.** 1 to 12
5. **Day of week.** 0 to 6, Sunday is 0 (7 also works in many versions)

*Five spaces, five fields. Anything left as * means “every possible value”.*

## Ten lines you will actually use

| Expression | When it runs |
| --- | --- |
| `* * * * *` | Every minute |
| `*/15 * * * *` | Every 15 minutes |
| `0 * * * *` | At the top of every hour |
| `30 2 * * *` | Every day at 2:30 am |
| `0 9 * * 1-5` | At 9:00 am, Monday to Friday |
| `0 0 * * 0` | Midnight every Sunday |
| `0 0 1 * *` | Midnight on the 1st of every month |
| `0 8,20 * * *` | At 8:00 am and 8:00 pm daily |
| `0 0 1 1 *` | Midnight on 1 January |
| `*/5 9-17 * * 1-5` | Every 5 minutes between 9:00 and 17:55 on weekdays |

*Ranges are inclusive: `9-17` includes the 17:xx hour, so the last run is 17:55.*

Try it: [Cron Expression Explainer](https://toolcloak.pages.dev/tools/cron-expression-explainer). Paste a cron line and it explains it in plain English, with the next run times.

## The same job, three ways

**“Run a report”: pick the line**

| When | What |
| --- | --- |
| 0 9 * * 1-5 | Every weekday morning. Good for a daily team summary. |
| 0 9 * * 1 | Every Monday at 9. Good for a weekly digest. |
| 0 9 1 * * | The first of each month at 9. Good for billing or monthly reports. |

*Only the last three fields change. Reading cron is mostly noticing where the stars are.*

## Six gotchas

1. **Time zone.** Cron uses the server's clock. A job “at 9” on a UTC server runs at 2:30 pm in India. Check with the `date` command.
2. **Day of month plus day of week.** `0 0 13 * 5` is not “Friday the 13th”. It runs on the 13th **and** on every Friday.
3. **Every minute by accident.** Leaving the minute field as `*` while setting an hour, as in `* 9 * * *`, runs 60 times during the 9 o'clock hour.
4. **`*/n` is not “every n from now”.** `*/45` in the minutes field runs at :00 and :45 every hour, then restarts at :00.
5. **Missed runs.** If the server was off at the time, standard cron does not catch up.
6. **Different dialects.** Some systems add a seconds field or special names like `@daily`. Check yours.

> **Test before you trust it:** Run the command by hand first, then schedule it a couple of minutes ahead and watch it fire. Write output to a log file, because cron does not show you errors on screen.

## Frequently asked questions

### What do the five fields in a cron expression mean?

In order: minute (0-59), hour (0-23), day of month (1-31), month (1-12) and day of week (0-6, Sunday is 0).

### How do I run a cron job every 15 minutes?

Use */15 * * * *. It runs at minute 0, 15, 30 and 45 of every hour.

### What does 0 9 * * 1-5 mean?

At 9:00 am, Monday through Friday.

### Which time zone does cron use?

The time zone of the machine running it, which is often UTC on servers. Convert your intended time before scheduling.

### Why did my cron job run on the wrong days?

If both day-of-month and day-of-week are set, standard cron runs the job when either one matches, not only when both do.



---

# Why a 1 TB drive shows 931 GB: KB, KiB, MB and Mbps explained

> You buy a 1 TB drive and your computer says 931 GB. Nothing is missing and nobody is cheating: two honest ways of counting meet in the same box. This guide shows where the 69 GB goes, why the gap grows with every size, and why your 100 Mbps plan downloads at 12 MB a second, with a picture for each step.

Published 2026-10-08 by Toolcloak. 7 minute read. Original: https://toolcloak.pages.dev/articles/why-a-1-tb-drive-shows-931-gb-kb-kib-mb-and-mbps-explained
Related free tool: [Data Storage Converter](https://toolcloak.pages.dev/tools/data-storage-converter)

A new 1 TB drive comes out of the box. You plug it in, open the file manager, and it says 931 GB. Within a minute most people are annoyed and a few are typing a complaint. But the drive is full-sized and the computer is not wrong. They simply use two different meanings of the word “giga”. Once you see the two systems side by side, the mystery disappears for good.

> **In short**
> - Drive makers count in **powers of 1,000**: 1 TB = 1,000,000,000,000 bytes. Many operating systems count in **powers of 1,024**: 1 GiB = 1,073,741,824 bytes.
> - So a **1 TB drive is about 931 GiB**, which Windows prints as “931 GB”. The gap is **6.9% at the GB level and 9.1% at the TB level**, and it grows with size.
> - **KB, MB, GB** (lowercase b is a different thing!) are bytes. **Kb, Mb, Gb** are bits. Internet speed is sold in **megabits** per second, so divide by 8 to get megabytes.
> - A **100 Mbps** line moves at most **12.5 MB a second**. A 1 GB file needs at least **80 seconds** even when everything is perfect.

## Two ways to count a kilobyte

In everyday life “kilo” means 1,000. A kilometre is 1,000 metres, a kilogram is 1,000 grams. Computers, however, work in binary, where the natural round numbers are powers of two. Two to the power of 10 is **1,024**, which is very close to 1,000, so early engineers called it a “kilobyte” and moved on. That shortcut was fine when files were small. It became a problem when everything got big.

To end the confusion, the standards bodies defined two separate sets of names. The **decimal** prefixes (kilo, mega, giga, tera) always mean powers of 1,000. The **binary** prefixes (kibi, mebi, gibi, tebi) always mean powers of 1,024. A kibibyte is written **KiB**, a gibibyte **GiB**. The trouble is that many people, and some software, still say “GB” when they really mean GiB.

| Name | Symbol | Exact size in bytes | Counting system |
| --- | --- | --- | --- |
| kilobyte | KB | 1,000 | Decimal (powers of 1,000) |
| kibibyte | KiB | 1,024 | Binary (powers of 1,024) |
| megabyte | MB | 1,000,000 | Decimal |
| mebibyte | MiB | 1,048,576 | Binary |
| gigabyte | GB | 1,000,000,000 | Decimal |
| gibibyte | GiB | 1,073,741,824 | Binary |
| terabyte | TB | 1,000,000,000,000 | Decimal |
| tebibyte | TiB | 1,099,511,627,776 | Binary |

*Each binary unit is a little bigger than its decimal twin. The difference is small at KB and large at TB.*

**Same word, two meanings**

| Decimal (drive makers, networks, most phones) | Binary (Windows file sizes, memory, many tools) |
| --- | --- |
| 1 GB = 1,000,000,000 B | 1 GiB = 1,073,741,824 B |
| Powers of 1,000 | Powers of 1,024 |
| Matches the metric system | Matches how memory chips are built |
| Used on the box of a drive or SD card | Often still labelled “GB” on screen |

*Neither side is lying. Problems start only when one side’s number is read with the other side’s meaning. Note that macOS and most phones now show decimal sizes, which is why a 1 TB drive often looks like 1 TB there.*

## Where the 69 GB goes

Take the box that says 1 TB. The maker means 1,000,000,000,000 bytes. Your operating system divides that by 1,073,741,824 (bytes in a GiB) and prints the result with the letters GB. The division gives **931.3**. That is the whole story, and no space has been taken away. The same number of bytes is written in a larger unit, so the number gets smaller.

**A 1 TB drive, as the box and as the screen count it**

| Step | Amount |
| --- | --- |
| Box says | 1,000 GB |
| Unit difference | −68.7 |
| Screen says | 931.3 GiB |

*1,000,000,000,000 ÷ 1,073,741,824 = 931.3. The drive still holds exactly the same number of bytes. A little more space is then used by the file system and system files, which is a separate and much smaller effect.*

The same arithmetic works for every drive size you can buy. A rule of thumb that is easy to remember: **the screen number is about 93% of the box number** for anything in gigabytes.

**What the box says and what a binary-counting screen shows (GB)**

|  | On the box | On the screen |
| --- | --- | --- |
| 128 GB | 128 GB | 119.2 GB |
| 256 GB | 256 GB | 238.4 GB |
| 512 GB | 512 GB | 476.8 GB |
| 1 TB | 1000 GB | 931.3 GB |
| 2 TB | 2000 GB | 1862.6 GB |

*Every pair shows the same 6.9% gap. A 2 TB drive appears to “lose” 137 GB, which sounds dramatic but is the same percentage as the 128 GB drive losing 8.8 GB.*

Try it: [Data Storage Converter](https://toolcloak.pages.dev/tools/data-storage-converter). Convert between bytes, KB, MB, GB, TB and the binary units to see both numbers for any size.

## The gap grows as the units get bigger

The ratio between binary and decimal is 1.024 for one step (kilo), 1.024 squared for two steps (mega), and so on. Each extra step multiplies in another 2.4%. That is why nobody noticed the difference in the floppy-disk era and everybody notices it now. Small files hide the gap and big drives expose it.

**How much smaller the binary number looks, by unit (percent)**

| Unit | Gap between decimal and binary |
| --- | --- |
| KB | 2.3 |
| MB | 4.6 |
| GB | 6.9 |
| TB | 9.1 |
| PB | 11.2 |

*The gap is 1 − (1000 ÷ 1024)ⁿ. At petabyte scale a “1 PB” system shows up as about 0.89 PiB, so storage vendors and cloud bills must be careful about which unit they mean.*

**A ladder of units, each one 1,000 times (or 1,024 times) the one before**

| When | What |
| --- | --- |
| 1 B | Byte. One character of plain English text takes about one byte. |
| 1 KB | Kilobyte. A short email without pictures is a few kilobytes. |
| 1 MB | Megabyte. A compressed photo from an older phone, or a minute of low-quality music. |
| 1 GB | Gigabyte. A short HD video or a few hundred photos. |
| 1 TB | Terabyte. A modern laptop drive. About 1,000 GB. |
| 1 PB | Petabyte. The scale of big data centres, not laptops. |

> **A fun fossil: the “1.44 MB” floppy disk:** A classic floppy held 1,474,560 bytes. That is 1,440 × 1,024. The label “1.44 MB” mixed both systems at once: it took 1,440 KiB and divided by 1,000, a reminder that the confusion is older than most of us.

## Bits versus bytes: the lowercase b matters

There is a second mix-up that costs people far more patience than the 931 GB one. A **byte** is 8 **bits**. Storage is measured in bytes (capital B): KB, MB, GB. Network speed is measured in bits (lowercase b): Kbps, **Mbps**, Gbps. When an internet plan says 100 Mbps, it means 100 megabits per second. A file that is 100 MB is 800 megabits.

**Four labels that look almost the same**

| Box | What it means |
| --- | --- |
| MB (capital B) | Megabytes. Used for file sizes, downloads and storage. 1 MB = 8 megabits. |
| Mb (small b) | Megabits. Used for the size of a stream or the speed of a link. 8 Mb = 1 MB. |
| MiB | Mebibytes, binary: 1,048,576 bytes. Shown by some tools as MB. |
| Mbps or Mb/s | Megabits per second. Speed, not size. Divide by 8 to get MB per second. |

*When a number looks wrong by a factor of eight, check the b. It is almost always bits versus bytes.*

**From your plan’s speed to your download time**

1. **Start with the plan.** For example 100 Mbps, which is 100 megabits a second.
2. **Divide by 8.** 100 ÷ 8 = 12.5 megabytes a second at the very best.
3. **Divide the file by that speed.** A 1,000 MB file takes 1,000 ÷ 12.5 = 80 seconds.
4. **Expect a little extra.** Wi-Fi, servers and protocol overhead usually leave you with 80 to 95 percent of the plan.

This is also why a speed test result and a download counter never match at first glance. The speed test says 94 Mbps and the browser says 11.7 MB/s. Divide 94 by 8 and you get 11.75. Both are right.

## How long will it really take?

Here is a quick reference of the best-case time for files of common sizes. It uses the formula size in MB × 8 ÷ speed in Mbps = seconds, and assumes the full speed is available all the time, which is optimistic.

**Best-case download time (file size against line speed)**

| File size | 10 Mbps | 50 Mbps | 100 Mbps | 500 Mbps | 1 Gbps |
| --- | --- | --- | --- | --- | --- |
| 100 MB | 80 s | 16 s | 8 s | 1.6 s | 0.8 s |
| 1 GB | 13 min 20 s | 2 min 40 s | 1 min 20 s | 16 s | 8 s |
| 5 GB | 1 h 7 min | 13 min 20 s | 6 min 40 s | 1 min 20 s | 40 s |
| 50 GB | 11 h 7 min | 2 h 13 min | 1 h 7 min | 13 min 20 s | 6 min 40 s |

**Three conversions worth remembering**

|  | Value | Note |
| --- | --- | --- |
| Mbps to MB per second | ÷ 8 | 100 Mbps is at best 12.5 MB/s |
| Box GB to screen GB | × 0.93 | 512 GB on the box is about 477 on screen |
| Box TB to screen TiB | × 0.91 | 1 TB is about 0.909 TiB |

*These are rules of thumb for the usual Windows case. Always check which unit your own tool is showing.*

## Why the sizes on your phone look different

Most phones, macOS and many web services now use decimal units, so a 128 GB phone shows 128 GB of storage, with the operating system taking a chunk. Windows still reports sizes in binary units but labels them with the decimal names. Linux tools often show both, with an option to switch (the “-h” and “--si” flags in common commands). So the same 1 TB drive can honestly look like 1 TB on a Mac and 931 GB on a Windows laptop.

- **Windows File Explorer:** binary counting, decimal-looking labels. 1 TB shows as about 931 GB.
- **macOS and iPhone:** decimal counting. A 1 TB drive shows about 1 TB.
- **Most Android phones:** vary by maker. Check the unit in the storage screen.
- **Cloud plans:** usually decimal, but read the fine print before you plan around the last few percent.

> **Memory is the exception:** RAM is still sold in binary sizes (8 GB of RAM really means 8 GiB), because memory chips are built from powers of two. Storage and memory are counted differently on the same computer.

## Practical tips

1. **Plan with the screen number.** If a project needs 900 GB, a 1 TB drive is too tight on Windows. Buy the next size up.
2. **Check the b.** For anything about speed, look for the lowercase b. Divide by 8 for MB per second.
3. **Keep 10 to 15 percent free.** Drives slow down and cannot reorganise files when they are almost full. That slack is far bigger than the unit gap.
4. **Compress before you upload.** Smaller files save time and storage. The [image compressor](https://toolcloak.pages.dev/tools/image-compressor) can cut a photo by 80 to 90 percent, and [this guide](https://toolcloak.pages.dev/articles/why-your-4mb-photo-is-really-a-400kb-photo) shows how.
5. **Remember Base64 adds a third.** If you embed files as text, the size grows by about 33 percent. See [Base64 is not encryption](https://toolcloak.pages.dev/articles/base64-is-not-encryption-and-it-makes-files-a-third-bigger).

If you enjoy the number side of this, the [binary, hex and decimal](https://toolcloak.pages.dev/articles/binary-hex-and-decimal-counting-in-bases-2-10-and-16) guide explains why 1,024 is such a natural number for computers.

## A quick self-test

A 512 GB SD card is full of files. How many GiB will Windows show? Multiply by 1,000,000,000 and divide by 1,073,741,824: about 476.8. And how long does a 2 GB video take to upload on a 20 Mbps upload line? 2,000 MB × 8 ÷ 20 = 800 seconds, about 13 minutes at best. Try your own numbers in the converter above and you will never be surprised by a drive label again.

## Frequently asked questions

### Why does my 1 TB drive show only 931 GB?

The maker counts 1 TB as 1,000,000,000,000 bytes. Windows divides by 1,073,741,824 to show gibibytes, but writes the unit as GB. The bytes are all there; the number is just written in a bigger unit.

### What is the difference between GB and GiB?

1 GB is 1,000,000,000 bytes (decimal). 1 GiB is 1,073,741,824 bytes (binary, 1,024 cubed). A GiB is about 7.4% larger than a GB.

### What is the difference between MB and Mbps?

MB means megabytes, a file size. Mbps means megabits per second, a speed. One byte is eight bits, so 100 Mbps is at best 12.5 MB per second.

### How many MB are in a GB?

1,000 MB in a GB by the decimal definition, and 1,024 MiB in a GiB by the binary definition.

### Why is RAM sold in binary sizes but drives in decimal?

Memory chips are built from powers of two, so their natural sizes are 4, 8, 16 and 32 GiB. Drive makers adopted the decimal prefixes because they match the metric system and give larger-looking numbers.

### Is the missing space a scam?

No. The decimal and binary systems are both documented, and the box states the real byte count. The confusion comes from software showing binary sizes with decimal-looking labels.



---

# Two dice and a coin: why seven wins, and why a streak proves nothing

> A fair die or coin is the simplest random machine there is, yet it fools people every day. This guide uses all 36 outcomes of two dice, the famous 17th-century gambling puzzle and a 5,000-flip experiment to show what chance really does.

Published 2026-10-08 by Toolcloak. 4 minute read. Original: https://toolcloak.pages.dev/articles/two-dice-and-a-coin-why-seven-wins-and-why-streaks-prove-nothing
Related free tool: [Dice Roller & Coin Flip](https://toolcloak.pages.dev/tools/dice-roller-coin-flip)

Every board game, every cricket toss and every “heads or tails” argument rests on the same idea: some outcomes are equally likely, so counting them gives you the odds. Counting is easy. What goes wrong is intuition, which finds patterns in streaks and believes that luck must “even out” in the next roll.

> **In short**
> - With two dice there are **36 equally likely outcomes**. A total of **7 can happen 6 ways (16.7%)**, while 2 and 12 happen only 1 way each (2.8%).
> - Rolling at least one six in 4 throws of one die is **51.8%**. At least one double-six in 24 throws of two dice is **49.1%**: slightly worse, not better.
> - A fair coin has **no memory**. After five heads in a row the next flip is still 50:50.
> - Over thousands of flips the **share of heads** settles near 50%. The **difference in count** does not shrink.

## One die: simple, and boring on purpose

A fair die has six faces, each with the same chance of landing up: 1 in 6, or about 16.7%. That means each number is “due” no more often than any other. Add a second die and things become interesting, because the **total** is no longer equally likely.

Try it: [Dice Roller & Coin Flip](https://toolcloak.pages.dev/tools/dice-roller-coin-flip). Roll one or several dice or flip a coin. Roll a few hundred times and watch the totals build up the shape described below.

## Two dice: counting the 36 outcomes

Treat the two dice as different (say one red, one blue). There are 6 × 6 = 36 equally likely pairs. The table below shows the total for every pair. Look along the diagonals: each diagonal has the same sum, and the middle one is the longest.

**Total of two dice, for every combination**

| Red die | 1 | 2 | 3 | 4 | 5 | 6 |
| --- | --- | --- | --- | --- | --- | --- |
| 1 | 2 | 3 | 4 | 5 | 6 | 7 |
| 2 | 3 | 4 | 5 | 6 | 7 | 8 |
| 3 | 4 | 5 | 6 | 7 | 8 | 9 |
| 4 | 5 | 6 | 7 | 8 | 9 | 10 |
| 5 | 6 | 7 | 8 | 9 | 10 | 11 |
| 6 | 7 | 8 | 9 | 10 | 11 | 12 |

**How many of the 36 outcomes give each total**

|  | Value |
| --- | --- |
| 2 | 1 |
| 3 | 2 |
| 4 | 3 |
| 5 | 4 |
| 6 | 5 |
| 7 | 6 |
| 8 | 5 |
| 9 | 4 |
| 10 | 3 |
| 11 | 2 |
| 12 | 1 |

*The totals form a triangle: 1, 2, 3, 4, 5, 6, 5, 4, 3, 2, 1. Seven has six ways and is the most likely total.*

**Chance of each total (%)**

|  | Chance |
| --- | --- |
| Total 2 | 2.8% |
| Total 3 | 5.6% |
| Total 4 | 8.3% |
| Total 5 | 11.1% |
| Total 6 | 13.9% |
| Total 7 | 16.7% |
| Total 8 | 13.9% |
| Total 9 | 11.1% |
| Total 10 | 8.3% |
| Total 11 | 5.6% |
| Total 12 | 2.8% |

*A total of 7 comes up about one roll in six. A total of 12 (double six) comes up about one roll in thirty-six.*

**Where the 36 outcomes sit**

|  | Value | Note |
| --- | --- | --- |
| Totals 6, 7 and 8 | 16 of 36 | 44% of all rolls land in these three |
| Totals 2 and 12 | 2 of 36 | Only 5.6% of rolls, one way each |
| Any double | 6 of 36 | 1+1 up to 6+6: one roll in six |

*The middle of the range collects most of the outcomes, which is why 6, 7 and 8 come up so often in dice games.*

> **Adding the dice is not the same as rolling a single 2–12 die:** A single die that gave 2 to 12 evenly would make every total equally likely. Two dice do not, because several different pairs lead to the same total. The more dice you add, the more bell-shaped the totals become. See [the bell curve](https://toolcloak.pages.dev/articles/the-bell-curve-68-95-99-7-rule-z-scores-and-what-rare-really-means) for why.

## A gambler’s puzzle from the 1600s

A French nobleman, the Chevalier de Méré, noticed that he made money betting on at least one six in four rolls of a die, but lost on at least one double six in 24 rolls of two dice. The two bets sound similar, since 4 is to 6 as 24 is to 36. The reason they are different is that “at least one” is easiest to count through its opposite: no success at all.

**Chance of at least one six in four rolls**

1. **Chance of no six on one roll.** 5 out of 6, which is about 0.833.
2. **Chance of no six in all four rolls.** (5/6)⁴ = 0.482, because the rolls are independent.
3. **Flip it around.** 1 − 0.482 = 0.518, so a six turns up 51.8% of the time.

**Out of 100 games of each bet, how many are won?**

| Group | Count |
| --- | --- |
| Win: a six in 4 rolls | 51.8% |
| Lose | 48.2% |

*A small edge for the player, enough to win steadily over many games.*

**The double-six bet in 24 rolls**

| Group | Count |
| --- | --- |
| Win: a double six in 24 rolls | 49.1% |
| Lose | 50.9% |

*Barely under half. Chance of no double six per roll is 35/36, and (35/36)²⁴ = 0.509. Pascal and Fermat solved this around 1654, which helped start probability theory.*

## The coin has no memory

Suppose a fair coin lands heads five times in a row. Many people feel tails is now “due”. It is not. Each flip is independent: the coin does not know what happened before. The chance of five heads in a row **before you start** is small, but the chance of a sixth head **after** five is just 50%.

**Chance of a run of heads in a row, from the start (%)**

|  | Chance |
| --- | --- |
| 1 head | 50% |
| 2 heads | 25% |
| 3 heads | 12.5% |
| 5 heads | 3.13% |
| 8 heads | 0.39% |
| 10 heads | 0.1% |

*Each extra head halves the chance. Five in a row is 3.1%, ten in a row is about 0.1%. Rare, but across many flips you will see them.*

> **The gambler’s fallacy:** Believing that a long streak must be corrected by the opposite outcome is called the gambler’s fallacy. A famous casino example from 1913 had the ball land on black 26 times in a row. Many players kept betting red, because it was “due”. Each spin was still close to 50:50 (a little less, because of the green zero).

## What really settles down

A coin was flipped by computer in an experiment, 5,000 times with a fixed seed so the result is repeatable. The share of heads wandered at first, then calmed down:

**Share of heads after more and more flips (%)**

| Number of flips | Share of heads | Fair value 50% |
| --- | --- | --- |
| 10 | 40 | 50 |
| 50 | 58 | 50 |
| 100 | 57 | 50 |
| 500 | 51.4 | 50 |
| 1,000 | 50.6 | 50 |
| 5,000 | 49.8 | 50 |

*This is the law of large numbers: the share of heads gets closer to 50% as flips grow.*

The common mistake is to think the count of heads and tails must also even out. It does not have to. A lead of 40 extra heads after 1,000 flips is 52% heads, but after 100,000 flips it is only 50.02%, even though the gap is exactly the same. Chance does not correct the past; it simply buries it under new flips.

## Quick practice

1. What is the chance of a total of 9 on two dice? 4 of 36, which is **11.1%**.
2. Which two totals are equally likely as 5? **9**, since both have 4 ways.
3. After four tails in a row, what is the chance the next flip is heads? **50%**.
4. What is the chance of at least one head in 3 flips? 1 − (1/2)³ = **87.5%**.

Randomness has a shape, and counting reveals it. For where randomness meets security, read about [UUID collisions](https://toolcloak.pages.dev/articles/can-two-uuids-ever-be-the-same-the-odds-in-pictures) and [password entropy](https://toolcloak.pages.dev/articles/password-entropy-bits-guesses-and-why-length-beats-symbols).

## Frequently asked questions

### What is the most likely total with two dice?

Seven. It can happen in 6 of the 36 ways, which is 16.7%.

### What is the chance of rolling a double six?

1 in 36, or about 2.8%.

### Is a coin more likely to land tails after many heads?

No. Each flip is independent. A fair coin is always 50:50, no matter what came before.

### What is the chance of rolling at least one six in four rolls?

About 51.8%. The chance of no six at all is (5/6)⁴ = 48.2%, so at least one six is the rest.

### What is the law of large numbers?

As the number of trials grows, the average result (such as the share of heads) approaches the true probability. It does not mean the raw counts must even out.

### Are online dice rollers truly random?

They use a pseudo-random number generator, which is fine for games. Anything that needs true unpredictability, such as passwords, should use a cryptographically secure source.



---

# 50% off plus 20% off is not 70% off: how stacked discounts really work

> A sale says “extra 20% off already reduced prices”. Your brain adds the two numbers. The shop does not. The real answer is 60 percent, and the gap gets wider the bigger the discounts get.

Published 2026-10-08 by Toolcloak. 3 minute read. Original: https://toolcloak.pages.dev/articles/50-percent-off-plus-20-percent-off-is-not-70-percent-off
Related free tool: [Discount Calculator](https://toolcloak.pages.dev/tools/discount-calculator)

You see a jacket marked 50 percent off. At the till, an extra 20 percent comes off. It feels like 70 percent. It is 60. The second discount is taken from the already reduced price, so it is worth less than it sounds.

> **In short**
> - Two discounts of a% and b% give a total discount of **100 − (100 − a)(100 − b) ÷ 100** percent.
> - 50% then 20% is **60%**, not 70%. 30% then 20% is **44%**, not 50%.
> - The order does not matter. Multiplying is the same either way.
> - A discount applied **before GST** and one applied **after** can give different bills.

## What actually happens to ₹100

Take an item priced at ₹100. The first discount of 50 percent removes ₹50, leaving ₹50. Now the second discount of 20 percent is taken from ₹50, not from ₹100. That is ₹10. You pay ₹40.

**Where each rupee of ₹100 goes**

| Group | Count |
| --- | --- |
| Gone with the first 50% off | ₹50 |
| Gone with the extra 20% off | ₹10 |
| What you pay | ₹40 |

*20 percent of ₹50 is ₹10, not ₹20. The second discount is applied to a smaller number.*

## The shortcut

> **Multiply what is left:** 50% off leaves 50% of the price (×0.50). 20% off leaves 80% (×0.80). Together: 0.50 × 0.80 = 0.40. So you pay 40 percent, which is a 60 percent discount.

| First | Then | What it feels like | Real discount |
| --- | --- | --- | --- |
| 20% | 10% | 30% | 28% |
| 30% | 20% | 50% | 44% |
| 40% | 30% | 70% | 58% |
| 50% | 20% | 70% | 60% |
| 50% | 50% | 100% | 75% |

*The bigger the discounts, the bigger the gap. Two 50 percent discounts are never a free item.*

**Real total discount for “50% off plus an extra …”**

|  | Value |
| --- | --- |
| +10% | 55% |
| +20% | 60% |
| +30% | 65% |
| +40% | 70% |
| +50% | 75% |

*Each extra step adds half as much as it seems, because only half of the price is left to discount.*

Try it: [Discount Calculator](https://toolcloak.pages.dev/tools/discount-calculator). Enter the price and the discount to see what you pay and what you save.

## Order does not matter, GST does

A 20 percent discount followed by a 50 percent one gives exactly the same final price as 50 then 20, because 0.8 × 0.5 equals 0.5 × 0.8. What can change your bill is **when tax is added**. In India, GST on the sale price is calculated on the price after the discount.

**A ₹2,000 item, 50% then 20% off, plus 18% GST**

| What the “70% off” thinking says | What you really pay |
| --- | --- |
| ₹708 | ₹944 |
| ₹2,000 less ₹1,400 = ₹600 | ₹2,000 × 0.5 × 0.8 = ₹800 |
| ₹600 plus 18% GST = ₹708 | ₹800 plus 18% GST = ₹944 |

*Both bills include GST. The difference is ₹236, from a discount that was 10 points smaller than it sounded. See [how GST is added or removed](https://toolcloak.pages.dev/articles/how-to-add-or-remove-gst-without-getting-it-wrong).*

## How to spot a misleading sale

- **“Up to 70% off”** means the best item is 70 percent off. Most are not.
- **“Extra 20% on already reduced”** is multiplied, not added.
- **Original prices that were never charged.** A 60 percent discount from an inflated price is not a bargain. Check the price history.
- **Discounts that remove a feature.** A cheaper model that does less is not the same product.

The same logic explains why a 50 percent loss needs a 100 percent gain to recover, as in [up 50 percent, down 50 percent](https://toolcloak.pages.dev/articles/up-50-percent-down-50-percent-and-you-are-not-back-to-zero). Percentages always apply to whatever is left.

## Frequently asked questions

### How do I calculate two discounts together?

Multiply what is left after each. For 50 percent and 20 percent off, 0.5 × 0.8 = 0.4, so you pay 40 percent of the price, which is a 60 percent total discount.

### Is 50% off plus 20% off the same as 70% off?

No. It is 60 percent off. The second discount is taken from the already reduced price.

### Does the order of the discounts matter?

No. Multiplication gives the same result in either order. 20 percent then 50 percent equals 50 percent then 20 percent.

### Is GST charged before or after a discount?

On the price after the discount. The taxable value is the discounted price, and GST is added on top of that.

### How do I find the discount percentage from two prices?

Subtract the sale price from the original, divide by the original and multiply by 100. A price falling from ₹800 to ₹600 is (200 ÷ 800) × 100 = 25 percent off.



---

# What your appliances cost per month: watts, hours and units explained

> Your bill is a count of units, and one unit is one kilowatt-hour. A 1.5-ton AC running 8 hours a day uses about 288 units a month, which at ₹8 a unit is ₹2,304. A ceiling fan for 10 hours costs ₹168. This guide shows the formula and which appliances really drive the bill.

Published 2026-10-08 by Toolcloak. 4 minute read. Original: https://toolcloak.pages.dev/articles/what-your-appliances-cost-per-month-watts-hours-and-units-explained
Related free tool: [Electricity Cost Calculator](https://toolcloak.pages.dev/tools/electricity-cost-calculator)

Electricity feels invisible until the bill arrives. Then it is one big number with no explanation. The good news is that the arithmetic behind it is only three steps, and once you can do it for one appliance you can find the two or three that are really costing you.

> **In short**
> - **Units (kWh) = watts × hours ÷ 1,000.** **Cost = units × rate per unit.**
> - At an example rate of **₹8 a unit**, a 1.5-ton AC for 8 hours a day costs about **₹2,304 a month**. A ceiling fan for 10 hours costs about **₹168**.
> - In our sample home, the AC is **about 60% of the bill**. Cooling and heating appliances matter far more than lights.
> - Tariffs differ by state, slab and season, and your wattage may differ from our figures. Treat this as a method, then plug in your own numbers.

## Watts, kilowatts and units

A **watt** measures how fast an appliance uses energy right now. A **kilowatt** is 1,000 watts. A **unit** on your bill is one kilowatt used for one hour, written kWh. So a 1,000 W heater running for one hour uses one unit.

> **Units per day = watts × hours ÷ 1,000:** A 100 W TV on for 4 hours: 100 × 4 ÷ 1,000 = 0.4 units a day. Over 30 days that is 12 units. At ₹8 a unit, ₹96 a month.

## A sample home, appliance by appliance

These are rough figures for a typical flat. An inverter AC draws less than its rating while the compressor idles, so we use an average of 1,200 W while it is on. A fridge cycles on and off, so we use a typical daily total.

| Appliance | Power | Daily use | Units a month | Cost at ₹8 |
| --- | --- | --- | --- | --- |
| 1.5-ton AC | 1,200 W | 8 hours | 288 | ₹2,304 |
| Geyser | 2,000 W | 1 hour | 60 | ₹480 |
| 3 ceiling fans | 70 W each | 10 hours | 63 | ₹504 |
| Refrigerator | cycles | 24 hours | 30 | ₹240 |
| TV | 100 W | 4 hours | 12 | ₹96 |
| 6 LED bulbs | 9 W each | 6 hours | 9.7 | ₹78 |
| Wi-Fi router | 12 W | 24 hours | 8.6 | ₹69 |

*Total about 471 units, or roughly ₹3,770 a month at a flat ₹8 a unit.*

**Monthly cost by appliance (₹, at ₹8 a unit)**

|  | Cost |
| --- | --- |
| 1.5-ton AC | ₹2304 |
| 3 ceiling fans | ₹504 |
| Geyser | ₹480 |
| Refrigerator | ₹240 |
| TV | ₹96 |
| 6 LED bulbs | ₹78 |
| Wi-Fi router | ₹69 |

*One appliance outspends the other six combined. If you want to save, start at the top.*

**Share of the sample bill**

|  | Share |
| --- | --- |
| AC | 61% |
| Fans | 13% |
| Geyser | 13% |
| Fridge | 6% |
| Everything else | 6% |

Try it: [Electricity Cost Calculator](https://toolcloak.pages.dev/tools/electricity-cost-calculator). Enter watts, hours a day and your rate to get the daily, monthly and yearly cost.

## What an hour more of AC costs

Because cost scales in a straight line with hours, you can feel the effect of each extra hour. For the 1,200 W AC at ₹8 a unit, each hour costs ₹9.60, and each extra hour a day adds about ₹288 a month.

**Monthly AC cost by hours of use per day (₹ thousand)**

| Hours a day | AC cost a month |
| --- | --- |
| 0 | 0 |
| 2 | 576 |
| 4 | 1152 |
| 6 | 1728 |
| 8 | 2304 |
| 10 | 2880 |
| 12 | 3456 |

*Going from 6 to 8 hours adds ₹576 a month. Setting the thermostat 1 degree higher, a common tip, saves a few percent of the AC’s consumption.*

## Small things add up differently

Replacing a 60 W old-style bulb with a 9 W LED saves 51 W. At 6 hours a day that is about 112 units a year per bulb, roughly ₹890 at ₹8 a unit. Across six bulbs, it comes to over ₹5,000 a year. That is a bigger win than most people expect, but still smaller than one fewer hour of AC each day all summer.

**One bulb, one year, 6 hours a day**

| 60 W old bulb | 9 W LED |
| --- | --- |
| 131 units | 20 units |
| About ₹1,051 a year at ₹8 | About ₹158 a year at ₹8 |
| Also gives off a lot of heat | Saves about ₹893 a year |

## Why your bill may not match the sum

- **Slabs.** Many states charge more per unit as your usage rises. The last units can cost more than the first, so the average rate is not the same as the top rate.
- **Fixed and other charges.** Fixed charges, fuel adjustments and taxes are added on top of the energy charge.
- **Standby power.** TVs, set-top boxes and chargers draw a little even when “off”. It is small per device but runs 24 hours a day.
- **Real watts differ from the label.** The sticker shows the maximum. Use a plug-in power meter if you want the real figure.
- **Season.** Summer and winter use differ a lot. Use a yearly average if you want to budget.

## Five ways to cut the bill

1. **Raise the AC thermostat and use a fan with it.** Every degree helps, and a fan moves the air at 70 W instead of 1,200 W.
2. **Cut geyser time.** Heat only what you need and switch off after use. A timer helps.
3. **Service the AC and fridge.** Clean filters and coils let them work with less power.
4. **Choose a high star rating when you replace a big appliance.** A 5-star unit uses less per year, which pays back over its life.
5. **Switch off at the wall.** Use a power strip for the TV, set-top box and chargers.

> **Find the big three first:** Write down the power and daily hours for every appliance over 500 W. Those usually explain most of your bill. Check small ones only after you have dealt with the big ones. If you also run a home business, our [break-even guide](https://toolcloak.pages.dev/articles/break-even-point-how-many-sales-before-you-make-a-profit) shows how to treat the bill as a fixed cost.

## Frequently asked questions

### How do I calculate electricity cost for an appliance?

Units = watts × hours ÷ 1,000. Cost = units × your rate per unit. For a 100 W TV used 4 hours a day, that is 0.4 units a day.

### What is one unit of electricity?

One unit is one kilowatt-hour (kWh): 1,000 watts used for one hour.

### How much does a 1.5-ton AC cost per month?

At about 1,200 W for 8 hours a day and ₹8 a unit, roughly ₹2,300 a month. Your actual cost depends on the model, the room, the weather and your tariff.

### Why is my bill higher than my calculation?

Slab rates, fixed charges, taxes and standby power are not in the simple formula, and real wattage can differ from the label.

### How can I find the real power of an appliance?

Use a plug-in energy meter, or read the label and remember that it shows the maximum draw, not the average.



---

# How big should an emergency fund be? A simple way to pick your number

> Three months, six months, twelve months: everyone has an opinion. The right answer depends on three things about your life, and you can work it out in two minutes with your own monthly spending.

Published 2026-10-08 by Toolcloak. 3 minute read. Original: https://toolcloak.pages.dev/articles/how-big-should-an-emergency-fund-be
Related free tool: [Emergency Fund Calculator](https://toolcloak.pages.dev/tools/emergency-fund-calculator)

A job ends, a hospital bill arrives, the laptop dies in the same week. An emergency fund is the money that turns those weeks from a crisis into an annoyance. The only real question is how much, and the honest answer is a number of **months**, not rupees.

> **In short**
> - Add up one month of must-pay costs, then multiply by 3 to 12 months depending on your situation.
> - Stable dual income: about 3 to 6 months. One income, dependents or irregular pay: 9 to 12.
> - Keep it easy to reach and boring. Its job is to be there, not to grow.

These are rules of thumb, not advice tailored to you. Your own job security, family and health matter more than any formula.

## Step 1: count only what must be paid

Do not use your full spending. In an emergency you cut the cinema and the takeaway. Add up the costs that keep going: rent or EMI, groceries, utilities, insurance premiums, loan repayments, school fees, transport. That is your **survival month**.

**An example survival month of ₹40,000**

|  | Share |
| --- | --- |
| Rent or EMI | ₹15,000 |
| Groceries and household | ₹9,000 |
| Bills, transport, phone | ₹6,000 |
| Insurance and loan payments | ₹7,000 |
| Everything else that cannot stop | ₹3,000 |

*Use your own figures. Look at three months of bank and card statements, then remove anything you could stop for a while.*

Try it: [Emergency Fund Calculator](https://toolcloak.pages.dev/tools/emergency-fund-calculator). Enter your monthly essentials and the number of months to get your target.

## Step 2: choose how many months

**How many months of cover, as a rough guide**

| Zone | From | To |
| --- | --- | --- |
| Lean | 0 | 3 |
| Typical | 3 | 6 |
| Cautious | 6 | 9 |
| Very safe | 9 | 12 |

*Months of survival costs. Where you sit on the line depends on the three questions below.*

- **How steady is your income?** A salaried job in a stable field leans lower. Freelance, commission or business income leans higher, because the gap between jobs or clients can be long.
- **Who depends on you?** Children, parents or a partner without income push the number up. A second earner in the house pulls it down.
- **How exposed are you to big surprises?** Health cover that is thin, a vehicle you rely on, or a home with ongoing repairs all argue for more.

| Months | On ₹40,000 a month | A fit for |
| --- | --- | --- |
| 3 | ₹1,20,000 | Two steady incomes, good health cover |
| 6 | ₹2,40,000 | A common starting target for most salaried people |
| 9 | ₹3,60,000 | Single income, or dependents |
| 12 | ₹4,80,000 | Self-employed, uneven income, or a risky industry |

*Simple multiplication. Replace 40,000 with your own survival month.*

**Target fund for a ₹40,000 survival month (₹ thousand)**

|  | Value |
| --- | --- |
| 3 months | 120k |
| 6 months | 240k |
| 9 months | 360k |
| 12 months | 480k |

*Each extra three months adds exactly three times the survival month.*

## Step 3: build it in a way you will stick to

A target of ₹2.4 lakh feels huge. Put ₹10,000 aside each month and you reach it in 24 months, with no investing skill at all. A smaller first milestone, one month of cover, comes after only four months.

**A gentle path at ₹10,000 a month**

| When | What |
| --- | --- |
| Month 4 | 1 month of cover: ₹40,000. The first breather. A broken phone no longer hurts. |
| Month 12 | 3 months of cover: ₹1,20,000. A short job gap is now survivable. |
| Month 24 | 6 months of cover: ₹2,40,000. The common target. You can stop adding, or slow down. |

*Ignoring interest, which only helps a little. Raise the monthly amount when pay rises and the dates move closer.*

## Where to keep it

**What the money needs to be**

| What it should be | What it should not be |
| --- | --- |
| Reachable | Locked or risky |
| Withdrawable within a day or two | Shares that may be down 30% when you need it |
| Stable value, no big swings | A deposit with a heavy penalty for breaking it |
| Separate from daily spending | The same account you spend from |

*Savings accounts, liquid funds and short deposits are the usual choices. Check current terms, since rates and rules change.*

> **Use it, then refill it:** An emergency fund that never gets used is doing its job, but one that is spent on a holiday is not. Decide in advance what counts: lost income, medical bills, urgent repairs. After using it, top it back up before anything else.

## Frequently asked questions

### How many months should an emergency fund cover?

A common starting point is 6 months of essential costs. Two steady incomes may need only 3, while a single or irregular income, or dependents, can call for 9 to 12.

### What counts as an emergency?

Unplanned, necessary costs: loss of income, medical bills, urgent home or vehicle repair. Sales, holidays and upgrades are not emergencies.

### Should I include my EMI in the calculation?

Yes. Loan payments and rent continue during an emergency, so they belong in your survival month.

### Where should I keep an emergency fund?

Somewhere you can reach within a day or two with stable value, like a savings account, a liquid fund or a short deposit. Avoid anything that could fall sharply or locks your money.

### Should I build the fund or pay debt first?

Many people build a small starter fund of about one month first, then pay down expensive debt, then grow the fund to the full target. Your situation may differ.



---

# A 7% fixed deposit: what is left after tax and inflation

> A ₹10 lakh FD at 7 percent shows ₹14 lakh after five years. After 30 percent tax on the interest it is ₹12.7 lakh. After 6 percent inflation, it buys what ₹9.5 lakh buys today. The deposit is safe. The purchasing power is not.

Published 2026-10-08 by Toolcloak. 3 minute read. Original: https://toolcloak.pages.dev/articles/fd-at-7-percent-after-tax-and-inflation-what-you-actually-earn
Related free tool: [FD Calculator](https://toolcloak.pages.dev/tools/fd-calculator)

Fixed deposits are famous for one thing: you know exactly what number you will get. What the number will buy is a different question. Three steps stand between the bank's headline rate and your real gain.

> **In short**
> - ₹10 lakh at 7% for 5 years: **₹14.03 lakh** before tax.
> - If the interest is taxed at **30 percent**, the effective rate is 4.9% and you get **₹12.70 lakh**.
> - With **6 percent inflation**, that ₹12.70 lakh buys about **₹9.49 lakh** worth of today's goods: slightly less than you started with.
> - Lower tax slabs and higher rates look better, but the exercise is worth doing with your own numbers.

## Step 1: the headline

Compounded yearly, ₹10 lakh at 7 percent grows to ₹14.03 lakh in five years. That is the number the bank puts on the brochure and the one the FD calculator shows first.

## Step 2: tax on the interest

FD interest is added to your income and taxed at your slab rate. For someone in the 30 percent slab, 30 percent of each year's interest goes in tax, so the 7 percent behaves like 4.9 percent (7 × 0.7). The lower your slab, the smaller the loss.

## Step 3: inflation

If prices rise 6 percent a year, ₹100 today needs ₹133.80 in five years to buy the same things. Dividing the after-tax balance by 1.338 gives its value in today's rupees.

**₹10 lakh in a 7% FD after 5 years**

| Step | Value |
| --- | --- |
| On the bank statement | ₹14.03 lakh |
| After 30% tax on interest | ₹12.70 lakh |
| In today's buying power (6% inflation) | ₹9.49 lakh |

*The last bar is below the ₹10 lakh you started with. The balance grew; what it buys shrank.*

**₹10 lakh in a 7% FD, year by year (lakh rupees)**

| Year | Before tax | After 30% tax | After tax, in today's rupees |
| --- | --- | --- | --- |
| 0 | 10 | 10 | 10 |
| 1 | 10.7 | 10.49 | 9.9 |
| 2 | 11.45 | 11 | 9.79 |
| 3 | 12.25 | 11.54 | 9.69 |
| 4 | 13.11 | 12.11 | 9.59 |
| 5 | 14.03 | 12.7 | 9.49 |

*The red line slopes gently down. Nothing visible happens. That is what makes it easy to miss.*

Try it: [FD Calculator](https://toolcloak.pages.dev/tools/fd-calculator). Enter the amount, rate and years to see the maturity value and interest.

## The same FD for different people

| Tax slab | Effective rate on a 7% FD | Real rate at 6% inflation |
| --- | --- | --- |
| No tax (income below limits) | 7.0% | about +0.9% |
| 20% slab | 5.6% | about −0.4% |
| 30% slab | 4.9% | about −1.0% |

*Real rate = (1 + effective rate) ÷ (1 + inflation) − 1. Slab names are simplified; the new and old regimes differ.*

## So are FDs bad?

No. They are the right tool for money you cannot afford to lose or will need soon: an emergency fund, a down payment due next year, fees due in two years. The mistake is using them for money you will not touch for 15 years and calling it safe.

**Two jobs for money**

| Needed in under 3 years | Needed in 10+ years |
| --- | --- |
| FD is fine | FD alone loses ground |
| Certainty matters more than growth | Tax and inflation compound against you |
| Inflation barely has time to bite | Consider growth assets for part of it |

*See [how big an emergency fund should be](https://toolcloak.pages.dev/articles/how-big-should-an-emergency-fund-be) for the case where FDs shine.*

> **A quick check:** Subtract your tax slab share from the FD rate, then subtract inflation. If the answer is near zero or below, the FD is preserving the number, not the value.

## Frequently asked questions

### Is FD interest taxable?

Yes. FD interest is added to your income and taxed at your slab rate, and banks may deduct TDS above certain limits.

### What is the real return on a 7% FD?

After 30 percent tax the effective rate is 4.9 percent. With 6 percent inflation, the real return is about −1 percent. For someone paying no tax it is about +0.9 percent.

### How much will ₹10 lakh become in a 7% FD in 5 years?

About ₹14.03 lakh with yearly compounding, before tax.

### Are FDs a bad investment?

They are good for short-term goals and emergency money, where safety matters most. For long goals, tax and inflation can leave little or no real gain.

### How do I calculate the real rate of return?

Divide one plus your after-tax rate by one plus the inflation rate, then subtract one. For 4.9 percent and 6 percent that gives about −1 percent.



---

# The Fibonacci sequence: from rabbits to the golden ratio, and why slow code loves it

> Add the last two numbers to get the next: 0, 1, 1, 2, 3, 5, 8, 13. That one rule hides a growth pattern, a famous ratio of about 1.618, a stair-climbing puzzle and the classic example of slow recursive code. This guide walks through all four with pictures and exact numbers.

Published 2026-10-08 by Toolcloak. 5 minute read. Original: https://toolcloak.pages.dev/articles/the-fibonacci-sequence-from-rabbits-to-the-golden-ratio-and-why-slow-code-loves-it
Related free tool: [Fibonacci Sequence Generator](https://toolcloak.pages.dev/tools/fibonacci-generator)

In 1202, Leonardo of Pisa, known as Fibonacci, posed a puzzle about rabbits: start with one pair, let every pair that is at least a month old produce a new pair each month, and count the pairs. The answer was a sequence that mathematicians, biologists, traders and programmers still talk about. The rule is so simple that a child can follow it, and the consequences are surprisingly deep.

> **In short**
> - Each term is the **sum of the two before it**: 0, 1, 1, 2, 3, 5, 8, 13, 21, 34, 55, 89…
> - The ratio of neighbouring terms settles at the **golden ratio, about 1.618**. Growth is exponential, roughly ×1.618 per step.
> - Climbing **n stairs** taking 1 or 2 steps at a time can be done in **F(n+1)** ways: 10 stairs give 89 routes.
> - Computing it with naive recursion is a classic trap: **F(30) needs 2.7 million calls**; a simple loop needs 30 steps.

## The rule and the first terms

Start with 0 and 1. Every next number is the sum of the previous two: 0 + 1 = 1, 1 + 1 = 2, 1 + 2 = 3, 2 + 3 = 5, and so on. Some books start at 1, 1 instead of 0, 1. The numbers after the start are identical.

| n | 0 | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 | 12 |
| --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- |
| F(n) | 0 | 1 | 1 | 2 | 3 | 5 | 8 | 13 | 21 | 34 | 55 | 89 | 144 |

*F(n) is the n-th term. F(10) = 55 and F(12) = 144.*

**How each new term is made**

1. **Take the last two.** For example 21 and 34.
2. **Add them.** 21 + 34 = 55.
3. **Slide along.** Now the last two are 34 and 55. Repeat.

Try it: [Fibonacci Sequence Generator](https://toolcloak.pages.dev/tools/fibonacci-generator). Choose how many terms you want and list them, or jump straight to the n-th number.

## The rabbit puzzle

Fibonacci’s puzzle assumes immortal rabbits that mate perfectly, so it is a model rather than zoology. Month 1: one pair, too young to breed. Month 2: still one pair, now mature. Month 3: the pair has a new pair, so two. Month 4: the old pair breeds again, the newer pair is maturing, so three. The pairs per month go 1, 1, 2, 3, 5, 8: exactly the sequence.

**Pairs of rabbits at the start of each month**

|  | Pairs |
| --- | --- |
| Month 1 | 1 |
| Month 2 | 1 |
| Month 3 | 2 |
| Month 4 | 3 |
| Month 5 | 5 |
| Month 6 | 8 |
| Month 7 | 13 |
| Month 8 | 21 |

*Each month the population equals last month’s pairs (who survive) plus the pairs born to everyone who was already mature two months ago.*

## It grows faster than it looks

The early terms are gentle, which hides how quickly the sequence takes off. Roughly every step multiplies the value by 1.618, so it is exponential growth, and the terms soon become huge.

**Some larger terms**

|  | Value | Note |
| --- | --- | --- |
| F(20) | 6,765 | Still small enough to read at a glance |
| F(30) | 832,040 | Over eight lakh |
| F(50) | 12,586,269,025 | More than 12 billion |

*F(90) is already 2,880,067,194,370,816,120, about 2.9 × 10¹⁸.*

**The first 16 terms (values)**

| Term number n | F(n) |
| --- | --- |
| 0 | 0 |
| 1 | 1 |
| 2 | 1 |
| 3 | 2 |
| 4 | 3 |
| 5 | 5 |
| 6 | 8 |
| 7 | 13 |
| 8 | 21 |
| 9 | 34 |
| 10 | 55 |
| 11 | 89 |
| 12 | 144 |
| 13 | 233 |
| 14 | 377 |
| 15 | 610 |

*The curve is flat for the first few terms and then bends steeply upward, the signature of exponential growth.*

If that shape looks familiar, it is the same curve you meet in [compound growth](https://toolcloak.pages.dev/articles/start-at-25-or-35-what-waiting-ten-years-really-costs) and the [rule of 72](https://toolcloak.pages.dev/articles/the-rule-of-72-how-long-money-takes-to-double). Fibonacci growth is like earning about 61.8% per step, with the nice twist that the “interest” comes from the sequence’s own past.

## The golden ratio appears

Divide each term by the one before it and watch what happens. The first ratios jump around (1, 2, 1.5, 1.67), then quickly calm down and approach **1.6180339…**, known as the **golden ratio** and written with the Greek letter phi (φ).

**Ratio of each term to the one before it: F(n+1) ÷ F(n)**

| n | F(n+1) ÷ F(n) | Golden ratio φ |
| --- | --- | --- |
| 1 | 1 | 1.618 |
| 2 | 2 | 1.618 |
| 3 | 1.5 | 1.618 |
| 4 | 1.6667 | 1.618 |
| 5 | 1.6 | 1.618 |
| 6 | 1.625 | 1.618 |
| 7 | 1.6154 | 1.618 |
| 8 | 1.619 | 1.618 |
| 9 | 1.6176 | 1.618 |
| 10 | 1.6182 | 1.618 |
| 11 | 1.618 | 1.618 |
| 12 | 1.6181 | 1.618 |

*The blue line oscillates around the amber line, getting closer each time. By n = 10 the ratio is correct to three decimals.*

> **φ = (1 + √5) ÷ 2 ≈ 1.6180339887:** It is the positive solution of x² = x + 1. That equation is the Fibonacci rule in disguise: ‘this term squared’ equals ‘this term plus one’ in ratio form, which is why the ratios of neighbours must settle on it.

Because of this, there is even a direct formula, Binet’s formula: F(n) = (φⁿ − (1 − φ)ⁿ) ÷ √5. Since (1 − φ) is about −0.618, its powers shrink to nothing, so F(n) is simply φⁿ ÷ √5 rounded to the nearest whole number. For n = 10: 1.618¹⁰ ÷ 2.236 ≈ 55.0.

> **Be careful with the nature claims:** Sunflower seed spirals and pinecone scales often have counts such as 34 and 55, or 21 and 34, which are Fibonacci numbers. That is a real pattern in many plants, not a law, and not every flower follows it. Many “golden ratio in art and the human face” claims are loose or untrue. Enjoy the real examples without stretching them.

## Climbing stairs and counting ways

Here is a puzzle that has nothing to do with rabbits. You can climb a staircase by taking either 1 or 2 steps at a time. How many different routes are there for n steps? The last move was either a single step (leaving n − 1 steps before it) or a double step (leaving n − 2). So ways(n) = ways(n − 1) + ways(n − 2): the Fibonacci rule again.

**Number of ways to climb n stairs with steps of 1 or 2**

|  | Routes |
| --- | --- |
| 5 stairs | 8 |
| 10 stairs | 89 |
| 15 stairs | 987 |
| 20 stairs | 10946 |

*The count for n stairs is F(n + 1). Twenty stairs already have 10,946 different routes.*

Another identity: the sum of the first n terms (from F(1)) is F(n + 2) − 1. Check with the first ten: 1 + 1 + 2 + 3 + 5 + 8 + 13 + 21 + 34 + 55 = 143, and F(12) − 1 = 144 − 1 = 143.

## Why programmers use it as a warning

Fibonacci is the standard example of recursion, a function that calls itself: fib(n) = fib(n − 1) + fib(n − 2). It is elegant, and it is slow, because the same values are recomputed again and again. fib(5) calls fib(4) and fib(3), and fib(4) calls fib(3) again, and so on.

**Two ways to compute F(30)**

| Naive recursion | A simple loop (or memoisation) |
| --- | --- |
| 2,692,537 calls | 30 steps |
| Recomputes the same terms repeatedly | Keeps only the last two values |
| Time roughly doubles with every extra step | Time grows in a straight line with n |
| F(40) needs about 331 million calls | Easily handles F(10,000) with big integers |

*The number of calls for naive recursion is 2 × F(n + 1) − 1, so the slowness is itself a Fibonacci sequence.*

**The fast way, step by step (for F(6))**

1. **Start.** a = 0, b = 1.
2. **Repeat 6 times.** Replace (a, b) with (b, a + b): (1,1), (1,2), (2,3), (3,5), (5,8), (8,13).
3. **Read the answer.** After 6 rounds, a = 8, which is F(6).

> **Watch your number types:** In JavaScript, ordinary numbers are exact only up to 9,007,199,254,740,991. F(78) = 8,944,394,323,791,464 still fits but F(79) = 14,472,334,024,676,221 does not, and results after that start to be wrong in the last digits. Use BigInt, or a big-integer type, for large terms. The same limit is why [Unix time](https://toolcloak.pages.dev/articles/unix-time-the-number-that-counts-seconds-since-1970) and IDs sometimes need 64 bits.

## Quick practice

1. What comes after 55, 89? **144** (55 + 89).
2. How many ways to climb 6 stairs with steps of 1 or 2? F(7) = **13**.
3. Estimate F(12) with φⁿ ÷ √5: 1.618¹² ≈ 321.99, divided by 2.236 gives 144.0. Rounded: **144**.
4. Is 1,000,000 a Fibonacci term? No. The terms near it are 832,040 (F(30)) and 1,346,269 (F(31)).

A rule you can say in one breath turns out to describe breeding rabbits, climbing stairs and a ratio that appears all over geometry, while also teaching every new programmer why smart algorithms matter. Try the generator above to see 50 terms in a flash, and notice how fast the numbers grow after the first few. For another kind of number pattern, read about [prime numbers](https://toolcloak.pages.dev/articles/prime-numbers-why-they-thin-out-and-why-your-bank-depends-on-them).

## Frequently asked questions

### What is the Fibonacci sequence?

A sequence in which each number is the sum of the two before it, starting 0, 1, 1, 2, 3, 5, 8, 13, 21, 34, 55.

### What is the golden ratio?

About 1.6180339887, the value that the ratio of consecutive Fibonacci numbers approaches. It equals (1 + √5) ÷ 2.

### What is the 10th Fibonacci number?

55, if you count F(0) = 0, F(1) = 1. Counting from F(1) = 1, F(2) = 1, the tenth term is also 55.

### Is there a formula for the n-th Fibonacci number?

Yes, Binet’s formula: F(n) = (φⁿ − (1 − φ)ⁿ) ÷ √5. In practice F(n) is φⁿ ÷ √5 rounded to the nearest integer.

### Why is recursive Fibonacci slow?

It recomputes the same values many times, so the number of calls grows exponentially. F(30) takes about 2.7 million calls, while a loop needs 30 steps.

### Do all flowers follow Fibonacci numbers?

No. Many sunflowers, pinecones and pineapples show Fibonacci counts in their spirals, but it is a common pattern, not a rule.



---

# The 25x rule: how much money is “enough” to stop working

> Multiply what you spend in a year by 25 and you get a first estimate of the savings that could pay for it. The rule is simple, famous and a little too optimistic for India. Here is how to use it honestly.

Published 2026-10-08 by Toolcloak. 3 minute read. Original: https://toolcloak.pages.dev/articles/the-25x-rule-how-much-is-enough-to-stop-working
Related free tool: [FIRE Calculator (Financial Independence Number)](https://toolcloak.pages.dev/tools/fire-calculator)

“How much is enough?” sounds like a question about money. It is really a question about spending. Once you know what a year of your life costs, a rule of thumb tells you roughly what needs to be saved. The rule is called the 25x rule, and the interesting part is what it hides.

> **In short**
> - FIRE number ≈ **yearly spending × 25**. That is the 4 percent rule, turned upside down.
> - At ₹50,000 a month, 25x is **₹1.5 crore**. A more cautious 3 percent rate needs **₹2 crore**.
> - The rule came from US data. With higher inflation, many Indian planners use 3 to 3.5 percent, not 4.
> - The number you choose matters less than **tracking what you really spend**.

## The rule in one line

> **Savings needed ≈ yearly spending ÷ withdrawal rate:** At a 4 percent withdrawal rate that is spending × 25. At 3 percent it is spending × 33. The idea is to take out that percentage in year one and then raise the amount each year with inflation.

The 4 percent rule comes from US studies of historical stock and bond returns. They asked how much a retiree could safely withdraw so that the money lasted about 30 years. 4 percent survived most historical periods. It is a rule of thumb, not a guarantee.

| Monthly spending | Yearly spending | At 4% (25x) | At 3% (33x) |
| --- | --- | --- | --- |
| ₹30,000 | ₹3.6 lakh | ₹90 lakh | ₹1.2 crore |
| ₹50,000 | ₹6 lakh | ₹1.5 crore | ₹2 crore |
| ₹75,000 | ₹9 lakh | ₹2.25 crore | ₹3 crore |
| ₹1,00,000 | ₹12 lakh | ₹3 crore | ₹4 crore |
| ₹1,50,000 | ₹18 lakh | ₹4.5 crore | ₹6 crore |

*Rounded. This is the corpus that must be invested, not your total net worth.*

Try it: [FIRE Calculator (Financial Independence Number)](https://toolcloak.pages.dev/tools/fire-calculator). Enter your spending, savings and expected returns. It estimates your number and the years to reach it.

## Why the withdrawal rate matters so much

Taking out a bit more each year looks small, but it drains the pot much faster. Here is a simple model: the money earns 8 percent a year, and you withdraw a fixed share of the starting amount, raised 6 percent each year for inflation.

**How many years the money lasts**

|  | Value |
| --- | --- |
| 3% | 51 |
| 4% | 33 |
| 5% | 24 |
| 6% | 19 |
| 8% | 14 |

*Illustrative model: 8% return, withdrawals rising 6% a year. Real returns are uneven, and bad early years hurt most. A 40-year retirement needs a lower rate than a 30-year one.*

Notice that the gap between 3 and 4 percent is not 33 percent. It is about 18 extra years. That is why early retirees use a lower rate than someone retiring at 65.

## What a corpus like this actually looks like

**Out of ₹100 of savings, what a 4 percent plan does in year one**

| Group | Count |
| --- | --- |
| Withdrawn to live on | ₹4 |
| Left invested | ₹96 |

*The other 96 stays invested and, if all goes well, grows enough to refill what was taken.*

## Four things the rule leaves out

- **Inflation is higher here.** Prices rising 6 percent a year double in about 12 years. See [what will ₹1 lakh buy in 2046](https://toolcloak.pages.dev/articles/what-will-one-lakh-buy-in-2046-inflation-in-pictures).
- **Healthcare is not smooth.** One hospital bill can dwarf a year of normal spending. Insurance and a separate buffer matter.
- **Taxes.** The rule works on what you spend, so the tax on withdrawals has to be added to your spending.
- **Your spending changes.** Children, parents and travel all move the number. Recalculate every year or two.

## A simple way to use it

1. **Track a real year of spending.** Not what you think you spend. What the bank statements say.
2. **Multiply by 25, then by 33.** The two numbers are your range.
3. **Divide by what you save a year** to get a rough number of years.
4. **Revisit it every year.** The number is a compass, not a destination.

## Frequently asked questions

### What is the 25x rule?

A rule of thumb that you need about 25 times your yearly expenses saved to retire. It is the 4 percent rule rewritten: a 4 percent withdrawal on 25 times your spending equals your spending.

### How much do I need to retire with ₹50,000 a month?

Spending ₹6 lakh a year gives about ₹1.5 crore at 4 percent and about ₹2 crore at 3 percent withdrawal.

### Is the 4 percent rule safe in India?

It was built on US data. Higher inflation in India means many planners use 3 to 3.5 percent for a long retirement. Treat 4 percent as the optimistic end of the range.

### Does the 25x rule include inflation?

Yes in the sense that it assumes you raise the withdrawal every year with inflation. If your own inflation is higher than the historical average used in the studies, you need a bigger corpus.

### Should I include my house in the FIRE number?

Only if you plan to sell it or rent it out. A home you live in does not pay for your expenses.



---

# GCD and LCM: Euclid’s 2,300-year-old shortcut, and when two buses meet again

> The greatest common divisor simplifies fractions, and the least common multiple tells you when repeating things line up again. This guide shows how to find both by hand, why Euclid’s method is so fast, and the one identity that links them.

Published 2026-10-08 by Toolcloak. 4 minute read. Original: https://toolcloak.pages.dev/articles/gcd-and-lcm-euclids-2300-year-old-shortcut-and-when-two-buses-meet-again
Related free tool: [GCD & LCM Calculator](https://toolcloak.pages.dev/tools/gcd-lcm-calculator)

Two small ideas hide behind most fraction homework and a surprising amount of real scheduling. The **greatest common divisor** (GCD, also called HCF) is the biggest number that divides two numbers exactly. The **least common multiple** (LCM) is the smallest number that both of them divide into. One tells you how to shrink, the other how to align.

> **In short**
> - GCD(48, 18) = **6** and LCM(48, 18) = **144**.
> - The two are linked: **GCD × LCM = a × b**. Check: 6 × 144 = 864 = 48 × 18.
> - Euclid’s method finds the GCD by repeated remainders: **GCD(a, b) = GCD(b, a mod b)** until the remainder is 0.
> - Use the **GCD to simplify fractions** (48/18 = 8/3) and the **LCM to line up repeating events** (every 12 and 18 minutes → every 36).

## Divisors you can see

A divisor of a number divides it with no remainder. The divisors of 48 are 1, 2, 3, 4, 6, 8, 12, 16, 24 and 48. The divisors of 18 are 1, 2, 3, 6, 9 and 18. The ones that appear in both lists are the **common divisors**: 1, 2, 3 and 6. The largest of them, 6, is the GCD.

**Divisors of 48 and 18, and what they share**

|  | Count |
| --- | --- |
| Divisors of 48 | 10 |
| Divisors of 18 | 6 |
| Common divisors | 4 (1, 2, 3, 6) |

*Listing divisors works for small numbers, but it gets slow quickly. The next two methods do not need long lists.*

**GCD as a tiling: a 48 × 18 rectangle in 6 × 6 squares**

| Group | Count |
| --- | --- |
| Squares of side 6 | 8 across × 3 down = 24 |

*The GCD is the side of the largest square that tiles a rectangle exactly. A 48 × 18 floor needs 24 square tiles of side 6.*

Try it: [GCD & LCM Calculator](https://toolcloak.pages.dev/tools/gcd-lcm-calculator). Enter two or more numbers and get the GCD, LCM and the working.

## Method 1: prime factors

Break each number into primes. 48 = 2⁴ × 3 and 18 = 2 × 3². For the GCD take the **smaller** power of each prime that both have. For the LCM take the **larger** power of every prime that appears.

| Prime | In 48 | In 18 | GCD uses (smaller) | LCM uses (larger) |
| --- | --- | --- | --- | --- |
| 2 | 2⁴ | 2¹ | 2¹ = 2 | 2⁴ = 16 |
| 3 | 3¹ | 3² | 3¹ = 3 | 3² = 9 |
| Product |  |  | 2 × 3 = 6 | 16 × 9 = 144 |

*The smaller powers give the GCD (6) and the larger powers give the LCM (144).*

> **GCD × LCM = a × b:** For any two positive numbers, the GCD times the LCM equals their product. For 48 and 18: 6 × 144 = 864 = 48 × 18. This is the easiest way to get the LCM after you have the GCD: LCM = a × b ÷ GCD.

## Method 2: Euclid’s algorithm

Written around 300 BC in Euclid’s Elements, this is one of the oldest algorithms still in everyday use. The idea: any number that divides both a and b also divides the remainder of a ÷ b. So replace the pair (a, b) by the smaller pair (b, remainder), and repeat until the remainder is zero. The last non-zero number is the GCD.

**GCD of 1071 and 462**

1. **1071 = 2 × 462 + 147.** Remainder 147, so now find GCD(462, 147).
2. **462 = 3 × 147 + 21.** Remainder 21, so now find GCD(147, 21).
3. **147 = 7 × 21 + 0.** Remainder 0. The last non-zero number, 21, is the GCD.

**Euclid shrinks the numbers fast (pair sizes)**

| Step | Amount |
| --- | --- |
| Start: 1071 | 1,071 |
| Next big number | → 462 |
| Then | → 147 |
| Then | → 21 |

*Each step replaces the bigger number by a much smaller remainder: 1071 → 462 → 147 → 21. Three divisions are enough.*

**How few steps Euclid needs**

|  | Value | Note |
| --- | --- | --- |
| GCD(123456, 7890) | 7 | Six-digit numbers, a handful of divisions |
| GCD(1000000, 999999) | 2 | Neighbours finish at once |
| GCD of two Fibonacci numbers | 28 | The slowest possible case |

*The worst cases are consecutive Fibonacci numbers, as proved by Gabriel Lamé in 1844. Even then, the step count grows with the number of digits, not the size of the numbers.*

See the [Fibonacci article](https://toolcloak.pages.dev/articles/the-fibonacci-sequence-from-rabbits-to-the-golden-ratio-and-why-slow-code-loves-it) for why those numbers are the slowest: each remainder is exactly the previous Fibonacci number, so the sequence shrinks as slowly as it can.

## The GCD in action: simplifying fractions

To reduce a fraction to its simplest form, divide the top and bottom by their GCD. For 48/18 the GCD is 6, so 48 ÷ 6 = 8 and 18 ÷ 6 = 3, giving 8/3. Using the GCD guarantees the answer is fully reduced in one step; dividing by 2 and then by 3 would take two.

**Reducing 48/18**

| Guessing small factors | Divide by the GCD |
| --- | --- |
| 2 steps | 1 step |
| 48/18 → 24/9 | GCD(48, 18) = 6 |
| 24/9 → 8/3 | 48 ÷ 6 = 8, 18 ÷ 6 = 3 |
| Easy to stop too early | Always fully simplified |

*Both reach 8/3. The GCD makes sure you never stop one step short.*

## The LCM in action: when things line up again

Two buses leave a stop together. One returns every 12 minutes, the other every 18. They next leave together at the LCM of 12 and 18, which is 36 minutes. The same idea schedules shifts, mixed repeating tasks and gear teeth that must return to the same pairing.

**Bus A (every 12 min) and bus B (every 18 min)**

| When | What |
| --- | --- |
| 0 min | Both leave together. The starting point. |
| 12 min | Bus A again. Bus B is still on its way. |
| 18 min | Bus B again. Bus A is waiting for its next trip. |
| 24 min | Bus A again. Still not together. |
| 36 min | Both leave together. LCM(12, 18) = 36. The pattern repeats from here. |

**LCM of different pairs: smaller cycles meet sooner**

|  | Minutes |
| --- | --- |
| 4 and 6 | 12 |
| 6 and 10 | 30 |
| 12 and 18 | 36 |
| 8 and 15 | 120 |
| 7 and 13 | 91 |

*When the two numbers share no common factor (7 and 13), the LCM is simply their product, 91. The more they share, the sooner they meet.*

> **Three or more numbers:** Apply the idea one pair at a time: LCM(4, 6, 10) = LCM(LCM(4, 6), 10) = LCM(12, 10) = 60. Three tasks that repeat every 4, 6 and 10 days all fall on the same day every 60 days. The same works for the GCD.

## Quick practice

1. GCD of 84 and 36? 84 = 2 × 36 + 12, 36 = 3 × 12, so **12**.
2. LCM of 84 and 36? 84 × 36 ÷ 12 = **252**.
3. Simplify 84/36. Divide both by 12 to get **7/3**.
4. Are 35 and 64 coprime? GCD(35, 64) = **1**, so yes. Their LCM is 2,240.

Two numbers with GCD 1 are called coprime. Coprime numbers matter in encryption, where choosing keys depends on them. Read about it in [prime numbers and RSA](https://toolcloak.pages.dev/articles/prime-numbers-why-they-thin-out-and-why-your-bank-depends-on-them).

## Frequently asked questions

### What is the GCD of 48 and 18?

6. It is the largest number that divides both 48 and 18 exactly.

### Are GCD and HCF the same thing?

Yes. HCF (highest common factor) and GCD (greatest common divisor) are different names for the same number.

### How do you find the LCM from the GCD?

LCM(a, b) = a × b ÷ GCD(a, b). For 48 and 18 that is 864 ÷ 6 = 144.

### What is the Euclidean algorithm?

A method that replaces (a, b) with (b, a mod b) over and over until the remainder is 0. The last non-zero value is the GCD.

### What if the GCD is 1?

Then the numbers are coprime (relatively prime). Their LCM is just their product.

### Where is the LCM used in everyday life?

Adding fractions with different denominators, scheduling repeating events, and gear or rotation problems all use the LCM.



---

# The gratuity formula (15/26): what your years of service are worth

> Gratuity is a lump sum paid when you leave a job after five years. The formula is last salary × 15 ÷ 26 × years. At ₹50,000 a month, ten years is about ₹2.88 lakh. Here is where the 26 comes from and what to watch.

Published 2026-10-08 by Toolcloak. 3 minute read. Original: https://toolcloak.pages.dev/articles/gratuity-formula-15-by-26-what-your-years-of-service-are-worth
Related free tool: [Gratuity Calculator](https://toolcloak.pages.dev/tools/gratuity-calculator)

Gratuity is the one part of your pay that grows with loyalty. It is a lump sum your employer pays when you leave, and it depends on only two numbers: your last drawn salary and how long you stayed.

> **In short**
> - Gratuity = **last drawn salary × 15 ÷ 26 × years of service**.
> - Last drawn salary here means **basic plus dearness allowance**, not your full CTC.
> - At ₹50,000 basic, **10 years pays about ₹2.88 lakh**, and 20 years about ₹5.77 lakh.
> - Eligibility normally starts after **5 years** of continuous service. Check the latest rules for exceptions and limits.

## The formula in plain words

> **Gratuity = (last drawn basic + DA) × 15 ÷ 26 × completed years:** The 15 is days of pay for each year of service. The 26 is the number of working days in a month. So you get 15 days of salary for each year you worked.

A part year of more than six months is rounded up to the next full year. So 7 years and 8 months counts as 8 years, while 7 years and 5 months counts as 7.

**Gratuity at a basic salary of ₹50,000 a month**

|  | Value |
| --- | --- |
| 5 yrs | ₹1.44 L |
| 10 yrs | ₹2.88 L |
| 15 yrs | ₹4.33 L |
| 20 yrs | ₹5.77 L |
| 30 yrs | ₹8.65 L |

*Each year adds about ₹28,846. The amount scales straight with years and with salary.*

Try it: [Gratuity Calculator](https://toolcloak.pages.dev/tools/gratuity-calculator). Enter your last basic plus DA and your years of service to see the gratuity.

## Worked example

**₹50,000 basic, 10 years**

| Step | Value |
| --- | --- |
| Last drawn basic + DA | ₹50,000 |
| × 15 ÷ 26 (about 0.577 of a month's pay per year) | ₹28,846 per year |
| × 10 years | ₹2,88,462 |

*The bars are not to scale. The point is the three steps: salary, per-year amount, total.*

## What people get wrong

- **Using CTC.** Only basic and DA count. A big CTC with a small basic means a small gratuity.
- **Counting years that were not continuous.** A real break in service can reset the clock.
- **Assuming everyone gets it.** The formula above applies to employees covered by the Payment of Gratuity Act. Employers not covered by the Act may use their own terms.
- **Forgetting the five-year rule.** There are exceptions, such as death or disablement, but resigning at 4 years 10 months normally means no gratuity.
- **Forgetting the cap and tax rules.** There is a legal ceiling on gratuity and a tax-free limit. Both are set by the government and have been revised before, so check the current figures.

> **Plan around it:** If you are close to five years, compare the gratuity you would gain by staying with what a better offer pays. At a ₹50,000 basic, the five-year mark is worth about ₹1.44 lakh.

Gratuity is one piece of your retirement and exit money. Compare it with what you build yourself in a [monthly SIP](https://toolcloak.pages.dev/articles/5000-rupees-a-month-for-20-years-what-a-sip-calculator-can-tell-you), where years of habit do the same quiet work.

## Frequently asked questions

### What is the gratuity formula?

Last drawn salary (basic plus dearness allowance) × 15 ÷ 26 × number of completed years of service. The 15 is the days of pay per year and 26 is the working days in a month.

### How much gratuity will I get for 10 years at ₹50,000 basic?

About ₹2,88,462.

### Is gratuity payable after 5 years?

Normally yes, after five years of continuous service. There are exceptions, such as death or disablement. Check the latest rules.

### Does CTC count for gratuity?

No. Only the last drawn basic salary plus dearness allowance is used.

### How is a part year counted?

A part year of more than six months is rounded up to a full year. Six months or less is not counted.



---

# Half-life: why things fade by half, again and again, and never quite reach zero

> Half-life is the time it takes for half of something to disappear, and it applies to radioactive atoms, caffeine in your body and the age of ancient bones. This guide builds the formula from simple halving, with pictures and exact worked examples.

Published 2026-10-08 by Toolcloak. 4 minute read. Original: https://toolcloak.pages.dev/articles/half-life-why-things-fade-by-half-again-and-again-and-never-quite-reach-zero
Related free tool: [Half-Life Calculator (Radioactive Decay)](https://toolcloak.pages.dev/tools/half-life-calculator)

Imagine a cup of coffee with 200 mg of caffeine. Five hours later, about 100 mg is left. After another five hours, about 50 mg. The amount does not fall by the same number each time. It falls by the same **fraction**: half. That fixed time for halving is the half-life, and the pattern shows up in physics, medicine, chemistry and even archaeology.

> **In short**
> - After one half-life **50%** remains, after two **25%**, after three **12.5%**. After **7 half-lives** under 1% is left.
> - Formula: **N = N₀ × (½)^(t ÷ T)**, where T is the half-life and t the time passed.
> - Caffeine with a 5-hour half-life: a 200 mg coffee leaves about **29 mg** after 14 hours.
> - To find the **time** from a known fraction, use **t = T × log₂(N₀ ÷ N)**. 25% left means exactly 2 half-lives.

## Halving again and again

Start with 100 units of something that decays with a half-life of one hour. After the first hour 50 remain. After the second, half of those 50: 25. It never reaches exactly zero, though after enough halvings what is left is too small to matter.

**What is left after each half-life (% of the start)**

|  | Value |
| --- | --- |
| 1 half-life | 50% |
| 2 | 25% |
| 3 | 12.5% |
| 4 | 6.25% |
| 5 | 3.13% |
| 6 | 1.56% |
| 7 | 0.78% |

*Each bar is half of the one before it: 50, 25, 12.5, 6.25, 3.13, 1.56, 0.78.*

**Halving a pile of 64 coins**

| Step | Value |
| --- | --- |
| Start | 64 |
| After 1 half-life | 32 |
| After 2 | 16 |
| After 3 | 8 |
| After 4 | 4 |

*After 6 half-lives a single coin is left. Real atoms decay at random, so the half-life describes a large group, not any single atom.*

Try it: [Half-Life Calculator (Radioactive Decay)](https://toolcloak.pages.dev/tools/half-life-calculator). Enter the half-life and the time passed, or the fraction left, and get the missing value.

## The formula

> **N = N₀ × (1/2)^(t ÷ T):** N₀ is the starting amount, N the amount now, T the half-life and t the elapsed time, both in the same unit. The exponent t ÷ T counts how many half-lives have passed, and it need not be a whole number.

For a non-whole number of half-lives, use a calculator. After 1.5 half-lives, 0.5^1.5 = 0.354, so about 35% remains: more than the 25% after 2 and less than the 50% after 1.

**How much iodine-131 (half-life about 8 days) is left after 20 days?**

1. **Count the half-lives.** 20 ÷ 8 = 2.5 half-lives.
2. **Raise ½ to that power.** 0.5^2.5 = 0.177.
3. **Read the answer.** About 17.7% of the original amount remains.

**Iodine-131 over 64 days (% remaining, half-life 8 days)**

| Days | Remaining |
| --- | --- |
| 0 | 100 |
| 8 | 50 |
| 16 | 25 |
| 24 | 12.5 |
| 32 | 6.25 |
| 40 | 3.13 |
| 48 | 1.56 |
| 56 | 0.78 |
| 64 | 0.39 |

*Eight days halves the amount to 50%. By day 64 (eight half-lives) only about 0.4% is left.*

## Real half-lives, from seconds to billions of years

Half-lives span an enormous range. Some atoms last a fraction of a second, while others outlast the Earth. A shorter half-life means faster decay, and a more intense radiation level for the same amount.

| Substance | Half-life (approx.) | Where it matters |
| --- | --- | --- |
| Technetium-99m | 6 hours | Medical scans, used because it fades quickly |
| Iodine-131 | 8 days | Thyroid treatment, fallout monitoring |
| Caffeine (healthy adult) | ≈ 5 hours (3 to 7) | Why an evening coffee affects sleep |
| Caesium-137 | ≈ 30 years | Long-lived fallout, industrial sources |
| Carbon-14 | 5,730 years | Dating organic remains |
| Uranium-238 | 4.5 billion years | Age of rocks and the Earth |

*The values are standard rounded figures. For caffeine, the half-life varies with genes, pregnancy, smoking and medication.*

**How a half-life shapes what you do with a substance**

| When | What |
| --- | --- |
| Hours | Technetium-99m. Gone in a day or two, so a patient’s exposure is brief. |
| Days | Iodine-131. Weeks of caution, then it is mostly gone. |
| Decades | Caesium-137. Takes generations to fade, so sites stay monitored. |
| Millennia | Carbon-14. Useful as a clock for ancient objects. |

## A coffee at 8 a.m.

Assume one 200 mg coffee and a 5-hour half-life. Every five hours the amount in your body halves. By late evening there is still a measurable amount.

**Caffeine left in the body after a 200 mg coffee (mg)**

| Hours after drinking | Caffeine remaining |
| --- | --- |
| 0h | 200 |
| 2h | 151.57 |
| 4h | 114.87 |
| 6h | 87.06 |
| 8h | 65.98 |
| 10h | 50 |
| 12h | 37.89 |
| 14h | 28.72 |
| 16h | 21.76 |
| 18h | 16.49 |
| 20h | 12.5 |
| 22h | 9.47 |
| 24h | 7.18 |

*At 8 a.m. the cup starts at 200 mg; by 10 p.m. (14 hours later) about 29 mg remains. Taking a second cup in the afternoon adds to what is already there.*

> **Doses add up before they fade:** Because the earlier dose has not finished decaying, a second drink stacks on top. A 100 mg coffee at 3 p.m. still leaves about 50 mg at 8 p.m., and 25 mg at 1 a.m. For sleep, people often choose a cut-off several half-lives before bed.

## Dating ancient things with carbon-14

Living things keep a steady proportion of carbon-14 while they are alive. After death the proportion falls with a half-life of 5,730 years. Measure the fraction left and you can work backwards to the age.

**Age of a sample by the share of carbon-14 left (years)**

|  | Age |
| --- | --- |
| 50% left | 5,730 yr |
| 25% left | 11,460 yr |
| 12.5% left | 17,190 yr |
| 10% left | 19,035 yr |
| 6.25% left | 22,920 yr |

*Use t = T × log₂(N₀ ÷ N). For 10% left: 5,730 × log₂(10) ≈ 19,000 years. Beyond about 50,000 years too little carbon-14 remains to measure.*

> **Other ways to say the same thing:** A decay constant λ = ln 2 ÷ T ≈ 0.693 ÷ T gives N = N₀ × e^(−λt). The mean lifetime is T ÷ ln 2 ≈ 1.44 × T. All three describe the same curve. It is the mirror image of compound growth: see [the rule of 72](https://toolcloak.pages.dev/articles/the-rule-of-72-how-long-money-takes-to-double) for doubling.

## Quick practice

1. A sample has a half-life of 10 days. How much of 80 g is left after 30 days? Three half-lives: 80 ÷ 8 = **10 g**.
2. How many half-lives until under 1%? **7** (0.78% remain).
3. Caffeine at 5-hour half-life: 160 mg after 10 hours? 160 ÷ 4 = **40 mg**.
4. A bone has 25% of its carbon-14. Age? Two half-lives: **11,460 years**.

Whenever something shrinks by a fixed fraction per unit of time, the half-life is the natural way to talk about it: discounted values, drug levels, even the “half-life” of news. Try different numbers in the calculator to feel how fast halving adds up.

## Frequently asked questions

### What is half-life?

The time it takes for half of a quantity to decay or disappear. After one half-life 50% is left, after two 25%, after three 12.5%.

### What is the half-life formula?

N = N₀ × (1/2)^(t ÷ T), where T is the half-life and t the elapsed time in the same unit.

### Does a substance ever fully disappear?

Mathematically no, since halving never reaches zero. In practice the amount becomes too small to matter, usually within 7 to 10 half-lives.

### What is the half-life of caffeine?

About 5 hours in healthy adults, with a typical range of 3 to 7 hours. Pregnancy, medication and genes change it.

### How does carbon dating work?

It measures the share of carbon-14 left in organic material and uses the 5,730-year half-life to calculate how long ago the organism died.

### Is a shorter half-life more dangerous?

A shorter half-life means more decays per second for the same amount, so higher intensity but a shorter time of exposure. Both matter.



---

# What a hash is, and why changing one letter changes everything

> Feed a hash function “hello” or the whole of Wikipedia and you get back a short fixed-size fingerprint. Change a single letter and the fingerprint looks completely different. That odd behaviour is what makes hashes useful.

Published 2026-10-08 by Toolcloak. 3 minute read. Original: https://toolcloak.pages.dev/articles/what-a-hash-is-and-why-changing-one-letter-changes-everything
Related free tool: [Hash Generator](https://toolcloak.pages.dev/tools/hash-generator)

Hash the word `hello` with SHA-256 and you get `2cf24dba5fb0a30e26e83b2ac5b9e29e1b161e5c1fa7425e73043362938b9824`. Change the last letter to make `hellp` and you get `fdd7585e08c4e2af…`. Not similar. Not nearby. Completely different. That is not a bug. It is the point.

> **In short**
> - A hash is a fixed-size fingerprint: same input, same output, and no way back.
> - Change one letter and about half of the 256 bits flip. That is the avalanche effect.
> - Use SHA-256 for files and data. Never use plain SHA-256 to store passwords.

A hash function takes any amount of data and boils it down to a fixed-length fingerprint. Same input always gives the same fingerprint. A different input, even by one character, gives a wildly different one. Once you accept those two rules, a lot of everyday technology makes sense: file checks, version control, password storage and more.

## Three promises a good hash makes

1. **Same input, same output.** Hash the same text a million times and you get the same fingerprint a million times, on any computer.
2. **Fixed size.** A single letter or a ten-gigabyte video both give you, for SHA-256, exactly 64 hexadecimal characters.
3. **One way.** You can go from the text to the fingerprint easily. Going back from the fingerprint to the text is, for practical purposes, impossible.

**SHA-256 in numbers**

|  | Value | Note |
| --- | --- | --- |
| Size of the fingerprint | 256 bits | That is 32 bytes |
| Hex characters shown | 64 | Four bits per character |
| Possible fingerprints | 2²⁵⁶ | Roughly 10⁷⁷, a 78-digit number |

*Whatever you feed in, the answer has exactly this shape.*

Try it: [Hash Generator](https://toolcloak.pages.dev/tools/hash-generator). Type or paste text and see its SHA-1, SHA-256 and SHA-512 fingerprints, calculated locally in your browser.

## One letter, a completely different fingerprint

Look at three tiny inputs that differ by one character, hashed with SHA-256. Only the beginning of each fingerprint is shown, but you will see that they have nothing in common.

| Input | SHA-256 (first 16 characters) |
| --- | --- |
| `hello` | `2cf24dba5fb0a30e` |
| `hellp` | `fdd7585e08c4e2af` |
| `Hello` | `185f8db32271fe25` |

*Changing e→p, or h→H, produces fingerprints with no resemblance at all. There is no “close” in the output, even when the input is very close.*

This is called the **avalanche effect**. If you compare the full 256 bits of `hello` against `hellp`, **131 of them differ**. Comparing `hello` and `Hello`, 125 differ. That is about half, which is exactly what you would expect from two unrelated random strings.

**How many of the 256 bits change when ‘hello’ becomes ‘hellp’**

|  | Share |
| --- | --- |
| Bits that changed | 131 |
| Bits that stayed | 125 |

*About half the bits flip, no matter how small the change. That is what stops anyone reading a pattern in the output.*

## What hashes are used for

- **Checking a download.** A site publishes the SHA-256 of a file. You hash your copy. If the fingerprints match, the file was not corrupted or tampered with on the way.
- **Finding duplicates.** Two files with the same hash are, for all practical purposes, identical, so you can find copies without comparing them byte by byte.
- **Version control.** Git identifies every commit by a hash of its contents, so history cannot be quietly rewritten without every later fingerprint changing.
- **Digital signatures and certificates.** Signing a long document is slow, so you hash it and sign the short fingerprint instead.
- **Storing passwords.** A site should keep a hash of your password, not the password. But read the warning below, because the type of hash matters a lot.

## Hashing is not encryption

Encryption is meant to be reversed with a key. A hash is meant to **not** be reversed at all, and there is no key. You cannot “decrypt” a SHA-256 fingerprint. What people do instead is **guess**: hash many candidate inputs and see which one matches.

> **Plain SHA-256 is the wrong tool for passwords:** SHA-256 is designed to be fast, and fast is exactly what an attacker wants when guessing billions of passwords. Passwords should be stored with a deliberately slow, salted method such as bcrypt, scrypt or Argon2. Use a general hash for files and data, not for passwords.

## Which algorithm, and which are retired

**Length of the output in hex characters**

|  | Characters |
| --- | --- |
| MD5 (128 bits) | 32 |
| SHA-1 (160 bits) | 40 |
| SHA-256 (256 bits) | 64 |
| SHA-512 (512 bits) | 128 |

*Longer output means more room for the fingerprint to be unique, but length is not the only thing that decides safety.*

- **MD5** and **SHA-1** have known collision attacks: researchers have built two different inputs that give the same fingerprint. Do not use them where an attacker is involved, such as signatures. They are still fine for catching accidental corruption.
- **SHA-256** and **SHA-512** are the sensible defaults today for integrity checks and signatures.
- **Larger is not always better.** SHA-512 is not “safer” than SHA-256 for most uses. It is just longer, and sometimes faster on 64-bit machines.

## A small trap: invisible differences

If you hash text and your friend hashes the “same” text but gets a different answer, look for invisible differences: a trailing space, a newline at the end, a different capital letter, or text saved in a different encoding. A hash sees every byte. Anything a human would ignore, the function counts.

## Frequently asked questions

### What is a hash?

A fixed-length fingerprint of some data produced by a hash function. The same input always gives the same output, and a tiny change to the input gives a completely different output.

### Is hashing the same as encryption?

No. Encryption can be reversed with a key. A hash is one-way and has no key, so you cannot decrypt it. You can only guess inputs and compare their hashes.

### Can a SHA-256 hash be reversed?

Not by calculation. People recover short or common inputs by guessing and hashing many candidates, which is why weak passwords are still at risk even when hashed.

### Should I use SHA-256 to store passwords?

No. It is too fast, which helps attackers. Use a slow, salted password hash such as bcrypt, scrypt or Argon2 instead.

### Why do MD5 and SHA-1 count as broken?

Because researchers have shown practical ways to create two different inputs with the same hash. That is unsafe for signatures and security checks, though fine for catching accidental file corruption.



---

# Heart rate zones explained: what each of the five zones is for

> Training zones split your effort into five bands, from easy walking to an all-out sprint. For a 30-year-old with an estimated maximum of 190 beats a minute, the bands run from 95 to 190. This guide shows how the zones are worked out, what each one feels like and how to spread your week across them.

Published 2026-10-08 by Toolcloak. 4 minute read. Original: https://toolcloak.pages.dev/articles/heart-rate-zones-explained-what-each-of-the-five-zones-is-for
Related free tool: [Heart Rate Zone Calculator](https://toolcloak.pages.dev/tools/heart-rate-zone-calculator)

A fitness tracker shows a number, but a number alone says little. Zones turn it into a plan: this much of your week should feel easy, this much hard, and almost none should be flat-out. It all starts with one estimate, your maximum heart rate.

> **In short**
> - The common estimate is **220 − age** for maximum heart rate. A 30-year-old gets **190 beats per minute (bpm)**.
> - The five zones are **50–60%, 60–70%, 70–80%, 80–90% and 90–100%** of that maximum. At 190, that is 95–114, 114–133, 133–152, 152–171 and 171–190 bpm.
> - The formula is an average and can be **off by 10 bpm or more** for an individual. Use the talk test beside it.
> - This is general fitness information, not medical advice. Check with a doctor before hard training if you have a heart condition or other health concerns.

## Step one: estimate your maximum

Your maximum heart rate is the highest rate your heart can reach during all-out effort. The only exact way to find it is a supervised test, so most people use a formula. The classic one is 220 minus your age. A newer one, often credited to Tanaka and colleagues, is 208 minus 0.7 times your age.

**Estimated maximum heart rate by age (bpm)**

|  | 220 − age | 208 − 0.7 × age |
| --- | --- | --- |
| Age 20 | 200 | 194 |
| Age 30 | 190 | 187 |
| Age 40 | 180 | 180 |
| Age 50 | 170 | 173 |
| Age 60 | 160 | 166 |

*The two formulas agree around age 40. The older formula gives a lower number for people over 40 and a higher one for people under 40.*

> **It is an estimate, not a measurement:** Two people of the same age can differ by 10 to 20 bpm in real maximum rate. Fitness level, genes, medicines and heat all play a part. If your zones feel wrong, trust how you feel and, if needed, ask a coach or doctor about a proper test.

## Step two: the five zones

Each zone is a band of percentages of your maximum. Higher zones feel harder and can be held for a shorter time.

**Zones for a 30-year-old with a maximum of 190 bpm**

| Zone | Range |
| --- | --- |
| Zone 1 · very light | 95–114 |
| Zone 2 · light | 114–133 |
| Zone 3 · moderate | 133–152 |
| Zone 4 · hard | 152–171 |
| Zone 5 · maximum | 171–190 |

*The needle sits at 140 bpm, which is in zone 3. Your own bands move with your maximum.*

Try it: [Heart Rate Zone Calculator](https://toolcloak.pages.dev/tools/heart-rate-zone-calculator). Enter your age (or a measured maximum) to get all five zones in bpm.

## What each zone feels like

**From easy to all-out**

| When | What |
| --- | --- |
| Zone 1 · 50–60% | Warm-up and recovery. Easy walking. You can chat in full sentences and sing without trouble. |
| Zone 2 · 60–70% | Base endurance. Comfortable jogging or cycling. You can still talk in full sentences. Most easy training lives here. |
| Zone 3 · 70–80% | Steady, moderate effort. Breathing is deeper. You can speak in short phrases. Useful, but easy to overdo. |
| Zone 4 · 80–90% | Hard, threshold effort. Only a few words at a time. Held for minutes, not hours. |
| Zone 5 · 90–100% | All-out intervals. You cannot talk. Held for seconds to a couple of minutes. |

A simple check is the **talk test**. If you can chat comfortably you are probably in zone 1 or 2. If you can say only short phrases you are likely in zone 3 or 4. If you cannot speak, you are in zone 5.

## Zones at other ages

Because the maximum falls with age, every zone moves down too. The table uses 220 minus age.

**Zone ranges by age (bpm, using 220 − age)**

| Age | Zone 1 | Zone 2 | Zone 3 | Zone 4 | Zone 5 |
| --- | --- | --- | --- | --- | --- |
| 20 | 100–120 | 120–140 | 140–160 | 160–180 | 180–200 |
| 30 | 95–114 | 114–133 | 133–152 | 152–171 | 171–190 |
| 40 | 90–108 | 108–126 | 126–144 | 144–162 | 162–180 |
| 50 | 85–102 | 102–119 | 119–136 | 136–153 | 153–170 |
| 60 | 80–96 | 96–112 | 112–128 | 128–144 | 144–160 |

## How to spread your week

Many coaches suggest that most training should be easy and a smaller share hard. A common guideline is about 80% easy (zones 1–2) and 20% hard (zones 4–5), with the middle zone used sparingly. It is a guideline, not a rule, and beginners often need even more easy time.

**A sample week of five hours of training**

|  | Zone 1–2 easy | Zone 3 moderate | Zone 4–5 hard |
| --- | --- | --- | --- |
| Beginner | 90% | 5 | 5 |
| Balanced plan | 80% | 5 | 15% |
| Common trap | 30% | 55% | 15% |

*The trap is making every session medium-hard. You get tired, but you never fully recover for the hard sessions or build a big easy base.*

## The “fat-burn zone” myth

Gym machines often label zone 2 as the fat-burn zone. It is true that a larger share of the calories burned come from fat at lower effort. But higher effort burns more calories per minute overall, so total fat used over a week depends far more on how much you do and what you eat than on staying in one zone.

## Practical tips

- **Measure resting heart rate in the morning.** A rate that is several beats above normal for a few days can mean you need rest.
- **Allow for heat, caffeine and sleep loss.** All can push your rate up for the same effort. Go by feel on those days.
- **Wrist trackers can lag.** On intervals they often show the number late. A chest strap is more accurate if it matters to you.
- **Sleep is part of training.** Our [sleep cycle guide](https://toolcloak.pages.dev/articles/sleep-in-90-minute-cycles-when-to-go-to-bed) helps you plan when to go to bed.
- **Stop and get help** if you feel chest pain, dizziness or unusual breathlessness.

> **Start with zone 2:** If you are new to exercise, spend several weeks mostly in zones 1 and 2. Add short zone 4 efforts once easy sessions feel easy. Check with your doctor first if you are unsure.

## Frequently asked questions

### How do I calculate my maximum heart rate?

A common estimate is 220 minus your age. Another is 208 minus 0.7 times your age. Both are averages and can be off by 10 bpm or more for one person.

### What are the five heart rate zones?

They are 50–60%, 60–70%, 70–80%, 80–90% and 90–100% of your maximum heart rate, from very light to maximum effort.

### Which zone is best for fat loss?

No single zone is best. Zone 2 uses a higher share of fat for fuel, but total calories burned over the week matter more than the zone.

### Is 220 minus age accurate?

It is a rough average. Individual maximum heart rates vary widely, so use the talk test or a supervised test if you need a precise number.

### What is a normal resting heart rate?

For most adults, a typical resting heart rate is about 60 to 100 beats a minute, and well-trained people are often lower. See a doctor if yours is unusually high or low or comes with symptoms.



---

# HRA exemption: the least of three rules, with a worked example

> The tax-free part of your House Rent Allowance is the smallest of three numbers: the HRA you actually get, a share of your basic salary, and the rent you pay minus 10% of your basic. With ₹40,000 basic, ₹20,000 HRA and ₹18,000 rent in a metro, ₹14,000 a month is exempt. Here is why, and how to improve it.

Published 2026-10-08 by Toolcloak. 4 minute read. Original: https://toolcloak.pages.dev/articles/hra-exemption-the-least-of-three-rules-with-a-worked-example
Related free tool: [HRA Exemption Calculator](https://toolcloak.pages.dev/tools/hra-exemption-calculator)

Your payslip shows HRA as one line, but the tax office does not treat the whole line as tax-free. It runs three tests and lets you keep the lowest result. Once you see which test is holding you back, you know exactly what to change.

> **In short**
> - Exempt HRA each month is the **least of three**: (1) HRA received, (2) **50% of basic + DA in a metro (40% elsewhere)**, (3) **rent paid minus 10% of basic + DA**.
> - With ₹40,000 basic, ₹20,000 HRA and ₹18,000 rent in a metro, the three numbers are ₹20,000, ₹20,000 and **₹14,000**. The exemption is **₹14,000 a month**.
> - Everything above the exemption is taxed as salary. Here that is ₹6,000 a month, ₹72,000 a year.
> - This exemption is generally available **only if you use the old tax regime** and actually pay rent. Rules and city lists are updated from time to time, so confirm the current version.

## The three tests

1. **The HRA your employer actually pays you.** If your payslip says ₹20,000, the exemption can never be more than that.
2. **A share of your salary.** 50% of basic plus dearness allowance if you live in a metro, 40% if you do not. “Salary” here means basic + DA, not your CTC.
3. **Rent paid minus 10% of basic + DA.** The first tenth of your salary is treated as the part you would spend on housing anyway, so only the rent above it counts.

The lowest of the three is your tax-free HRA. The rest of the HRA is added to your taxable income.

**The three numbers for ₹40,000 basic, ₹20,000 HRA, ₹18,000 rent (metro)**

| Tests that were not the limit | The test that decides |
| --- | --- |
| ₹20,000 | ₹14,000 |
| Test 1: HRA received is ₹20,000 | Test 3: ₹18,000 rent − ₹4,000 (10% of ₹40,000) |
| Test 2: 50% of ₹40,000 is ₹20,000 | This is the lowest, so ₹14,000 is exempt |
| Both are higher than test 3 | ₹6,000 of your HRA is taxable |

**What happens to ₹20,000 of monthly HRA**

| Step | Amount |
| --- | --- |
| HRA received | ₹20,000 |
| Taxable part | −₹6,000 |
| Exempt HRA | ₹14,000 |

*Over a year the exemption is ₹1,68,000. At a 20% slab that saves roughly ₹35,000 in tax, including 4% cess.*

Try it: [HRA Exemption Calculator](https://toolcloak.pages.dev/tools/hra-exemption-calculator). Enter your basic, DA, HRA and rent to see the exempt and taxable parts.

## How rent changes the answer

Because test 3 depends on your rent, the exemption rises as your rent rises, until test 1 or 2 takes over. Keep the same salary and HRA, and only change the rent:

**Monthly HRA split by rent paid (₹40,000 basic, ₹20,000 HRA, metro)**

|  | Value |
| --- | --- |
| Rent ₹10k | ₹6,000 exempt |
| Rent ₹14k | ₹10,000 exempt |
| Rent ₹18k | ₹14,000 exempt |
| Rent ₹22k | ₹18,000 exempt |
| Rent ₹26k | ₹20,000 exempt |

*Each bar is the ₹20,000 HRA. Below ₹24,000 of rent, test 3 is the limit. From ₹24,000 upward, test 1 and 2 cap the exemption at ₹20,000 and extra rent no longer helps.*

## Metro versus non-metro

Test 2 changes only if the 50% share changes to 40%. In our example, a non-metro city turns test 2 into ₹16,000 (40% of ₹40,000). That is still above test 3, so the exemption stays at ₹14,000. The city matters only when the 40% or 50% test is the lowest of the three.

> **Which cities count as metro?:** Delhi, Mumbai, Kolkata and Chennai have long been the metro cities for this rule. The list has been the subject of changes and proposals, so check the latest rules for your city before you file.

**Annual view of the same example**

|  | Value | Note |
| --- | --- | --- |
| HRA received in the year | ₹2.40 L |  |
| Exempt HRA | ₹1.68 L | Not taxed |
| Taxable HRA | ₹72,000 | Added to income |

## Five things people get wrong

- **Using CTC for the 50% test.** Only basic plus DA counts. If your basic is a small part of CTC, the exemption can be small however large the HRA looks.
- **Claiming HRA without paying rent.** You must actually pay rent. If it is above ₹1 lakh a year, your landlord’s PAN is normally needed, so keep it handy.
- **Assuming it works in the new regime.** The exemption is generally an old-regime benefit. Compare both regimes before you choose, because dropping HRA can still be worthwhile if your other deductions are small.
- **Forgetting to submit proof.** If you do not give rent receipts to your employer, tax is deducted on the full HRA during the year and you reclaim it only when you file your return.
- **Paying rent to parents without a record.** It is allowed when the rent is real and the parent reports it as income. Use a proper agreement and bank transfers.

## A checklist before you submit proofs

1. Find your monthly basic and DA on the payslip, and the HRA component.
2. Write down the real rent you pay for each month of the year.
3. Work out the three tests and take the lowest.
4. Collect rent receipts. You can make them with the [rent receipt generator](https://toolcloak.pages.dev/tools/rent-receipt-generator) if your landlord does not give any.
5. Check the old-versus-new regime choice with your employer’s tax tool or a tax adviser.

> **Ask about your salary structure:** Because test 2 and test 3 both use basic pay, a higher basic with a lower special allowance can raise the exemption. Many employers allow a small change at the start of the year. It is worth asking, but it also raises your PF contribution, so compare the take-home as well.

## Frequently asked questions

### How is HRA exemption calculated?

It is the least of three: the HRA you receive, 50% of basic + DA for metros (40% for others), and the rent you pay minus 10% of basic + DA.

### Can I claim HRA if I live in my parents’ house?

Yes, if you actually pay them rent, they receive it as income, and you keep records such as a rent agreement and bank transfers.

### Do I need rent receipts?

Employers usually ask for receipts as proof. If your annual rent is above ₹1 lakh, the landlord’s PAN is normally required as well.

### Is HRA exempt under the new tax regime?

The HRA exemption is generally available only under the old regime. Compare both regimes with your own numbers before choosing.

### What happens to the part of HRA that is not exempt?

It is added to your taxable salary and taxed at your slab rate.

### Does DA count for the HRA calculation?

Yes, basic salary plus dearness allowance that forms part of retirement benefits is used. Bonuses and special allowances are not.



---

# HTTP status codes: the five families, and the dozen you will actually meet

> Nobody needs to memorise sixty codes. Learn the five families by their first digit and a dozen common ones, and you can read almost any error message, API response or browser warning.

Published 2026-10-08 by Toolcloak. 3 minute read. Original: https://toolcloak.pages.dev/articles/http-status-codes-the-five-families-you-need-to-know
Related free tool: [HTTP Status Code Lookup](https://toolcloak.pages.dev/tools/http-status-codes)

Every time a browser or app asks a server for something, the server answers with a three-digit number before it sends anything else. That number is the status code. It is the shortest honest summary of what just happened.

> **In short**
> - The **first digit** is the family: 1 info, 2 success, 3 redirect, 4 your mistake, 5 the server's mistake.
> - **4xx** means the request was wrong. **5xx** means the server failed on a good request.
> - 401 means “who are you?” and 403 means “I know who you are, and no”.
> - 502, 503 and 504 all mean a problem behind the front door, but each tells a different story.

## The five families

**Read the first digit**

|  | Value | Note |
| --- | --- | --- |
| 1xx Informational | 100 | Hold on, keep going |
| 2xx Success | 200 | It worked |
| 3xx Redirection | 301 | Look somewhere else |
| 4xx Client error | 404 | The request was the problem |
| 5xx Server error | 500 | The server was the problem |

*If you can only remember one thing, remember that 4 and 5 split the blame between asker and answerer.*

## The dozen you will meet

| Code | Name | What it really means |
| --- | --- | --- |
| 200 | OK | Request worked and here is the result |
| 201 | Created | Something new was made, such as a new record |
| 204 | No Content | Worked, with nothing to send back |
| 301 | Moved Permanently | This address is gone for good. Update your links |
| 302 | Found | Temporarily elsewhere. Keep using the old address |
| 304 | Not Modified | Nothing changed, use your cached copy |
| 400 | Bad Request | The request was malformed |
| 401 | Unauthorized | You are not logged in or the credentials are wrong |
| 403 | Forbidden | Logged in, but not allowed |
| 404 | Not Found | Nothing lives at this address |
| 429 | Too Many Requests | Slow down: you hit a rate limit |
| 500 | Internal Server Error | The server crashed handling a good request |
| 503 | Service Unavailable | Overloaded or down for maintenance |

*Names follow the standard. “Unauthorized” is a historic misnomer: it really means unauthenticated.*

Try it: [HTTP Status Code Lookup](https://toolcloak.pages.dev/tools/http-status-codes). Search any status code and get a plain-English meaning with an example.

## Tell them apart

**Four codes people confuse**

| Box | What it means |
| --- | --- |
| 401 Unauthorized | We do not know who you are. Log in or send valid credentials. |
| 403 Forbidden | We know who you are. You are not allowed here. |
| 301 Moved Permanently | Update your bookmarks and links. Search engines move the ranking to the new address. |
| 302 Found | Use the new address this time only. The old one stays the main one. |

*Mixing up 301 and 302 is a classic SEO mistake. A temporary redirect does not hand over the old page's standing in the same way.*

## 502, 503, 504: the gateway trio

**You, the front door and the app behind it**

1. **You.** Send a request to the website
2. **Front door (proxy or load balancer).** Passes it to the real application
3. **The application.** Does the work, or fails to

*Each of these three codes is the front door telling you what it saw happen to the application.*

- **502 Bad Gateway:** the front door got a broken or invalid answer from the app.
- **503 Service Unavailable:** the app is too busy or switched off on purpose. Try again soon.
- **504 Gateway Timeout:** the app took too long, so the front door gave up waiting.

## For people who build things

1. **Use the right code.** A wrong password is 401, not 200 with an error message inside.
2. **Do not send 200 for failures.** Monitoring tools and caches read the code, not your message.
3. **Be honest about 404 and 410.** 410 Gone tells search engines the page was removed on purpose.
4. **Return 429 with a Retry-After header** so clients know when to try again.

> **A fun one:** 418 “I’m a teapot” comes from an April Fools’ joke standard from 1998. It is real, and some servers still answer it, but you will never need it.

## Frequently asked questions

### What do the first digits of HTTP status codes mean?

1xx informational, 2xx success, 3xx redirection, 4xx client error and 5xx server error.

### What is the difference between 401 and 403?

401 means you are not authenticated: log in or send valid credentials. 403 means you are authenticated but not permitted to do this.

### What is the difference between 301 and 302?

301 is a permanent move and passes the old page's standing to the new one. 302 is a temporary redirect, and the original address stays the main one.

### What causes a 502, 503 or 504 error?

502 means the gateway got a bad response from the app, 503 means the app is overloaded or down, and 504 means the app took too long to answer.

### Should an API return 200 for errors?

No. Use the appropriate 4xx or 5xx code so clients, caches and monitoring can tell that something failed.



---

# CTC to in-hand salary: where a ₹12 lakh package actually goes

> A ₹12 lakh CTC does not arrive as ₹1 lakh a month. Employer PF, gratuity, your own PF and professional tax come out first. In a typical split with a 40% basic, about ₹88,000 a month reaches your account before income tax. Here is the step-by-step walk from CTC to credit alert.

Published 2026-10-08 by Toolcloak. 4 minute read. Original: https://toolcloak.pages.dev/articles/ctc-to-in-hand-salary-where-a-12-lakh-package-actually-goes
Related free tool: [In-Hand Salary Calculator (CTC to Take-Home, India)](https://toolcloak.pages.dev/tools/in-hand-salary-calculator)

When an offer letter says ₹12 lakh, most people divide by twelve and plan for ₹1 lakh a month. The first salary credit is a shock. Nobody lied. CTC stands for cost to company, and it counts several things the company spends on you that never pass through your hands.

> **In short**
> - **CTC is what the employer spends, not what you receive.** It includes employer PF, gratuity and sometimes insurance or bonuses paid later.
> - In our example, ₹12,00,000 CTC with a 40% basic leaves about **₹10.59 lakh a year, or ₹88,276 a month, before income tax**.
> - A **higher basic share lowers take-home** because PF and gratuity are calculated on the basic. Moving from 30% to 50% basic changes the monthly figure by about ₹5,800.
> - Every company’s structure is different. This is an illustration with stated assumptions, so check your own offer letter.

## The assumptions

To keep the arithmetic clear, we use a simple structure. Real offer letters vary, and some include variable pay, insurance premiums, meal cards or joining bonuses.

- CTC ₹12,00,000 a year.
- Basic salary is 40% of CTC: ₹4,80,000 a year (₹40,000 a month).
- Employer PF is 12% of basic: ₹57,600 a year.
- Employee PF is also 12% of basic: ₹57,600 a year.
- Gratuity provision is 4.81% of basic: ₹23,088 a year.
- Professional tax is ₹2,400 a year (₹200 a month). It depends on the state, and some states charge none.
- Income tax is **not** included yet, because it depends on your regime and deductions.

## From CTC to the bank, step by step

**₹12 lakh CTC to yearly in-hand pay before income tax (₹ lakh)**

| Step | Amount |
| --- | --- |
| CTC | ₹12.00 L |
| Employer PF | −0.58 |
| Gratuity | −0.23 |
| Gross salary | ₹11.19 L |
| Your PF | −0.58 |
| Prof. tax | −0.02 |
| In-hand | ₹10.59 L |

*Gross salary is what the company pays out as monthly salary (₹93,276). After your own PF and professional tax, ₹88,276 a month remains before income tax.*

Two things in that picture are not lost. The employer PF and your own PF both go into your provident fund account, and the gratuity provision is paid when you leave after the required service. They are part of what you earn, they are just not spendable today. Our [gratuity guide](https://toolcloak.pages.dev/articles/gratuity-formula-15-by-26-what-your-years-of-service-are-worth) shows how large that gratuity can become.

Try it: [In-Hand Salary Calculator (CTC to Take-Home, India)](https://toolcloak.pages.dev/tools/in-hand-salary-calculator). Enter your CTC and basic to see a month-by-month take-home estimate.

**The monthly view**

|  | Value | Note |
| --- | --- | --- |
| CTC ÷ 12 | ₹1,00,000 | What most people expect |
| Gross monthly salary | ₹93,276 | After employer PF and gratuity |
| In-hand before tax | ₹88,276 | After your PF and prof. tax |

## Why the basic share matters

Provident fund and gratuity are both calculated on basic pay. If a larger part of your CTC is labelled basic, more goes into PF and gratuity and less is available as cash each month. The money is not gone, but you cannot spend it now.

**Monthly in-hand before income tax for ₹12 lakh CTC, by basic share**

|  | Value |
| --- | --- |
| Basic 30% | ₹91,157 |
| Basic 40% | ₹88,276 |
| Basic 50% | ₹85,395 |

*Each 10-point step in the basic share changes take-home by about ₹2,900 a month. In return you build a bigger PF and gratuity.*

## Other CTC levels

**Monthly in-hand before income tax (₹)**

| CTC a year | Basic 30% | Basic 40% | Basic 50% |
| --- | --- | --- | --- |
| ₹6 lakh | ₹45,479 | ₹44,038 | ₹42,598 |
| ₹12 lakh | ₹91,157 | ₹88,276 | ₹85,395 |
| ₹18 lakh | ₹1,36,836 | ₹1,32,514 | ₹1,28,193 |
| ₹24 lakh | ₹1,82,514 | ₹1,76,752 | ₹1,70,990 |

## Where income tax comes in

Income tax is taken from your gross salary through the year (TDS). How much depends on the tax regime you choose, your deductions and any other income. Employers apply the regime you declare, and many switch you to the default if you do not choose. That is why two colleagues with the same CTC can see very different take-home.

> **Check the rules for the current year:** Slabs, rebates, the standard deduction and regime defaults are revised in the Union Budget. Use your employer’s declaration form or a current tax calculator before you rely on a number.

## Questions to ask before accepting an offer

1. **What is the basic salary and what is the PF base?** Some employers calculate PF on the full basic, others on a ceiling of ₹15,000 of wages. The difference is large.
2. **Is any part of the CTC variable?** A bonus that is “up to” a figure may be paid late, in part or not at all.
3. **Are insurance, food coupons or reimbursements inside the CTC?** These are real benefits, but they do not arrive as cash.
4. **How much is the joining or retention bonus, and is it repayable?** If you leave early you may have to give it back.
5. **What is the notice period buyout?** It can matter more than a ₹50,000 difference in pay.

> **Compare offers on in-hand, not on CTC:** Two offers can have the same CTC and a ₹4,000 a month difference in take-home. Put both into a calculator with the same assumptions and compare the monthly figure, then add the value of PF, gratuity and insurance on top.

## What to do with the difference

Once you know your real monthly figure, build the budget from it, not from the CTC. Set aside an [emergency fund](https://toolcloak.pages.dev/articles/how-big-should-an-emergency-fund-be) first, and then start a monthly investment. Even a small one adds up, as shown in [start at 25 or 35](https://toolcloak.pages.dev/articles/start-at-25-or-35-what-waiting-ten-years-really-costs).

## Frequently asked questions

### How do I convert CTC to in-hand salary?

Subtract the employer PF, gratuity and any other employer-side benefits to get gross salary. Then subtract your own PF, professional tax, income tax and any other deductions.

### Why is my in-hand salary less than CTC divided by 12?

CTC includes employer PF, gratuity and benefits that are not paid in cash each month, and your own PF, taxes and professional tax are deducted before the credit.

### Does a higher basic salary reduce in-hand pay?

Usually yes, because PF and gratuity are calculated on basic. The money goes to your PF and gratuity instead of your bank account.

### Is PF part of my salary?

Yes. Both the employer and employee contributions are your money and earn interest in your PF account, but you cannot spend them until withdrawal conditions are met.

### What is the professional tax?

It is a state tax on salaried people. The amount depends on the state and slab, and some states do not charge it.



---

# What will ₹1 lakh buy in 2046? Inflation in pictures

> Nothing happens to cash in a drawer. The note stays the same. The shop prices do not. At 6 percent inflation, today’s ₹1 lakh buys about ₹31,000 worth of today’s goods in twenty years. Here is the slow leak, drawn out.

Published 2026-10-08 by Toolcloak. 3 minute read. Original: https://toolcloak.pages.dev/articles/what-will-one-lakh-buy-in-2046-inflation-in-pictures
Related free tool: [Inflation Calculator](https://toolcloak.pages.dev/tools/inflation-calculator)

Put ₹1,00,000 in a locker today and take it out in 2046. The number will not have changed by a single rupee. What it buys will have changed a great deal: at 6 percent a year, it will buy roughly what **₹31,000 buys today**.

> **In short**
> - At 6% inflation, ₹1 lakh today buys about ₹31,000 worth of today’s goods in 20 years.
> - Real return ≈ (1 + interest) ÷ (1 + inflation) − 1. A 7% deposit can be close to zero in real terms.
> - Write goals in future rupees and test a few inflation rates, such as 4, 6 and 8 percent.

Inflation is hard to feel because it never announces itself. No bill arrives saying “you lost ₹69,000”. The loss shows up as slightly higher prices at the tea stall, the school and the petrol pump, year after year, until one day the old numbers look unreal. This article puts numbers and pictures on that slow leak.

## How a rupee shrinks

If prices rise by a steady rate every year, the buying power of a fixed amount of money falls by the same compounding. After `n` years the money is worth `1 ÷ (1 + inflation)ⁿ` of what it was. This is the same maths as compound interest, running the wrong way.

**What ₹1 lakh of today’s buying power becomes (value in lakh rupees)**

| Year | At 6% inflation | At 4% inflation | At 8% inflation |
| --- | --- | --- | --- |
| 0 | 1 | 1 | 1 |
| 5 | 0.75 | 0.82 | 0.68 |
| 10 | 0.56 | 0.68 | 0.46 |
| 15 | 0.42 | 0.56 | 0.32 |
| 20 | 0.31 | 0.46 | 0.21 |
| 25 | 0.23 | 0.38 | 0.15 |
| 30 | 0.17 | 0.31 | 0.1 |

*Drag along the lines. At 6 percent, ₹1 lakh keeps three quarters of its power after five years, a bit over half after ten, and under a third after twenty. The nominal amount never changes.*

Try it: [Inflation Calculator](https://toolcloak.pages.dev/tools/inflation-calculator). Enter an amount, a number of years and an inflation rate to see the future cost or the buying power left.

## The same shrinking, as prices

Looking at it from the shop’s side is just as sobering. Something that costs ₹100 today costs this much in twenty years:

**What a ₹100 item costs in 20 years**

|  | Value |
| --- | --- |
| Today | ₹100 |
| 4% a year | ₹219 |
| 6% a year | ₹321 |
| 8% a year | ₹466 |

*Rounded to the nearest rupee. Two percentage points of extra inflation, 6 versus 4, make the twenty-year price about 47 percent higher.*

**How long until prices double**

|  | Value | Note |
| --- | --- | --- |
| At 4% inflation | 17.7 yrs | 72 ÷ 4 = 18, the rule of thumb |
| At 6% inflation | 11.9 yrs | 72 ÷ 6 = 12 |
| At 8% inflation | 9 yrs | 72 ÷ 8 = 9 |

*Also the time for your cash to lose half its buying power. The rule-of-72 article shows where this shortcut comes from.*

## The number that matters: your real return

A deposit that pays 7 percent sounds comfortable. But if prices rise 6 percent, you are only about 1 percent ahead in what you can actually buy. That left-over is called the **real return**, and it is the only return that counts.

> **Real return ≈ (1 + interest) ÷ (1 + inflation) − 1:** 7 percent interest with 6 percent inflation: 1.07 ÷ 1.06 − 1 = 0.0094, about 0.9 percent. Close to the shortcut “interest minus inflation”, but the division is the exact one.

Now add tax. If the interest is taxed at a 30 percent slab rate, a 7 percent deposit leaves 4.9 percent. Against 6 percent inflation, 1.049 ÷ 1.06 − 1 is about −1 percent. In buying power you went **backwards** while the balance looked like it grew. Your own tax rate decides this, so check yours.

**A 7 percent deposit, in two ways**

| Looking at the balance | Looking at what it buys |
| --- | --- |
| +7% | ≈ −1% |
| The number in the account keeps rising | After 6% inflation and a 30% tax slab |
| Statements look reassuring | You can afford slightly less each year |

*An example, not a forecast. Rates and tax slabs differ, and a different mix can come out ahead. The method is what to take away.*

## Which rate should you use?

Nobody knows the future rate of inflation. In recent years the headline figure in India has varied from year to year, and your own spending is not the same as the national basket. A sensible habit is to run **two or three** rates, such as 4, 6 and 8 percent, and plan so that the plan survives the middle one.

- **Education and healthcare** have often risen faster than general prices. If you are saving for these, use a higher rate than the headline number.
- **Rent, food and fuel** follow general inflation more closely.
- **Electronics** often get cheaper for the same quality, so inflation on them can be negative.

## What to do with the number

1. **Write goals in future money.** If you need ₹50 lakh in today’s buying power in 20 years, at 6 percent you need about ₹1.6 crore on the day. Use the calculator to get the exact figure.
2. **Keep cash for short-term needs only.** Money you will spend within a year or two does not lose much. Money you will need in fifteen years does.
3. **Aim for returns above inflation after tax.** If a place for your money cannot beat the real-rate test over your time horizon, it is a parking space, not a plan.
4. **Revisit once a year.** If actual prices rose faster than your assumed rate, adjust the goal rather than hoping.

## Frequently asked questions

### What will ₹1 lakh be worth in 20 years?

At 6 percent inflation, about ₹31,000 in today’s buying power. At 4 percent it is about ₹46,000, and at 8 percent about ₹21,500.

### How do I calculate purchasing power?

Divide the amount by (1 + inflation rate) raised to the number of years. ₹1,00,000 at 6 percent for 20 years is 1,00,000 ÷ 3.207, which is about ₹31,180.

### What is a real return?

The return after removing inflation. It is roughly interest minus inflation, and exactly (1 + interest) ÷ (1 + inflation) − 1.

### How long until prices double at 6 percent inflation?

About 12 years. Divide 72 by the rate, and you get a quick estimate of the doubling time.

### Which inflation rate should I assume for planning?

Nobody knows the future figure. Test a few rates, such as 4, 6 and 8 percent, and use a higher one for costs that tend to rise faster, like education and healthcare.



---

# CIDR and subnets: what /24, /26 and /20 really mean

> The number after the slash in an address like 192.168.1.0/24 counts how many of the 32 bits belong to the network. The rest are for hosts. A /24 holds 256 addresses, a /26 holds 64 and a /20 holds 4,096. This guide shows how to read, split and size a subnet.

Published 2026-10-08 by Toolcloak. 4 minute read. Original: https://toolcloak.pages.dev/articles/cidr-and-subnets-what-slash-24-slash-26-and-slash-20-really-mean
Related free tool: [IP Subnet Calculator](https://toolcloak.pages.dev/tools/ip-subnet-calculator)

Every device on a network has an address, and every address has two parts: which network it belongs to and which device it is on that network. CIDR notation is a compact way to say where the line between the two parts falls. Once that clicks, subnet masks stop looking like magic.

> **In short**
> - An IPv4 address has **32 bits**. In `192.168.1.0/24`, the **/24** means the first 24 bits name the network and the last 8 name the host.
> - Addresses in a subnet = **2 ^ (32 − prefix)**. So /24 = 256, /26 = 64, /20 = 4,096.
> - Two addresses are reserved in most subnets: the **network address** (first) and the **broadcast address** (last). Usable hosts = addresses − 2.
> - A bigger number after the slash means a **smaller** subnet.

## An address is 32 bits

An IPv4 address such as 192.168.1.10 is four numbers from 0 to 255. Each number is 8 bits, so the whole thing is 32 bits. The slash number says how many of those 32 bits, counted from the left, are fixed for the network.

**The 32 bits of a /24 network**

| Group | Count |
| --- | --- |
| Network bits (fixed) | 24 bits |
| Host bits (free to change) | 8 bits |

With 8 host bits there are 2⁸ = 256 different host values, from 0 to 255. Those 256 addresses are the whole /24 subnet.

## How big is each prefix?

Each step up in the prefix halves the subnet. Each step down doubles it. Here are the common sizes, with the subnet mask that goes with each:

| Prefix | Subnet mask | Addresses | Usable hosts |
| --- | --- | --- | --- |
| /16 | 255.255.0.0 | 65,536 | 65,534 |
| /20 | 255.255.240.0 | 4,096 | 4,094 |
| /24 | 255.255.255.0 | 256 | 254 |
| /25 | 255.255.255.128 | 128 | 126 |
| /26 | 255.255.255.192 | 64 | 62 |
| /27 | 255.255.255.224 | 32 | 30 |
| /28 | 255.255.255.240 | 16 | 14 |
| /29 | 255.255.255.248 | 8 | 6 |
| /30 | 255.255.255.252 | 4 | 2 |

*A /31 has 2 addresses and is used for point-to-point links. A /32 is a single host.*

**Usable hosts, by prefix length**

|  | Value |
| --- | --- |
| /30 | 2 |
| /29 | 6 |
| /28 | 14 |
| /27 | 30 |
| /26 | 62 |
| /25 | 126 |
| /24 | 254 |

*Each step to the right doubles the hosts (minus the two reserved addresses). The columns grow quickly, which is why the choice of prefix matters.*

Try it: [IP Subnet Calculator](https://toolcloak.pages.dev/tools/ip-subnet-calculator). Enter an address and prefix to see the network, broadcast, mask and usable range.

## Reading one subnet in full

Take 192.168.1.0/26. The host part is the last 6 bits, so there are 64 addresses.

**192.168.1.0/26**

|  | Value | Note |
| --- | --- | --- |
| Network address | 192.168.1.0 | First address, names the subnet |
| Usable hosts | .1 – .62 | 62 devices |
| Broadcast address | 192.168.1.63 | Last address, talks to all |

The subnet mask 255.255.255.192 is the same idea written in decimal: 26 ones followed by 6 zeros. The last number is 192 because the first two bits of the last byte are ones (128 + 64 = 192).

## Splitting a /24 into four /26 subnets

Say you have 192.168.1.0/24 and want four separate groups, such as staff, guests, cameras and servers. Borrow two bits from the host side and you get four /26 blocks of 64 addresses.

**192.168.1.0/24 split into four /26 blocks (last number of the address)**

|  | 0–63 | 64–127 | 128–191 | 192–255 |
| --- | --- | --- | --- | --- |
| The /24 | .0–.63 | .64–.127 | .128–.191 | .192–.255 |

*Each block has its own network and broadcast address, for example 192.168.1.64 and 192.168.1.127 for the second one.*

**How to size a subnet in four steps**

1. **Count the devices.** Include servers, phones, printers and room to grow.
2. **Add 2.** The network and broadcast addresses are not usable.
3. **Round up to a power of 2.** Subnet sizes are 4, 8, 16, 32, 64 and so on.
4. **Find the prefix.** Prefix = 32 minus the number of host bits. For 64 addresses, 6 host bits, so /26.

## Private ranges you will see everywhere

Three blocks are reserved for private use inside homes and companies. They are not routed on the public internet.

| Block | Range | Addresses | Often used for |
| --- | --- | --- | --- |
| 10.0.0.0/8 | 10.0.0.0 – 10.255.255.255 | 16,777,216 | Large companies and cloud networks |
| 172.16.0.0/12 | 172.16.0.0 – 172.31.255.255 | 1,048,576 | Medium networks and containers |
| 192.168.0.0/16 | 192.168.0.0 – 192.168.255.255 | 65,536 | Home routers and small offices |

## Common mistakes

- **Forgetting the two reserved addresses.** A /30 has 4 addresses but only 2 usable hosts, which is why it is used for links between two routers.
- **Choosing a subnet that is too tight.** A /27 holds 30 hosts. If you have 28 devices today, you have no room for growth.
- **Overlapping subnets.** 10.0.0.0/16 and 10.0.5.0/24 overlap, because the second sits inside the first. Overlaps cause routing trouble, especially when joining two networks by VPN.
- **Reading the mask as a number to add.** The mask is a pattern of ones and zeros. 255.255.255.192 is 26 ones, not a number you subtract.
- **Mixing it up with IPv6.** IPv6 uses 128 bits and prefix lengths like /64, but the idea of network bits and host bits is the same.

> **A shortcut for the last number:** In a /26 the block size is 64, so subnets start at 0, 64, 128 and 192. For a prefix that falls inside the last byte, block size = 256 minus the last number of the mask (256 − 192 = 64). If the address you are checking is 192.168.1.100/26, it belongs to the block that starts at 64.

## Frequently asked questions

### What does /24 mean in an IP address?

It means the first 24 bits are the network part and the last 8 bits are for hosts. That gives 256 addresses, of which 254 are usable.

### How many hosts are in a /26?

A /26 has 64 addresses and 62 usable hosts, because the first and last addresses are reserved.

### What is the subnet mask for /20?

255.255.240.0. It has 20 ones followed by 12 zeros and covers 4,096 addresses.

### What is a broadcast address?

It is the last address in a subnet. A message sent to it is delivered to every device in that subnet.

### Is a bigger prefix number a bigger network?

No, the opposite. A bigger prefix means more network bits and fewer host bits, so the subnet is smaller.

### Which IP ranges are private?

10.0.0.0/8, 172.16.0.0/12 and 192.168.0.0/16. They are for use inside private networks and are not routed on the public internet.



---

# Inside a JWT: three parts, two dots, and why decoding is not the same as trusting

> A JSON Web Token looks like a random string with two dots. Split it at the dots and each part is readable. The catch: anyone can read it, and only the signature proves nobody changed it.

Published 2026-10-08 by Toolcloak. 3 minute read. Original: https://toolcloak.pages.dev/articles/inside-a-jwt-three-parts-one-dot-at-a-time
Related free tool: [JWT Decoder](https://toolcloak.pages.dev/tools/jwt-decoder)

Log in to many apps and the server hands you a token that looks like `xxxxx.yyyyy.zzzzz`. That is a JWT. It is the app's way of saying, without looking anything up, “this person was already checked.” Understanding what is inside makes login bugs much easier to fix.

> **In short**
> - A JWT has **three parts** joined by dots: header, payload, signature.
> - Header and payload are only **Base64URL-encoded JSON**. They are **not encrypted**, so anyone can read them.
> - The **signature** is what stops tampering. Decoding a token does not check it.
> - Never put passwords or secrets in the payload.

## The three parts

**header.payload.signature**

1. **Header.** Says which algorithm signed it, for example HS256 or RS256, and that the type is JWT
2. **Payload.** The claims: who the user is, who issued it, and when it expires
3. **Signature.** A fingerprint made with a secret or private key. Change one character in the first two parts and it stops matching

*The first two parts are Base64URL text. A browser or a decoder tool can read them in a blink.*

## The claims you will see

| Claim | Meaning | Example |
| --- | --- | --- |
| `sub` | Subject: who the token is about | A user ID |
| `iss` | Issuer: who created it | Your auth server's address |
| `aud` | Audience: who it is meant for | Your API |
| `exp` | Expiry time, as a Unix timestamp | 1790000000 |
| `iat` | Issued-at time | Unix timestamp |
| `nbf` | Not valid before this time | Unix timestamp |
| `jti` | A unique ID for this token | Random string |

*Times are Unix seconds. See [Unix time](https://toolcloak.pages.dev/articles/unix-time-the-number-that-counts-seconds-since-1970) for how to read them.*

Try it: [JWT Decoder](https://toolcloak.pages.dev/tools/jwt-decoder). Paste a token to see its header and payload, with the expiry turned into a readable date. Nothing leaves your browser.

## Decoding is not verifying

**Two very different actions**

| Box | What it means |
| --- | --- |
| Decode | Turn the first two parts back into JSON. Anyone can do it, no key needed. Fine for debugging. |
| Verify | Recompute the signature with the key and compare. This is what proves the token is genuine. |
| Trust the payload after decoding only | Dangerous. An attacker can write any payload they like. |
| Trust it after verifying and checking exp | Correct. Verify the signature, then the time and audience. |

*Your server must verify every token on every request. A decoder tool is for reading, not for trusting.*

## Why “it is just encoded” matters

**What a JWT is, and is not**

| Is | Is not |
| --- | --- |
| Signed | Secret |
| Tamper-evident | Anyone holding it can read it |
| Self-contained: the server needs no lookup | Not encrypted by default, so no passwords or private data inside |

*If you need the contents hidden, use an encrypted token (JWE) or keep the data on the server and send only an ID.*

## Common JWT mistakes

1. **Not checking `exp`.** Without it, a stolen token works forever.
2. **Accepting `alg: none`.** Always pin the allowed algorithms on the server.
3. **Long expiry for everything.** Keep access tokens short, refresh them with a separate token.
4. **Storing tokens where scripts can read them.** If a page can be injected, the token can be stolen.
5. **Pasting a live token into random websites.** Use a tool that decodes locally, and treat real tokens like passwords.

## Frequently asked questions

### What is a JWT?

A JSON Web Token: a signed string made of a header, a payload of claims and a signature, joined by dots. Servers use it to prove a user was already authenticated.

### Is a JWT encrypted?

Not by default. The header and payload are only Base64URL-encoded, so anyone can read them. The signature only prevents tampering.

### What is the difference between decoding and verifying a JWT?

Decoding just reads the contents and needs no key. Verifying recomputes the signature with the key to prove the token is genuine and unaltered.

### What does exp mean in a JWT?

The expiry time, as a Unix timestamp in seconds. After that moment the token must be rejected.

### Is it safe to paste a JWT into an online decoder?

Only if the tool decodes in your browser and you are not sending a live production token to a service you do not trust. Treat real tokens like passwords.



---

# Why 2100 is not a leap year (and 2000 was)

> Every fourth year gets a 29 February. Except some do not. The full rule has three parts, and it exists because a year is not a whole number of days.

Published 2026-10-08 by Toolcloak. 3 minute read. Original: https://toolcloak.pages.dev/articles/why-2100-is-not-a-leap-year
Related free tool: [Leap Year Checker](https://toolcloak.pages.dev/tools/leap-year-checker)

Most people learn: a leap year is any year divisible by 4. It works for your whole life, so it feels complete. But 1900 was not a leap year, 2000 was, and 2100 will not be. The shortcut is nearly right, and the gap is the interesting part.

> **In short**
> - Leap year if divisible by 4, except century years, unless also divisible by 400.
> - So 2000 was a leap year, but 1900 and 2100 are not.
> - The rule gives an average year of 365.2425 days, very close to the real 365.2422.

## The full rule

1. If the year is divisible by **400**, it is a leap year. (2000, 2400)
2. Otherwise, if it is divisible by **100**, it is **not** a leap year. (1900, 2100, 2200)
3. Otherwise, if it is divisible by **4**, it is a leap year. (2024, 2028, 2032)
4. Anything else is a normal 365-day year.

Try it: [Leap Year Checker](https://toolcloak.pages.dev/tools/leap-year-checker). Type a year to see whether it is a leap year, and which rule applies.

| Year | Divisible by 4 | By 100 | By 400 | Leap year? |
| --- | --- | --- | --- | --- |
| 2024 | Yes | No | No | Yes |
| 2026 | No | No | No | No |
| 1900 | Yes | Yes | No | No |
| 2000 | Yes | Yes | Yes | Yes |
| 2100 | Yes | Yes | No | No |

*Century years are the only ones where the “every four years” rule is overruled.*

## Why a leap year exists at all

Earth takes about **365.2422 days** to go round the Sun. A calendar with only 365 days falls behind by nearly a quarter of a day each year. After four years it is almost a whole day out, and the seasons slowly drift against the dates. The extra day on 29 February pulls the calendar back.

## Adding a day every four years is too much

A leap day every fourth year makes the average year 365.25 days. That is a hair too long, by about 0.0078 days a year. Small, but it builds up: roughly **one extra day every 128 years**.

**Three calendars, three average years**

|  | Value | Note |
| --- | --- | --- |
| Plain year | 365 | Drifts by about 1 day every 4 years |
| Leap day every 4 years | 365.25 | Drifts by 1 day every ~128 years |
| The full rule | 365.2425 | Drifts by 1 day in roughly 3,000 years |

*The real tropical year is about 365.2422 days. The Gregorian rule is off by only around 0.0003 days a year.*

The fix is to skip the leap day in century years, except every fourth century. In 400 years that is 100 years divisible by 4, minus 4 century years, plus 1 for the one divisible by 400: **97 leap years**. And 365 + 97/400 = 365.2425.

**Leap years in each 400-year cycle**

|  | Value |
| --- | --- |
| Every 4th year | 100 |
| Minus centuries | 96 |
| Plus 400s | 97 |

*Start with 100, remove the 4 century years, then add one back for the year divisible by 400. That leaves 97 per cycle.*

## A bit of history

**How the calendar got fixed**

| When | What |
| --- | --- |
| 45 BC | The Julian calendar. A leap day every four years, no exceptions. Average year of 365.25 days. |
| 1582 | The Gregorian reform. Ten days were removed: 4 October was followed by 15 October. The century rule began. |
| Later | Countries switched at different times. Britain and its colonies changed in 1752. Others waited until the twentieth century. |

*Because countries changed on different dates, old records from different places can disagree by days.*

## Useful facts

- **Next leap years:** 2028, 2032 and 2036. The year 2100 will skip it.
- **A 29 February birthday** comes up once in 1,461 days, because four years hold 365 × 4 + 1 days.
- **Software bugs.** Programs that only check “divisible by 4” fail on 2100. Programs that forget the 400 rule fail on 2000, and some did.
- **Days in a leap year:** 366, so a year has 52 weeks and 2 days, which is why dates shift weekdays by two in a leap year and one in a normal year.

## Frequently asked questions

### Is 2100 a leap year?

No. It is divisible by 100 but not by 400, so the century rule removes it.

### Was 2000 a leap year?

Yes. It is divisible by 400, which overrides the century exception.

### What is the leap year rule?

Divisible by 4 is a leap year, except years divisible by 100, unless they are also divisible by 400.

### Why are there leap years?

Earth’s orbit takes about 365.2422 days, so a 365-day calendar drifts. A leap day keeps dates aligned with the seasons.

### How many leap years are there in 400 years?

97. That gives an average calendar year of 365.2425 days.



---

# AI tokens and API cost: why the bill depends on the answer length and the chat history

> AI services charge by the token, not by the word or the message. Output tokens usually cost several times more than input, and a long conversation resends its whole history every turn. This guide explains tokens with simple rules of thumb, works through a monthly-bill example, and shows four ways to cut the bill without hurting quality.

Published 2026-10-08 by Toolcloak. 6 minute read. Original: https://toolcloak.pages.dev/articles/ai-tokens-and-api-cost-why-the-bill-depends-on-output-and-chat-history
Related free tool: [AI API Cost Calculator](https://toolcloak.pages.dev/tools/llm-cost-calculator)

You build a small chatbot, test it for an afternoon, and it costs almost nothing. Then real users arrive, and the bill looks nothing like your estimate. The reason is rarely a mystery: AI services meter usage in tokens, and tokens behave in ways that are easy to misjudge. Once you understand three things, what a token is, why output costs more, and why history keeps growing, you can predict the bill before you ship.

> **In short**
> - A **token** is a chunk of text, roughly **4 characters or 0.75 words** in English. 1,000 words is about **1,300 tokens**.
> - You pay for **input tokens** (what you send) and **output tokens** (what the model writes). Output is usually priced at several times the input rate.
> - In a chat, every turn **resends the earlier messages**, so input grows turn by turn. A 10-turn chat can use **12 times** the input of its new text alone.
> - To cut cost: **shorten answers, trim history, keep prompts tight, and cache or batch** where the provider allows.

## What a token is

Language models do not read letters or whole words. They read **tokens**, pieces of text chosen so that common fragments get a single token. “the” is one token. A long or unusual word may be split into two or three pieces. Spaces, punctuation and numbers also count. The exact split depends on the model’s tokenizer, so different services count the same sentence slightly differently.

For plain English, two rules of thumb work well enough for budgeting: about **4 characters per token**, or about **three quarters of a word per token**. They are estimates. Code, numbers and rare words use more tokens. Text in scripts other than the Latin alphabet, such as Hindi in Devanagari, often needs noticeably more tokens for the same meaning, so a Hindi or mixed-language chatbot can cost more than the same bot in English.

**Roughly how many tokens different texts use (English)**

|  | Tokens |
| --- | --- |
| A tweet (280 characters) | ≈ 70 |
| A 100-word email | ≈ 133 |
| A 500-word page | ≈ 667 |
| A 2,000-word article | ≈ 2,670 |
| A 25,000-word report (about 50 pages) | ≈ 33,000 |

*Tokens ≈ words ÷ 0.75. Use these only for planning. For exact counts, run your text through a [token counter](https://toolcloak.pages.dev/tools/token-counter).*

Try it: [AI API Cost Calculator](https://toolcloak.pages.dev/tools/llm-cost-calculator). Enter prompt size, answer size, price per million tokens and number of calls to estimate the cost.

## Input, output and the price per million

Providers publish prices per **million tokens**, with separate rates for input and output. Output costs more because the model has to produce it one token at a time, which uses far more computing than reading the input in one pass. A common pattern is output at around three to five times the input rate.

> **Cost = (input tokens × input price + output tokens × output price) ÷ 1,000,000:** In this article we use example prices of **$3 per million input tokens** and **$15 per million output tokens**. These are made-up round numbers for the arithmetic. Check your provider’s current price list before budgeting.

Take a customer-support chatbot. Each conversation sends about 800 tokens in (instructions plus the customer’s message) and gets 300 tokens back. The cost of one conversation is (800 × 3 + 300 × 15) ÷ 1,000,000 = **$0.0069**, under one cent. That feels free, until you multiply.

**A support bot with 1,000 conversations a day**

|  | Value | Note |
| --- | --- | --- |
| Per conversation | $0.0069 | 800 tokens in, 300 out |
| Per 1,000 conversations | $6.90 | One day of traffic |
| Per month | $207 | 30 days of the same load |

*A tenth of a cent here and there becomes real money at scale. Always multiply by your expected volume.*

**Where the $207 a month goes**

|  | Share |
| --- | --- |
| Input tokens (800 per chat) | $72 |
| Output tokens (300 per chat) | $135 |

*Output is only 27% of the tokens (300 of 1,100) but 65% of the cost. That is why the length of the answer is the first thing to look at.*

## Why chat history makes the bill grow

A model has no memory between requests. To make a chat feel continuous, the app sends the **entire conversation so far** each time. On turn 1 you send only the first message. On turn 5 you send turns 1 to 4 again, plus the new one. The new text each turn stays small, but the input keeps climbing.

Suppose each user message is 200 tokens and each reply 300 tokens. Turn 1 sends 200 tokens. Turn 2 sends 200 + (200 + 300) = 700. Turn 3 sends 1,200. And so on, 500 more every turn.

**Input tokens sent on each turn of a 10-turn chat**

| Turn number | Input actually sent (history + new message) | New text only |
| --- | --- | --- |
| 1 | 200 | 200 |
| 2 | 700 | 200 |
| 3 | 1200 | 200 |
| 4 | 1700 | 200 |
| 5 | 2200 | 200 |
| 6 | 2700 | 200 |
| 7 | 3200 | 200 |
| 8 | 3700 | 200 |
| 9 | 4200 | 200 |
| 10 | 4700 | 200 |

*By turn 10 you are sending 4,700 tokens to ask a 200-token question. The grey line is what it would be if history were free; the blue line is what you pay.*

**Cost of one 10-turn chat, with and without the history effect (example prices)**

|  | Input | Output |
| --- | --- | --- |
| If only new text counted | $0.006 | $0.045 |
| With history resent | $0.0735 | $0.045 |

*Total input for the 10 turns is 24,500 tokens against only 2,000 of genuinely new text, 12 times as much. The output stays at 3,000 tokens either way, so the history effect hits the input side.*

> **Long chats get expensive, not just long:** The total input of a chat grows roughly with the **square** of its length: doubling the number of turns more than doubles the input. That is why “start a new conversation” is good cost advice as well as good usability advice.

## The context window: a ceiling on the sum

Every model has a **context window**, the maximum number of tokens it can handle in one request, counting input and output together. A window of 128,000 tokens holds roughly 96,000 words, close to a full-length novel. If your history plus the new question plus the expected answer exceeds the window, something must be dropped or the call fails.

**What has to fit in one request**

1. **Instructions.** The system prompt that tells the model how to behave.
2. **History and documents.** Earlier turns, retrieved notes and any files you pasted in.
3. **The new message.** What the user just asked.
4. **Room for the answer.** Output tokens count against the same window.

*Plan the window like a suitcase: everything has to fit at once. The [context window calculator](https://toolcloak.pages.dev/tools/ai-context-window-calculator) shows how much space is left.*

If a document is too big, split it into pieces and process them one by one with an [AI text chunker](https://toolcloak.pages.dev/tools/ai-text-chunker), rather than forcing it into a single call.

## How big prompts and big answers change the cost

The next table shows the cost of **100,000 calls** at the example prices for different prompt and answer sizes. Moving down means a bigger prompt, moving right means a longer answer.

**Cost of 100,000 calls (US dollars, example prices)**

| Input tokens | 100 out | 500 out | 1,500 out |
| --- | --- | --- | --- |
| 500 in | $300 | $900 | $2,400 |
| 2,000 in | $750 | $1,350 | $2,850 |
| 8,000 in | $2,550 | $3,150 | $4,650 |

## Four ways to cut the cost

Because the bill is a simple formula, each lever is easy to test. Using the support-bot example (800 in, 300 out, $6.90 per 1,000 conversations), here is what each change does.

**Cost per 1,000 conversations after each change (example prices)**

|  | Cost |
| --- | --- |
| Today | $6.90 |
| Halve the prompt (400 in) | $5.70 |
| Halve the answer (150 out) | $4.65 |
| Halve both | $3.45 |

*Halving the answer saves 33%, halving the prompt only 17%, because output costs five times as much per token. Halving both cuts the bill by half.*

1. **Ask for shorter answers.** A line such as “answer in under 100 words unless asked for more” can cut output dramatically. Set a maximum output length as a safety net.
2. **Trim the history.** Keep the last few turns and a short summary of the rest, instead of the whole transcript.
3. **Tighten the instructions.** A 2,000-token system prompt sent on every call is a fixed tax. Remove examples and rules the model does not need. The prompt tools on this site, such as the [prompt generator](https://toolcloak.pages.dev/tools/prompt-generator), help you write compact ones.
4. **Cache and batch where offered.** Some providers discount repeated prompt prefixes and offer cheaper rates for jobs that can wait. Check the documentation, since terms vary.
5. **Use the right-sized model.** Easy tasks such as classification or short rewrites rarely need the most expensive model. Test a smaller one on a sample.

> **Do not forget the retries:** Failed calls, retries and “try again” buttons are billed like any other call. If 10% of requests are retried, your real volume is 10% higher than your user count suggests.

## A back-of-envelope method you can reuse

**Estimate the monthly bill in five steps**

1. **Estimate tokens per call.** Prompt plus history in, answer out. Use words ÷ 0.75 if you have only word counts.
2. **Multiply by the price.** Input tokens × input rate plus output tokens × output rate, each per million.
3. **Multiply by volume.** Calls per day × 30.
4. **Add a safety margin.** Add 20 to 30 percent for retries, long chats and growth.
5. **Test with real traffic.** Run a small pilot and compare the real token counts with your guess.

A single AI call is nearly free. A million of them is a line item in the budget. The skill is simply in multiplying early, watching the output length, and keeping history under control. If you like thinking in compounding terms, the same “small things times many” idea appears in [the minimum payment trap](https://toolcloak.pages.dev/articles/the-minimum-payment-trap-what-a-credit-card-balance-really-costs) and [the appliance cost guide](https://toolcloak.pages.dev/articles/what-your-appliances-cost-per-month-watts-hours-and-units-explained). Put your own numbers in the calculator above and see what your app will cost before you launch it.

## Frequently asked questions

### What is a token in AI?

A token is a small chunk of text, such as a short word or part of a longer word, which language models use as their unit of reading and writing. In English, one token is about four characters or three quarters of a word.

### How many tokens is 1,000 words?

About 1,300 to 1,400 tokens in English. Code, numbers and non-English text usually use more.

### Why do output tokens cost more than input tokens?

The model writes output one token at a time, which takes much more computing than reading the input. Providers usually price output at several times the input rate.

### Why does a long chat cost more per message?

The app resends the earlier messages with each new one, so the input grows every turn. Total input grows roughly with the square of the chat length.

### What is a context window?

The maximum number of tokens, input and output combined, that a model can handle in a single request.

### How can I lower my AI API bill?

Ask for shorter answers, trim the conversation history, shorten the instructions, use caching or batch discounts where offered, and pick a smaller model for easy tasks.



---

# One extra EMI a year: how a 20-year home loan quietly becomes a 17-year loan

> Pay a single extra EMI every year on a ₹50 lakh loan and you finish about 3 years early and save more than ₹10 lakh in interest. Here is exactly why prepayment works so well, and when it does not.

Published 2026-10-08 by Toolcloak. 3 minute read. Original: https://toolcloak.pages.dev/articles/one-extra-emi-a-year-how-a-home-loan-shrinks
Related free tool: [Loan Prepayment Calculator (Interest Saved)](https://toolcloak.pages.dev/tools/loan-prepayment-calculator)

A home loan feels fixed: a number, a tenure, an EMI. But it is the one financial contract where a small extra payment now has an outsized effect later. One extra EMI a year, paid in a month you barely notice, takes years off a loan.

> **In short**
> - On a ₹50 lakh loan at 8.5 percent for 20 years, the EMI is about **₹43,391** and the total interest about **₹54.1 lakh**.
> - Paying **one extra EMI every year** ends the loan in about 16 years 9 months and saves about **₹10.3 lakh** in interest.
> - Early prepayments save the most. A ₹1 lakh prepayment in year 1 saves about ₹3.8 lakh of interest, in year 15 only about ₹0.5 lakh.

## Why prepaying early is so powerful

Interest is charged on what you still owe. An extra payment goes straight to the principal, so from that day onward you pay interest on a smaller number, and that smaller number keeps compounding in your favour for the rest of the loan. The earlier it happens, the more months it works.

**₹50 lakh, 8.5 percent, 20 years**

|  | Value | Note |
| --- | --- | --- |
| Monthly EMI | ₹43,391 | Same in both cases |
| Interest if you do nothing | ₹54.1 L | Over 20 years |
| Interest with 1 extra EMI a year | ₹43.9 L | Over about 16 yrs 9 months |

*The extra EMI is paid once a year, for example with your annual bonus. Interest saved is about ₹10.3 lakh.*

**The same loan, two ways**

| Pay only the EMI | Add 1 extra EMI a year |
| --- | --- |
| 240 months | 201 months |
| Debt-free in 20 years | Debt-free about 3 years earlier |
| Interest paid: about ₹54.1 lakh | Interest paid: about ₹43.9 lakh |

*You paid about ₹6.9 lakh in extra EMIs in total and got back about ₹10.3 lakh in interest saved.*

Try it: [Loan Prepayment Calculator (Interest Saved)](https://toolcloak.pages.dev/tools/loan-prepayment-calculator). Enter your loan, rate and a prepayment amount. It shows the new end date and the interest you save.

## Why a loan is mostly interest at the start

In the first years of a long loan, most of each EMI is interest. Only a small slice reduces the debt. That is exactly why an early lump sum has such leverage: it removes principal that the loan would otherwise take years to chip away.

**How much a ₹1 lakh prepayment saves, by when you pay it**

| Step | Value |
| --- | --- |
| In year 1 (saves 11 months) | ₹3.8 lakh saved |
| In year 5 (saves 8 months) | ₹2.5 lakh saved |
| In year 10 (saves 5 months) | ₹1.3 lakh saved |
| In year 15 (saves 3 months) | ₹0.5 lakh saved |

*A one-time ₹1 lakh prepayment on the same ₹50 lakh, 8.5 percent, 20-year loan, keeping the EMI the same. Paid in year 1 it saves about 7 times more interest than the same lakh in year 15.*

## Shorten the tenure or lower the EMI?

When you prepay, most lenders let you pick. Both reduce the interest, but not equally.

| Choice | What changes | Best for |
| --- | --- | --- |
| Keep EMI, shorter tenure | You are debt-free sooner and save the most interest | People who can afford the current EMI |
| Lower EMI, same tenure | Monthly outgo falls, saving is smaller | People who need cash-flow breathing room |

*If you can keep paying the same EMI, shortening the tenure almost always saves more.*

## When prepaying is not the best move

- **No emergency fund.** Prepaid money cannot be taken back out. Build 6 months of expenses first. See the emergency fund article.
- **Higher-interest debt exists.** A credit card at 36 percent should be cleared before a home loan at 8.5.
- **Prepayment charges apply.** Floating rate home loans for individuals usually have none, but check your loan terms.
- **You would use the money better elsewhere.** If your investments reliably beat your loan rate after tax, it is a closer call. Most people value the certainty.

## Frequently asked questions

### How much does one extra EMI a year save?

On a ₹50 lakh loan at 8.5 percent for 20 years, one extra EMI a year ends the loan about 3 years earlier and saves about ₹10.3 lakh in interest.

### Is it better to prepay early or late in the loan?

Early. In the first years most of your EMI is interest, so every extra rupee removes principal that would have kept generating interest for a long time.

### Should I reduce my EMI or my tenure after prepaying?

If you can afford the current EMI, keeping it and shortening the tenure saves more interest. Lowering the EMI helps cash flow but saves less.

### Are there charges for prepaying a home loan?

Floating-rate home loans taken by individuals usually carry no prepayment charges, but fixed-rate loans and some others may. Check your loan agreement.

### Are these numbers exact?

They are calculated monthly at a fixed 8.5 percent and are for illustration. Your lender's schedule, rate changes and payment dates will shift them slightly.



---

# Lumpsum investment: what ₹1 lakh becomes in 10, 20 and 30 years

> A lumpsum is money invested once and left alone. At 12% a year, ₹1 lakh becomes ₹3.11 lakh in 10 years, ₹9.65 lakh in 20 and ₹29.96 lakh in 30. At 8% the 30-year figure is only ₹10.06 lakh. Here is the formula, the effect of the rate, and when lumpsum beats a SIP.

Published 2026-10-08 by Toolcloak. 3 minute read. Original: https://toolcloak.pages.dev/articles/lumpsum-investment-what-1-lakh-becomes-in-10-20-and-30-years
Related free tool: [Lumpsum Investment Calculator](https://toolcloak.pages.dev/tools/lumpsum-calculator)

A bonus, an inheritance or a maturing deposit: sooner or later most people hold a single large amount and wonder what it could become. A lumpsum calculation answers that with one formula, and it also shows how much the return rate and the number of years each matter.

> **In short**
> - **Future value = amount × (1 + rate) ^ years.** ₹1 lakh at 12% for 20 years is 1 lakh × 9.65 = **₹9.65 lakh**.
> - The **rate and the time both multiply**. 30 years at 12% gives ₹29.96 lakh, nearly **3 times** the 20-year result.
> - A 4-point lower rate matters hugely over long periods: **₹10.06 lakh at 8% against ₹29.96 lakh at 12%** after 30 years.
> - Returns on market-linked investments are never fixed. These are illustrations, not forecasts.

## The formula

> **FV = P × (1 + r)ⁿ:** P is the amount you invest, r is the yearly rate as a decimal (12% = 0.12) and n is the number of years. For ₹1,00,000 at 12% for 10 years: 1,00,000 × 1.12¹⁰ = 1,00,000 × 3.106 = ₹3.11 lakh.

Because the amount is multiplied by the same factor every year, the growth in rupees gets bigger each year even though the percentage is the same. That is compounding.

**₹1 lakh at three different rates (lakh rupees)**

| Years | 8% a year | 10% a year | 12% a year |
| --- | --- | --- | --- |
| 0 | 1 | 1 | 1 |
| 5 | 1.47 | 1.61 | 1.76 |
| 10 | 2.16 | 2.59 | 3.11 |
| 15 | 3.17 | 4.18 | 5.47 |
| 20 | 4.66 | 6.73 | 9.65 |
| 25 | 6.85 | 10.83 | 17 |
| 30 | 10.06 | 17.45 | 29.96 |

*The three paths start together and fan out. The gap is small at year 5 and enormous at year 30.*

Try it: [Lumpsum Investment Calculator](https://toolcloak.pages.dev/tools/lumpsum-calculator). Enter an amount, a rate and a number of years to see the future value.

## The table to keep

**What ₹1 lakh becomes (lakh rupees)**

| Rate | 10 years | 20 years | 30 years |
| --- | --- | --- | --- |
| 8% | ₹2.16 L | ₹4.66 L | ₹10.06 L |
| 10% | ₹2.59 L | ₹6.73 L | ₹17.45 L |
| 12% | ₹3.11 L | ₹9.65 L | ₹29.96 L |
| 15% | ₹4.05 L | ₹16.37 L | ₹66.21 L |

## Which matters more: rate or time?

Both. But look at the 10% row: doubling the time from 10 to 20 years multiplies the result by 2.6, and tripling it to 30 years by 6.7. Time has an accelerating effect. Adding years is often the easier lever, because you control it more than the market return.

**₹1 lakh at 10%: value at each age of the investment (lakh rupees)**

|  | Value |
| --- | --- |
| 5 yrs | ₹1.61 L |
| 10 yrs | ₹2.59 L |
| 15 yrs | ₹4.18 L |
| 20 yrs | ₹6.73 L |
| 25 yrs | ₹10.83 L |
| 30 yrs | ₹17.45 L |

*The last five years add ₹6.6 lakh, more than the first 20 years combined (₹5.7 lakh of growth).*

## Lumpsum or SIP?

A SIP invests a fixed amount every month. A lumpsum invests everything at once. If you already have the money, the lumpsum starts compounding sooner. If you do not, a SIP builds the habit from your income.

**₹6 lakh, ten years, 12% a year**

| Lumpsum today | ₹5,000 a month for 10 years |
| --- | --- |
| ₹19.8 lakh | ₹11.6 lakh |
| All ₹6 lakh invested on day one | Same ₹6 lakh in total, spread over 120 months |
| Needs the full sum now | Fits a monthly salary |
| Fully exposed to the market at the start | Spreads the buying price across market ups and downs |

*The lumpsum is larger because the money works for longer. But they solve different problems. Most people have a salary, not a spare ₹6 lakh. See [start at 25 or 35](https://toolcloak.pages.dev/articles/start-at-25-or-35-what-waiting-ten-years-really-costs) for the SIP side.*

## Real returns are lower than the headline

- **Inflation.** ₹29.96 lakh in 30 years buys far less than ₹29.96 lakh today. See [what ₹1 lakh buys in 2046](https://toolcloak.pages.dev/articles/what-will-one-lakh-buy-in-2046-inflation-in-pictures).
- **Tax.** Gains are taxed when you sell, at rates that depend on the asset and holding period. Check the current rules.
- **Costs.** Fund expenses and fees reduce your return a little every year.
- **Risk.** An average of 12% can hide a −30% year. The order of returns changes what you end up with if you need money at a fixed date.
- **Averages versus the CAGR.** A 12% simple average and a 12% annual compounded return are different. Read [average return vs CAGR](https://toolcloak.pages.dev/articles/average-return-vs-cagr-why-20-percent-is-not-what-you-earned).

> **Large amount, nervous about timing?:** You can split a lumpsum into 6 or 12 monthly instalments and keep the rest in a safe place meanwhile. It lowers the chance of investing everything just before a fall, at the cost of less time in the market.

## The rule of 72 shortcut

To estimate how long it takes to double, divide 72 by the rate. At 12% that is 6 years, at 8% it is 9 years. See [the rule of 72](https://toolcloak.pages.dev/articles/the-rule-of-72-how-long-money-takes-to-double) for how it works.

## Frequently asked questions

### What is the lumpsum formula?

Future value = P × (1 + r)ⁿ, where P is the amount, r is the yearly rate and n is the years.

### What will ₹1 lakh be in 20 years at 12%?

About ₹9.65 lakh, if it earns 12% a year compounded annually. Actual returns will vary.

### Is a lumpsum better than a SIP?

A lumpsum gets more time to grow if you have the money now. A SIP suits regular income and spreads the buying price. They solve different problems.

### How does the rate change the result?

Over long periods a small change in the rate changes the result a lot. At 8% ₹1 lakh becomes ₹10.06 lakh in 30 years, at 12% it becomes ₹29.96 lakh.

### Are these returns guaranteed?

No. Market-linked investments move up and down. These figures show how the maths works for a steady rate.



---

# Markup vs margin: why a 50% markup is only a 33% margin

> Markup is profit as a share of cost. Margin is profit as a share of the selling price. A product that costs ₹600 and sells for ₹1,000 has a 66.7% markup and a 40% margin. Mixing the two up quietly cuts your profit, and this guide shows exactly how.

Published 2026-10-08 by Toolcloak. 3 minute read. Original: https://toolcloak.pages.dev/articles/markup-vs-margin-why-a-50-percent-markup-is-only-a-33-percent-margin
Related free tool: [Markup & Margin Calculator](https://toolcloak.pages.dev/tools/markup-margin-calculator)

Two shopkeepers both say they “make 30% on everything”. One earns ₹23 on every ₹100 sale, the other earns ₹30. They are not lying. One of them is talking about markup, the other about margin, and the two words sit on different bases.

> **In short**
> - **Markup = profit ÷ cost × 100.** **Margin = profit ÷ selling price × 100.**
> - Margin is always the smaller number. A **50% markup is a 33.3% margin**, and a **100% markup is only a 50% margin**.
> - To get a margin of M%, price at **cost ÷ (1 − M)**. Adding M% to the cost gives a lower margin than you wanted.
> - A discount hits margin much harder than it looks: **10% off on a 25% margin product cuts profit by 40%**.

## Same profit, two different percentages

Take a product that costs you ₹600 and that you sell for ₹1,000. Your profit is ₹400. Now ask what percentage that profit is, and you have to choose what to divide by.

**From selling price to profit (₹ per piece)**

| Step | Amount |
| --- | --- |
| Selling price | ₹1,000 |
| Cost of the item | −₹600 |
| Profit | ₹400 |

*The profit of ₹400 is 66.7% of the cost (₹600) but only 40% of the selling price (₹1,000).*

**The same ₹400 profit**

| Markup | Margin |
| --- | --- |
| 66.7% | 40% |
| ₹400 ÷ ₹600 cost | ₹400 ÷ ₹1,000 price |
| Answers: how much did I add on top of what I paid? | Answers: how much of each sale rupee do I keep? |
| Common in trading and wholesale | Common in accounts, reports and investor decks |

## The two formulas

> **Markup % = (price − cost) ÷ cost × 100      Margin % = (price − cost) ÷ price × 100:** The top line of both is the same profit. Only the bottom line changes. Because the price is always bigger than the cost, dividing by the price always gives the smaller percentage.

You can convert one to the other without knowing any rupee amounts. Margin = markup ÷ (100 + markup). Markup = margin ÷ (100 − margin). The second formula explodes as margin rises: a 90% margin means a 900% markup.

**Margin you actually get for each markup**

|  | Markup | Margin |
| --- | --- | --- |
| Markup 10% | 10% | 9.1% |
| Markup 25% | 25% | 20% |
| Markup 50% | 50% | 33.3% |
| Markup 100% | 100% | 50% |
| Markup 200% | 200% | 66.7% |

*Each pair is the same sale. Margin climbs slowly while markup shoots up. At 100% markup you double your cost and keep exactly half of each sale.*

Try it: [Markup & Margin Calculator](https://toolcloak.pages.dev/tools/markup-margin-calculator). Enter a cost and a markup or margin to get the selling price and the other percentage.

## The pricing mistake that costs real money

Suppose you want a 30% margin on an item that costs ₹700. There are two ways to price it, and only one is right.

**Where each ₹100 of price goes**

|  | Cost | Profit |
| --- | --- | --- |
| Add 30% to cost (price ₹910) | ₹76.9 cost | ₹23.1 |
| Price for a 30% margin (₹1,000) | ₹70 cost | ₹30 profit |

*₹700 × 1.30 = ₹910 gives a profit of ₹210, which is 23.1% of ₹910. To get 30%, divide the cost by 0.70: ₹700 ÷ 0.70 = ₹1,000.*

On a business that sells ₹10 lakh a month, that gap is ₹69,000 of profit every month, just because the wrong word was used in the pricing sheet.

**Selling price needed for each margin on a ₹700 cost**

| Margin | Price | Profit | Markup needed |
| --- | --- | --- | --- |
| 20% | ₹875 | ₹175 | 25% |
| 25% | ₹933 | ₹233 | 33.3% |
| 30% | ₹1,000 | ₹300 | 42.9% |
| 40% | ₹1,167 | ₹467 | 66.7% |
| 50% | ₹1,400 | ₹700 | 100% |

## Why a discount bites harder than you think

A product costs ₹75 and sells at ₹100, a 25% margin and ₹25 profit. Offer 10% off. The price becomes ₹90, the cost is still ₹75, so profit falls from ₹25 to ₹15. The margin falls to 16.7%, and the profit drops by 40%.

**Profit per ₹100 item as the discount grows**

|  | Value |
| --- | --- |
| No discount | ₹25 |
| 5% off | ₹20 |
| 10% off | ₹15 |
| 15% off | ₹10 |
| 20% off | ₹5 |
| 25% off | ₹0 |

*At a 25% margin, a 25% discount means selling at cost. To make up for 10% off you need to sell about two thirds more units (25 ÷ 15 = 1.67).*

Stacked offers are worse still. See why in [50% off plus 20% off is not 70% off](https://toolcloak.pages.dev/articles/50-percent-off-plus-20-percent-off-is-not-70-percent-off).

## Gross margin, net margin and where GST fits

- **Gross margin** counts only the direct cost of the goods. It ignores rent, salaries and electricity.
- **Net margin** subtracts every cost, including tax. It is the share of each sale you really keep.
- **GST is not your revenue.** If you charge GST, take it out before you work out margin. [Adding or removing GST](https://toolcloak.pages.dev/articles/how-to-add-or-remove-gst-without-getting-it-wrong) is a separate step, done on the price first.
- **Losses show up in both.** A ₹100 item sold for ₹80 has a margin of −25% (loss ÷ price) and a markup of −20% (loss ÷ cost). Same ₹20 loss, two different percentages.

> **Check the word before you compare:** When a supplier says “we work on 40%”, ask which one. A 40% markup is a 28.6% margin. A 40% margin is a 66.7% markup. Same sentence, very different business.

## A quick routine for pricing any product

1. Write down the full cost per piece, including packing and delivery.
2. Decide the **margin** you want on the selling price, not a markup.
3. Divide the cost by (1 − margin) to get the price.
4. Add GST on top if you are registered, and check the result against what rivals charge.
5. Test a discount before you offer it, using the profit-per-piece figure and not the percentage.

Keep markup for talking with traders who use it. Keep margin for judging whether the business is healthy.

## Frequently asked questions

### What is the difference between markup and margin?

Markup is profit divided by cost. Margin is profit divided by selling price. The profit is the same, but margin is always the smaller percentage because the selling price is bigger than the cost.

### How do I convert markup to margin?

Margin = markup ÷ (100 + markup). For example a 50% markup is 50 ÷ 150 = 33.3% margin.

### How do I convert margin to markup?

Markup = margin ÷ (100 − margin). A 25% margin is 25 ÷ 75 = 33.3% markup.

### How do I set a price for a target margin?

Selling price = cost ÷ (1 − margin). For a ₹700 cost and a 30% margin, price = 700 ÷ 0.70 = ₹1,000.

### Can markup be more than 100%?

Yes. A product that costs ₹100 and sells for ₹300 has a 200% markup. Margin can never be above 100%.

### Is gross margin the same as net margin?

No. Gross margin subtracts only the direct cost of the goods. Net margin subtracts every cost, including overheads and tax.



---

# Mean vs median vs mode: why “the average salary” is so often misleading

> A company of nine people announces an average salary of ₹11.5 lakh. Eight of them earn between ₹4 and ₹7 lakh. Nobody is lying; the word “average” is. This guide explains the mean, median and mode, shows how one big number bends the mean, and gives a simple rule for picking the right one.

Published 2026-10-08 by Toolcloak. 6 minute read. Original: https://toolcloak.pages.dev/articles/mean-vs-median-vs-mode-why-the-average-salary-is-misleading
Related free tool: [Statistics Calculator](https://toolcloak.pages.dev/tools/statistics-calculator)

Here is a job advertisement: “Our team of nine earns an average of ₹11.5 lakh a year.” It sounds great until you look at the payslips. Eight people earn between ₹4 and ₹7 lakh. The ninth, the founder, takes ₹60 lakh. The ad is technically true, and still misleading. The trick is that “average” can mean three different things, and they can disagree wildly.

> **In short**
> - **Mean** = add everything up and divide by the count. **Median** = the middle value when sorted. **Mode** = the value that appears most often.
> - In our example the mean is **₹11.5 lakh**, the median **₹5.5 lakh** and the mode **₹5 lakh**. The one big salary pulled only the mean.
> - When data is **skewed** or has **outliers** (incomes, house prices, response times), the **median** describes a typical person better.
> - Use the **mean** for fairly symmetric data and for totals, the **median** for typical values, and the **mode** for categories like shoe sizes.

## The three averages in one minute

All three try to answer the same question, “what is a typical value?”, in different ways. Take our nine salaries in lakh rupees per year: 4, 4.5, 5, 5, 5.5, 6, 6.5, 7 and 60.

**The nine salaries (₹ lakh a year)**

|  | Value |
| --- | --- |
| A | 4 L |
| B | 4.5 L |
| C | 5 L |
| D | 5 L |
| E | 5.5 L |
| F | 6 L |
| G | 6.5 L |
| H | 7 L |
| Founder | 60 L |

*Eight bars are close to each other. The ninth is almost ten times taller. The mean has to reflect all nine, so it is dragged toward the tall one.*

**Three answers to “what is typical?”**

|  | Value | Note |
| --- | --- | --- |
| Mean | ₹11.5 L | 103.5 ÷ 9. Every value counts, including the 60. |
| Median | ₹5.5 L | The 5th of 9 sorted values. Half earn more, half less. |
| Mode | ₹5 L | The most common value. Two people earn exactly 5. |

*The mean is more than double the median. A new hire expecting ₹11.5 lakh would be disappointed.*

## How to find each one

### Mean

Add up all the values and divide by how many there are. Here 4 + 4.5 + 5 + 5 + 5.5 + 6 + 6.5 + 7 + 60 = 103.5, and 103.5 ÷ 9 = 11.5. The mean uses every number, which is its strength (it is fair to all data) and its weakness (it is easily dragged around by one extreme number).

### Median

Sort the values and pick the middle one. With nine values the middle is the fifth, which is 5.5. If the count is even, there are two middle values, and you average them. For 3, 5, 7, 9 the median is (5 + 7) ÷ 2 = 6. The median ignores how large the extreme values are. It only cares about their position.

### Mode

The value that appears most often. In our data 5 appears twice and everything else once, so the mode is 5. A set can have no mode (all values different) or several. The mode is the only one of the three that works for categories: the most common shirt size, the most popular bus route.

**Finding the median step by step**

1. **Sort the numbers.** 4, 4.5, 5, 5, 5.5, 6, 6.5, 7, 60.
2. **Count them.** There are 9 values. The middle position is (9 + 1) ÷ 2 = 5.
3. **Read the 5th value.** It is 5.5. If the count were even, average the two middle ones.

Try it: [Statistics Calculator](https://toolcloak.pages.dev/tools/statistics-calculator). Paste a list of numbers to get the mean, median, mode, range and standard deviation at once.

## How one number bends the mean

Let us do an experiment. Keep the other eight salaries fixed and change only the ninth. As the top salary grows, the mean rises steadily. The median does not move at all, because the top person is still the top person and the middle is still the fifth value.

**Mean and median as the top salary grows (₹ lakh)**

| The ninth person’s salary (₹ lakh) | Mean | Median |
| --- | --- | --- |
| 10 | 5.9 | 5.5 |
| 20 | 7.1 | 5.5 |
| 40 | 9.3 | 5.5 |
| 60 | 11.5 | 5.5 |
| 100 | 15.9 | 5.5 |

*With a top salary of 10 the two are close (5.9 and 5.5). At 100, the mean is 15.9 and the median is still 5.5. This is why the median is called a robust measure.*

**From the median to the mean in our example (₹ lakh)**

| Step | Amount |
| --- | --- |
| Median | 5.5 |
| Pull of the founder | +6.0 |
| Mean | 11.5 |

*The gap between the two (6 lakh) is a quick signal that the data is lopsided. When the mean is far above the median, a few big values are at work.*

## Skew: when the three averages disagree

In a perfectly symmetric bell curve all three averages sit at the same point. In a **right-skewed** distribution, which has a long tail of big values, the order is usually mode, then median, then mean, from left to right. Incomes, house prices, hospital stays and the time to fix a bug all look like this. Few values are huge, and many are small to medium.

**A right-skewed distribution: the three averages in order**

| Value (arbitrary units) | How common each value is |
| --- | --- |
| 0 | 0 |
| 0.5 | 30.3 |
| 1 | 36.8 |
| 1.5 | 33.5 |
| 2 | 27.1 |
| 2.5 | 20.5 |
| 3 | 14.9 |
| 3.5 | 10.6 |
| 4 | 7.3 |
| 4.5 | 5 |
| 5 | 3.4 |
| 5.5 | 2.2 |
| 6 | 1.5 |

*Drawn from a standard gamma curve: mode 1, median about 1.68, mean 2. The long right tail pulls the mean furthest. If the mean is above the median, suspect a right skew.*

**Two honest sentences about the same data**

| “The average salary is ₹11.5 lakh” | “The typical salary is ₹5.5 lakh” |
| --- | --- |
| The mean | The median |
| True, and technically correct | Half the team earns more, half less |
| Raised by the founder’s ₹60 lakh | Not moved by the founder’s pay |
| Describes nobody in the room | Matches what a new hire will likely see |

*Whenever you read a headline with “average”, ask: mean or median?*

## Which one should you use?

**A simple guide to choosing**

| Box | What it means |
| --- | --- |
| Use the mean when… | The data is roughly symmetric, without wild outliers, and you care about totals. Example: average daily sales over a month to forecast the revenue. |
| Use the median when… | The data is skewed or has outliers, and you want a typical case. Examples: salaries, house prices, waiting times, delivery delays. |
| Use the mode when… | The data is in categories, or you want the most common choice. Examples: shirt sizes, most-used payment method, the busiest hour. |
| Show more than one when… | You are not sure. Giving mean and median together tells readers how lopsided the data is. |

*Good reports give two numbers, such as “median ₹5.5 lakh, mean ₹11.5 lakh”. The gap itself is information.*

## Other kinds of average you will meet

### The weighted mean

Sometimes values do not count equally. A course gives 20% to assignments, 30% to a midterm and 50% to the final. You scored 80, 60 and 70. The plain mean is (80 + 60 + 70) ÷ 3 = 70. The weighted mean is 0.2 × 80 + 0.3 × 60 + 0.5 × 70 = 16 + 18 + 35 = **69**. The final counts the most, so it pulls harder.

**Plain mean against weighted mean for the same three marks**

|  | Score |
| --- | --- |
| Assignments (20%) | 80 |
| Midterm (30%) | 60 |
| Final (50%) | 70 |
| Plain mean | 70 |
| Weighted mean | 69 |

*The weights decide the result. A grade calculator or the statistics tool above does this arithmetic for you.*

### The trimmed mean

Drop the highest and lowest values before averaging. Judges in diving and gymnastics do this to blunt a biased score. For our nine salaries, drop 4 and 60 and the mean of the remaining seven is 39.5 ÷ 7 ≈ **5.6 lakh**, very close to the median.

### The geometric mean

For growth rates you multiply rather than add, so the right average is the geometric mean. A fund that gains 50% one year and loses 50% the next has an arithmetic mean of 0%, but your money is down: 1.5 × 0.5 = 0.75. The true yearly rate is √0.75 − 1 ≈ −13.4%. The articles on [up 50 percent, down 50 percent](https://toolcloak.pages.dev/articles/up-50-percent-down-50-percent-and-you-are-not-back-to-zero) and [average return versus CAGR](https://toolcloak.pages.dev/articles/average-return-vs-cagr-why-20-percent-is-not-what-you-earned) walk through this trap.

## Five ways averages mislead

1. **The outlier hides in the mean.** One huge value shifts it. Always check the median.
2. **Two groups hiding in one.** If data has two clusters (say, junior and senior staff), no single average describes either group. Split it first.
3. **Averages of averages.** The mean of two city averages is not the national average unless the cities are the same size. Use a weighted mean.
4. **Percentages are not added.** Average the underlying counts, not the percentages.
5. **The mean of nothing.** The average family has 2.4 children, but no family does. The mean need not be a possible value, which is fine for planning and silly for describing.

> **A habit worth building:** When you see an average, ask three questions: which kind is it, how many values are behind it, and what is the spread? For the spread, see the [bell curve and standard deviation](https://toolcloak.pages.dev/articles/the-bell-curve-68-95-99-7-rule-z-scores-and-what-rare-really-means) guide.

## Quick practice

Five friends spend ₹200, ₹250, ₹300, ₹300 and ₹1,950 on a trip dinner. The mean is 3,000 ÷ 5 = ₹600, the median ₹300, and the mode ₹300. Which should the restaurant quote as the typical bill? The median, because one person ordered an expensive meal. Which should the group use to split the total fairly? Neither: the total of ₹3,000 is what matters, so the mean times five is the figure to split. The same data, two different questions, two different averages.

## Frequently asked questions

### What is the difference between mean, median and mode?

The mean is the sum divided by the count, the median is the middle value of the sorted list, and the mode is the most frequent value.

### Why is the median better for salaries?

A few very high salaries pull the mean up but do not change the middle value, so the median shows what a typical person earns.

### Can the mean be higher than every value except one?

Yes. With eight values near 5 and one value of 60, the mean is 11.5, which is above eight of the nine values.

### What if there are two middle values?

When the count is even, the median is the average of the two middle values. For 3, 5, 7, 9 it is 6.

### Can a data set have more than one mode?

Yes. If two values tie for the most frequent, the data is bimodal. If no value repeats, it has no mode.

### What is a weighted average?

An average in which some values count more than others, found by multiplying each value by its weight and adding. Grades often use it.



---

# Morse code: why E is a single dot and Q takes four signs

> Morse code looks like random dots and dashes, but it is a carefully designed compression scheme: common letters get short codes, rare letters get long ones. This guide shows the timing rules, the maths of why it works, and how fast operators really send.

Published 2026-10-08 by Toolcloak. 4 minute read. Original: https://toolcloak.pages.dev/articles/morse-code-why-e-is-one-dot-and-q-is-four-signs-long
Related free tool: [Morse Code Translator](https://toolcloak.pages.dev/tools/morse-code-translator)

Every letter of the alphabet can be sent down a wire using just two signals: a short one (dot) and a long one (dash). Morse code was built for the telegraph in the 1800s and is still used by radio amateurs, pilots checking a beacon and people who like puzzles. What makes it clever is not the dots and dashes. It is the order in which the codes were handed out.

> **In short**
> - A **dash lasts 3 dots**. The gap between signs is 1 dot, between letters 3 dots and between words 7 dots.
> - Common letters get short codes: **E is one dot, T is one dash**. Rare ones are long: Q is `--.-`.
> - That design makes average English about **19% shorter** to send than if the same codes were handed out in alphabetical order.
> - **SOS** is `... --- ...`: 27 dot-lengths of signal. It does not stand for any words.

## The whole alphabet

International Morse code has 26 letters, 10 digits and a handful of punctuation marks. Here are all the letters. Look for the pattern: the shortest codes belong to letters you meet all the time.

| Letter | Code | Letter | Code | Letter | Code |
| --- | --- | --- | --- | --- | --- |
| A | .- | B | -... | C | -.-. |
| D | -.. | E | . | F | ..-. |
| G | --. | H | .... | I | .. |
| J | .--- | K | -.- | L | .-.. |
| M | -- | N | -. | O | --- |
| P | .--. | Q | --.- | R | .-. |
| S | ... | T | - | U | ..- |
| V | ...- | W | .-- | X | -..- |
| Y | -.-- | Z | --.. |  |  |

*Dot is a short signal, dash a long one. Digits are always five signs: 1 is `.----`, 5 is `.....` and 0 is `-----`.*

Try it: [Morse Code Translator](https://toolcloak.pages.dev/tools/morse-code-translator). Type any sentence and see it as dots and dashes, or paste Morse and read it back as text.

## Time, not symbols, is what counts

A dash is not just “a different symbol”. It is a signal held three times as long as a dot, and the silences have exact lengths too. Without them, `...---...` could not be told apart from other letter splits. All timing is counted in one basic unit, the length of a dot.

**How long each part lasts, in dot units**

|  | Units |
| --- | --- |
| Dot | 1 |
| Dash | 3 |
| Gap inside a letter | 1 |
| Gap between letters | 3 |
| Gap between words | 7 |

*Everything else follows from these five numbers. Skilled operators can hear the gaps as clearly as the signs.*

**Building SOS, piece by piece (dot units)**

| Step | Amount |
| --- | --- |
| S (. . .) | 5 |
| Letter gap | +3 |
| O (- - -) | +11 |
| Letter gap | +3 |
| S (. . .) | +5 |

*S is three dots plus two gaps: 1+1+1+1+1 = 5. O is three dashes plus two gaps: 3+1+3+1+3 = 11. Total 27 units.*

> **SOS is a pattern, not an abbreviation:** SOS was chosen as a distress signal because `... --- ...` is simple and hard to mistake, not because the letters stand for “Save Our Souls”. Operators send it as one run with no letter gaps, so it sounds like a single rhythm.

## Why common letters get short codes

Imagine a telegraph operator who has to send thousands of words a day. If the most frequent letter, E, takes 1 unit and a rare one like Q takes 13, the rare letters barely matter because they hardly appear. The result is a shorter message overall. This is the same idea behind modern file compression.

**Time to send each of the ten most common letters (dot units)**

|  | Value |
| --- | --- |
| E | 1 |
| T | 3 |
| A | 5 |
| O | 11 |
| I | 3 |
| N | 5 |
| S | 5 |
| H | 7 |
| R | 7 |
| D | 7 |

*E takes 1 unit, T and I take 3, A, N and S take 5. Even the longest of these, O at 11, is not long.*

**Time to send the five rarest letters (dot units)**

|  | Value |
| --- | --- |
| Q | 13 |
| J | 13 |
| X | 11 |
| Z | 11 |
| K | 9 |

*Q and J take 13 units each. They appear in about one letter in a thousand, so the cost barely adds up.*

**Average time per letter of English, by how the codes are assigned (dot units)**

|  | Units |
| --- | --- |
| Real Morse (by frequency) | 6.1 |
| Same codes, alphabetical order | 7.5 |
| Worst case: rare letters short | 10.6 |

*Weighted by typical English letter frequencies (E 12.7%, T 9.1% and so on). The same set of codes can be about 1.7 times slower if handed out the wrong way round.*

## How fast is “fast”?

Speed is measured in words per minute, using the word PARIS as the standard. PARIS plus the gap after it takes exactly 50 units, so at 1 WPM a dot lasts 1.2 seconds. The faster the speed, the shorter the unit.

**Length of one dot at each speed (milliseconds)**

|  | Value |
| --- | --- |
| 5 | 240 ms |
| 10 | 120 ms |
| 15 | 80 ms |
| 20 | 60 ms |
| 25 | 48 ms |
| 30 | 40 ms |

*Labels are words per minute. At 20 WPM a dot is 60 ms and a dash 180 ms. Beginners usually learn at 5 to 10 WPM; fast operators pass 30.*

**Sending a few real words**

|  | Value | Note |
| --- | --- | --- |
| SOS | 27 units | About 1.6 s at 20 WPM |
| HELLO | 49 units | About 2.9 s at 20 WPM |
| PARIS (standard word) | 43 + 7 | Exactly 50 with the word gap |

*Seconds are units × 0.06 at 20 WPM. Longer words take longer, but not in proportion to their letter count.*

**Reading Morse by ear, the way operators learn**

1. **Hear the rhythm, not the dots.** Learn each letter as a sound pattern. “Di-dah” is A, not “dot dash”.
2. **Listen for the gaps.** A longer pause means a new letter. A much longer pause means a new word.
3. **Copy at a comfortable speed.** Start at 15 WPM with long gaps between letters, then shrink the gaps.

## Where Morse still shows up

- **Amateur radio**: many licence holders still use it, because a narrow Morse signal can be heard when voice cannot.
- **Navigation beacons**: some radio beacons identify themselves with a short Morse code.
- **Accessibility**: some people with limited mobility can type using two switches and Morse.
- **Phones and watches**: a vibration or light flash can send a message when you cannot speak.

> **Learn it faster:** Do not memorise a chart of dots and dashes. Learn the sound of each letter at full speed and practise for ten minutes a day. Most people can read the full alphabet in a few weeks this way.

The idea of giving the shortest codes to the most common symbols is exactly what compression formats do. For a related story about encoding text, read about [Base64](https://toolcloak.pages.dev/articles/base64-is-not-encryption-and-it-makes-files-a-third-bigger), and for another simple cipher you can crack by counting letters, see the [Caesar cipher](https://toolcloak.pages.dev/articles/caesar-cipher-why-only-25-keys-means-anyone-can-crack-it).

## Frequently asked questions

### What does SOS mean in Morse code?

SOS is `... --- ...`. It is a distress signal chosen because the pattern is simple and unmistakable. It does not stand for any words.

### How long is a dash compared with a dot?

Three times as long. The gap between signs of one letter is one dot, between letters three dots and between words seven dots.

### Why is E just one dot?

E is the most common letter in English, so it was given the shortest possible code. T, the second most common, gets the single dash.

### How do you write numbers in Morse code?

Every digit is five signs long: 1 is `.----`, 2 is `..---`, 3 is `...--`, 4 is `....-`, 5 is `.....`, 6 is `-....`, 7 is `--...`, 8 is `---..`, 9 is `----.` and 0 is `-----`.

### How fast can people send Morse code?

Beginners start at 5 to 10 words per minute. Experienced operators send 20 to 30, and top competitors reach well over 40.

### Is Morse code still used?

Yes, mainly by amateur radio operators, in some navigation beacons and as an accessibility input method.



---

# The bell curve: the 68-95-99.7 rule, z-scores and what “rare” really means

> Heights, exam marks, delivery times and measurement errors all tend to pile up in the same bell shape. Once you know the mean and the spread, three numbers (68, 95 and 99.7 percent) tell you where almost everything lies, and a z-score tells you how unusual any single value is. Here is the whole idea, with pictures and worked examples.

Published 2026-10-08 by Toolcloak. 7 minute read. Original: https://toolcloak.pages.dev/articles/the-bell-curve-68-95-99-7-rule-z-scores-and-what-rare-really-means
Related free tool: [Normal Distribution Calculator (Probability)](https://toolcloak.pages.dev/tools/normal-distribution-calculator)

Your exam score is 75. The class average is 60. Is that good? It depends on how spread out the marks are. If everyone scored between 55 and 65, a 75 is outstanding. If marks ranged from 20 to 100, it is merely above average. The bell curve gives you a way to answer this question with a single number, and it works the same for marks, heights, blood pressure, delivery times and machine tolerances.

> **In short**
> - A **normal distribution** is the symmetric bell shape described by two numbers: the **mean** (centre) and the **standard deviation** (typical spread).
> - About **68%** of values lie within 1 standard deviation of the mean, **95%** within 2 and **99.7%** within 3.
> - A **z-score** = (value − mean) ÷ standard deviation. It says how many standard deviations a value is from the centre.
> - A score of 75 when the mean is 60 and the standard deviation is 10 has **z = 1.5**, which is about the **93rd percentile** (better than 93% of people).

## What the bell shape means

Imagine measuring the height of 10,000 adults and stacking the results into bars. Most people are close to the average. Very tall and very short people are rare, and the bars get lower the further you go from the middle. Draw a smooth line over the tops of those bars and you get the famous bell.

The bell appears whenever a result is the sum of many small, independent influences, such as genes, diet and sleep for height, or tiny random errors in a measurement. This is why it shows up everywhere, and why statisticians treat it as the default model for many everyday measurements. It is not universal, though. Incomes, website visits and waiting times are usually skewed, and we will come to that.

**The standard bell curve: how likely each value is (height of the curve)**

| Standard deviations from the mean | Relative likelihood (×100) |
| --- | --- |
| −4 | 0.01 |
| −3 | 0.44 |
| −2 | 5.4 |
| −1 | 24.2 |
| 0 | 39.9 |
| 1 | 24.2 |
| 2 | 5.4 |
| 3 | 0.44 |
| 4 | 0.01 |

*The curve is highest at the mean and falls away equally on both sides. The values are the density of the standard normal curve multiplied by 100 so they fit on the axis. At ±3 standard deviations the curve is almost flat on the floor.*

## The two numbers that describe it

Only two values are needed to draw a normal curve. The **mean** (written μ) says where the centre is. The **standard deviation** (written σ) says how wide the bell is. A small σ gives a tall narrow bell. A large σ gives a low wide one. Same shape, different scale.

**Same mean, different spread**

| Small standard deviation | Large standard deviation |
| --- | --- |
| Tall and narrow | Low and wide |
| Values cluster tightly | Values are scattered |
| A big gap from the mean is unusual | A big gap from the mean is common |
| Typical of precise machines | Typical of mixed groups |

*Two classes can have the same average mark of 60 and very different stories. The standard deviation is what tells them apart.*

> **Standard deviation in plain words:** It is roughly “the typical distance of a value from the average”. If the mean height is 170 cm and the standard deviation is 7 cm, most adults are within about 7 cm of 170, and nearly all are within 21 cm.

## The 68-95-99.7 rule

This rule, also called the empirical rule, is the part worth memorising. Walk out from the mean in steps of one standard deviation and count how much of the data you have captured.

**Share of all values in each band around the mean (percent)**

|  | Within 1σ | Between 1σ and 2σ | Between 2σ and 3σ | Beyond 3σ |
| --- | --- | --- | --- | --- |
| Both sides together | 68.3% | 27.2% | 4.3% | 0.27% |

*68.27% lies within one σ. Another 27.18% lies between one and two σ (so 95.45% within two). A further 4.28% lies between two and three σ (99.73% within three). Only 0.27% is beyond three σ.*

**Out of 100 people, where do they fall?**

| Group | Count |
| --- | --- |
| Within 1σ of the mean | 68 |
| Between 1σ and 2σ | 27 |
| Further than 2σ | 5 |

*The exact figures are 68.3, 27.2 and 4.6, rounded to whole people.*

**Remember these three**

|  | Value | Note |
| --- | --- | --- |
| Within 1 standard deviation | 68% | About 2 in 3 |
| Within 2 standard deviations | 95% | About 19 in 20 |
| Within 3 standard deviations | 99.7% | About 997 in 1,000 |

*Many process checks, from factory quality control to blood-test reference ranges, are built on exactly these bands.*

Try it: [Normal Distribution Calculator (Probability)](https://toolcloak.pages.dev/tools/normal-distribution-calculator). Enter a mean, a standard deviation and a value to find its z-score, percentile and the share above or below it.

## Z-scores: how unusual is this value?

A raw number such as 75 means little without context. A z-score turns it into a universal unit: **how many standard deviations the value is above or below the mean**. After that you can compare things that have nothing in common, like a maths mark and a running time.

> **z = (value − mean) ÷ standard deviation:** Mean 60, standard deviation 10, score 75: z = (75 − 60) ÷ 10 = 1.5. A negative z means below the mean. A z of 0 is exactly average.

**From a raw score to “better than X percent”**

1. **Subtract the mean.** 75 − 60 = 15 marks above average.
2. **Divide by the standard deviation.** 15 ÷ 10 = 1.5. The score is one and a half spreads above the centre.
3. **Look up the percentile.** A z of 1.5 has about 93.3% of the curve below it.
4. **Say it in words.** The score is better than about 93 of every 100 students, or in the top 7%.

**Percentile for common z-scores (share of values below)**

| z-score | Percentile | Share above | Roughly 1 in… |
| --- | --- | --- | --- |
| 0 | 50.0% | 50.0% | 2 |
| 0.5 | 69.1% | 30.9% | 3 |
| 1.0 | 84.1% | 15.9% | 6 |
| 1.5 | 93.3% | 6.7% | 15 |
| 2.0 | 97.7% | 2.3% | 44 |
| 2.5 | 99.4% | 0.6% | 161 |
| 3.0 | 99.9% | 0.13% | 741 |

**What a z-score feels like**

| Zone | From | To |
| --- | --- | --- |
| Very low | -3 | -2 |
| Below | -2 | -1 |
| Typical | -1 | 1 |
| Above | 1 | 2 |
| Very high | 2 | 3 |

*Anything inside −1 to +1 is ordinary. Beyond ±2 is unusual enough that you would notice it. These are common rules of thumb, not strict laws.*

## Worked examples

### 1. An exam

Marks have a mean of 60 and a standard deviation of 10. A score of 80 is two standard deviations above the mean, so z = 2 and about **2.3%** of students did better. A score of 45 has z = −1.5, so it is better than only about **6.7%** of the class.

### 2. IQ scores

IQ tests are designed to give a mean of 100 and a standard deviation of 15. A score of 130 is z = 2, so roughly 2.3% of people score higher. About 95% of people fall between 70 and 130, which is exactly the two-σ band. This is a design choice of the test, not a law of nature.

### 3. A delivery promise

A courier’s average delivery time is 40 minutes with a standard deviation of 8 minutes. If you promise 56 minutes (two σ above the mean), about 97.7% of orders arrive in time and 2.3% are late. Promise 64 minutes (three σ) and only about 0.13% are late. Each extra σ of cushion costs time but cuts late deliveries sharply.

**Share of orders that would be late, by the time you promise (mean 40 min, σ 8 min)**

|  | Late orders |
| --- | --- |
| 40 min (z 0) | 50% |
| 48 min (z 1) | 15.9% |
| 56 min (z 2) | 2.3% |
| 64 min (z 3) | 0.13% |

*Promising the average time means half of the orders are late. Adding a little cushion gives enormous gains at first, then smaller ones.*

## How rare is “rare”?

The further out you go, the faster the odds collapse. People talk about “five sigma” events in physics and “six sigma” in manufacturing. Here is what those numbers mean for the chance of landing beyond that many standard deviations from the mean on either side.

**How often a value lands beyond k standard deviations (both sides)**

| When | What |
| --- | --- |
| 1σ | About 1 in 3. 31.7% of values. Hardly rare at all. |
| 2σ | About 1 in 22. 4.6% of values. You would see this in a class of thirty. |
| 3σ | About 1 in 370. 0.27% of values. Rare enough to check for a cause. |
| 4σ | About 1 in 15,800. 0.0063% of values. A real surprise. |
| 5σ | About 1 in 1.7 million. The usual bar in particle physics for claiming a discovery. |
| 6σ | About 1 in 500 million. The pure normal-curve figure. Quality programmes quote 3.4 defects per million because they allow the mean to drift by 1.5σ. |

*Real-world data often has heavier tails than the bell predicts, so extreme events can happen more often than these numbers suggest.*

## When the bell curve does not fit

The normal curve is a model. It is excellent for many natural measurements and poor for others. If the data has a hard floor at zero and a long tail on the right, such as salaries, house prices, city sizes or the time before a customer calls support, the mean is dragged upward by a few giants and the 68-95-99.7 rule fails.

- **Check the shape first.** A quick histogram will show if the data is symmetric.
- **Be careful with small samples.** Ten values rarely show a clean bell, and the estimated standard deviation is itself uncertain.
- **Beware of outliers.** One mistaken entry can inflate the standard deviation and hide real surprises. The [mean, median and mode](https://toolcloak.pages.dev/articles/mean-vs-median-vs-mode-why-the-average-salary-is-misleading) guide shows how a single value can change everything.
- **Do not confuse “normal” with “healthy” or “good”.** In statistics, normal only describes the shape.

> **The bell curve is not a guarantee:** Saying “99.7% of values are within 3σ” is true only if the data really is normal. For financial returns, internet traffic and insurance claims, the extremes are fatter and more frequent than the bell suggests. Treat the rule as a starting point, not a promise.

## A short checklist

1. Find the **mean** and the **standard deviation** of your data.
2. Compute **z = (value − mean) ÷ σ** for any value you care about.
3. Read the percentile from a table or the calculator above.
4. Check that the data is roughly symmetric before trusting the answer.
5. Report it in plain words: “better than 93%”, “1 in 22”, “within the usual range”.

Rare events sound alarming and ordinary ones sound dull, but the z-score gives both the same fair measure. Pair it with the odds ideas in [can two UUIDs ever be the same](https://toolcloak.pages.dev/articles/can-two-uuids-ever-be-the-same-the-odds-in-pictures), and you have a solid feel for how likely, or unlikely, things really are.

## Frequently asked questions

### What is the 68-95-99.7 rule?

In a normal distribution, about 68% of values fall within one standard deviation of the mean, about 95% within two, and about 99.7% within three.

### What is a z-score?

A z-score is the number of standard deviations a value is from the mean: (value − mean) ÷ standard deviation. Positive means above the mean, negative means below.

### What does a z-score of 2 mean?

The value is two standard deviations above the mean. About 97.7% of values are lower and about 2.3% are higher.

### How do I turn a z-score into a percentile?

Use a standard normal table or a calculator. For example, z = 1 is about the 84th percentile and z = 1.5 is about the 93rd.

### Is all data normally distributed?

No. Heights and measurement errors are close to normal, but incomes, prices and waiting times are usually skewed with long right tails.

### What does six sigma mean?

In pure normal-curve terms it means the nearest limit is six standard deviations away. Six Sigma programmes quote 3.4 defects per million because they allow for a 1.5σ drift in the process mean.



---

# Binary, hex and decimal: counting in bases 2, 10 and 16

> Computers store everything as bits, but people read decimal, and programmers often use hexadecimal as a shortcut. The same number, 202, is 11001010 in binary and CA in hex. This guide shows how to convert between them by hand, why one hex digit equals four bits, and where you meet these systems every day.

Published 2026-10-08 by Toolcloak. 4 minute read. Original: https://toolcloak.pages.dev/articles/binary-hex-and-decimal-counting-in-bases-2-10-and-16
Related free tool: [Number Base Converter](https://toolcloak.pages.dev/tools/number-base)

You count in tens because you have ten fingers. A computer counts in twos because a switch has two states. Hexadecimal, base 16, sits between them as a tidy way for people to write long strings of bits. None of these systems is harder than decimal. They only change what each position is worth.

> **In short**
> - A **base** is how many symbols a number system uses: 10 for decimal, 2 for binary (0 and 1), 16 for hex (0–9 and A–F).
> - In binary each position is worth **double** the one to its right: 128, 64, 32, 16, 8, 4, 2, 1.
> - **One hex digit is exactly four bits**, so a byte (8 bits) is always two hex digits, from 00 to FF.
> - The number **202** is **11001010** in binary and **CA** in hex. All three are the same value.

## Place value: the idea behind every base

In decimal, the number 305 means 3 hundreds, 0 tens and 5 ones. Each place is worth ten times the one on its right. Binary does the same with two. Hex does the same with sixteen.

**What each binary place is worth (8 bits)**

|  | Value |
| --- | --- |
| bit 7 | 128 |
| bit 6 | 64 |
| bit 5 | 32 |
| bit 4 | 16 |
| bit 3 | 8 |
| bit 2 | 4 |
| bit 1 | 2 |
| bit 0 | 1 |

*Every place is double the one to its right. Add up the places that are switched on to get the number. All eight on gives 255.*

## Reading a binary number

Take 11001010. Write the place values above each bit and add the ones that are on.

| Place value | 128 | 64 | 32 | 16 | 8 | 4 | 2 | 1 |
| --- | --- | --- | --- | --- | --- | --- | --- | --- |
| Bit | 1 | 1 | 0 | 0 | 1 | 0 | 1 | 0 |
| Counts? | 128 | 64 | – | – | 8 | – | 2 | – |

*128 + 64 + 8 + 2 = 202.*

**202 as eight switches (filled = on)**

| Group | Count |
| --- | --- |
| On: 128 and 64 | bits 7 and 6 |
| Off: 32 and 16 | bits 5 and 4 |
| On: 8 | bit 3 |
| Off: 4 | bit 2 |
| On: 2 | bit 1 |
| Off: 1 | bit 0 |

## Decimal to binary by dividing by 2

To convert, divide by 2 again and again, and write down each remainder. Read the remainders from the last to the first.

**Converting 202 to binary**

1. **202 ÷ 2 = 101.** Remainder 0. This is the last bit (bit 0).
2. **101 ÷ 2 = 50.** Remainder 1.
3. **50 ÷ 2 = 25 · 25 ÷ 2 = 12.** Remainders 0 and 1.
4. **12 ÷ 2 = 6 · 6 ÷ 2 = 3.** Remainders 0 and 0.
5. **3 ÷ 2 = 1 · 1 ÷ 2 = 0.** Remainders 1 and 1. Stop when you reach 0.
6. **Read from the bottom up.** 11001010, which is 202.

Try it: [Number Base Converter](https://toolcloak.pages.dev/tools/number-base). Type a number in any base and see it in binary, octal, decimal and hex at once.

## Why hex exists

Long strings of ones and zeros are hard for people to read and easy to mistype. Hex is a shortcut. Because 16 is 2⁴, one hex digit stands for exactly four bits. You can convert a binary number to hex by splitting it into groups of four from the right.

| Decimal | Binary | Hex | Decimal | Binary | Hex |
| --- | --- | --- | --- | --- | --- |
| 0 | 0000 | 0 | 8 | 1000 | 8 |
| 1 | 0001 | 1 | 9 | 1001 | 9 |
| 2 | 0010 | 2 | 10 | 1010 | A |
| 3 | 0011 | 3 | 11 | 1011 | B |
| 4 | 0100 | 4 | 12 | 1100 | C |
| 5 | 0101 | 5 | 13 | 1101 | D |
| 6 | 0110 | 6 | 14 | 1110 | E |
| 7 | 0111 | 7 | 15 | 1111 | F |

*Hex uses the letters A to F for the values 10 to 15.*

For 11001010, split into 1100 and 1010. 1100 is C (12) and 1010 is A (10). So the hex is CA. Check it: C × 16 + A = 12 × 16 + 10 = 202.

**The same byte, three ways**

| For people | For machines and programmers |
| --- | --- |
| 202 | 11001010 = 0xCA |
| Decimal, easy to read | Binary shows every bit |
| Hard to see which bits are on | Hex is a short form: two digits per byte |

## How big can a number get?

**The largest value that fits in each size**

|  | Value | Note |
| --- | --- | --- |
| 8 bits (1 byte) | 255 | Hex FF |
| 16 bits | 65,535 | Hex FFFF |
| 32 bits | 4.29 billion | 4,294,967,295 · IPv4 addresses |
| 64 bits | 1.8 × 10¹⁹ | 18,446,744,073,709,551,615 |

*Each extra bit doubles the number of values. n bits can hold 2ⁿ different values, from 0 up to 2ⁿ − 1.*

## Where you meet these every day

- **Web colours.** `#6366f1` is three bytes: 63, 66 and F1 in hex, which are 99, 102 and 241 in decimal. That is red 99, green 102, blue 241. The [contrast checker guide](https://toolcloak.pages.dev/articles/is-your-text-readable-the-contrast-ratio-explained) uses these values.
- **IP addresses.** Each part of 192.168.1.10 is one byte (8 bits). See [what /24 and /26 mean](https://toolcloak.pages.dev/articles/cidr-and-subnets-what-slash-24-slash-26-and-slash-20-really-mean).
- **File permissions.** The chmod number 755 is octal (base 8). Each digit is three bits: 7 is 111 (read, write, execute), 5 is 101 (read and execute).
- **Characters.** The letter A is 65 in decimal, 41 in hex and 01000001 in binary.
- **Hashes.** A SHA-256 hash is 256 bits, shown as 64 hex digits. See [what a hash is](https://toolcloak.pages.dev/articles/what-a-hash-is-and-why-changing-one-letter-changes-everything).

## Quick tips

1. Memorise the place values 1, 2, 4, 8, 16, 32, 64, 128. Everything else is addition.
2. Learn the sixteen four-bit patterns above. Then hex and binary conversions become instant.
3. Write 0x before a hex number in code (0xCA) and 0b before binary (0b11001010) so they are not mistaken for decimal.
4. Check by converting back. If you end up at the number you started with, the conversion is right.

> **Odd-looking numbers are often powers of two:** 256, 512, 1024, 65536. They are 2⁸, 2⁹, 2¹⁰ and 2¹⁶. Once you know that bits double, those numbers stop looking arbitrary.

## Frequently asked questions

### How do I convert binary to decimal?

Write the place values 128, 64, 32, 16, 8, 4, 2, 1 above the bits and add the values where the bit is 1. For 11001010 that is 128 + 64 + 8 + 2 = 202.

### How do I convert decimal to binary?

Divide by 2 repeatedly and write down the remainders. Read the remainders from last to first.

### Why does hexadecimal use letters?

Base 16 needs sixteen symbols. After 0 to 9, the letters A to F stand for 10 to 15.

### Why is one hex digit four bits?

Because 16 = 2⁴. Four bits can hold exactly 16 different values, which one hex digit represents.

### What does 0x mean?

In many programming languages, 0x in front of a number means it is written in hexadecimal, for example 0xFF is 255.



---

# Password entropy: what “bits” really mean, and why length beats symbols

> An 8-character password with every symbol on the keyboard falls in days. A six-word passphrase of plain lowercase words outlasts the age of the universe. The difference is a number called entropy, and it is easy to read once you see it drawn.

Published 2026-10-08 by Toolcloak. 3 minute read. Original: https://toolcloak.pages.dev/articles/password-entropy-bits-guesses-and-why-length-beats-symbols
Related free tool: [Password Entropy Calculator](https://toolcloak.pages.dev/tools/password-entropy-calculator)

Websites love to tell you to add a capital letter, a number and a symbol. They rarely tell you the one thing that matters more: how long it is. Entropy, measured in bits, is the honest score, and it explains why.

> **In short**
> - Entropy = length × log₂(size of the character set). Each extra **bit doubles** the work for an attacker.
> - Going from 8 to 12 characters adds far more than switching from lowercase to all symbols.
> - A random six-word passphrase is about **77 bits**, about as strong as a random 12-character mix of all keys, and far easier to remember.
> - The formula only holds for **randomly chosen** passwords. A human-made “P@ssw0rd!” is far weaker than its length suggests.

## What a bit of entropy means

Entropy counts how many equally likely passwords an attacker would have to try. One bit means 2 possibilities, two bits means 4, ten bits means 1,024. So 40 bits is about a trillion, and 80 bits is a trillion times a trillion. The number grows so fast that every added bit doubles the attacker's work.

> **The formula:** Bits = length × log₂(characters available). Lowercase only has 26 characters, about 4.7 bits each. All printable keys have 94, about 6.55 bits each. A 4-word phrase from a 7,776-word list gives about 12.9 bits per word.

## Length against variety

Here is the surprising part. Look at how many bits each choice buys.

**Bits of entropy for random passwords**

|  | Value |
| --- | --- |
| 8 lowercase | 38 |
| 8 all keys | 52 |
| 12 lowercase | 56 |
| 12 all keys | 79 |
| 16 lowercase | 75 |
| 6 random words | 78 |

*Twelve plain lowercase letters (56 bits) beat eight characters of every kind (52 bits). Sixteen lowercase letters beat almost anything short.*

## How long would it take to crack?

Assume the worst realistic case: an attacker has stolen a database of weakly hashed passwords and can test **10 billion guesses a second** on their own hardware. On average they find the password after trying half of all possibilities.

| Random password | Bits | Average time to crack |
| --- | --- | --- |
| 8 lowercase letters | 37.6 | about 10 seconds |
| 8 letters and digits (62 symbols) | 47.6 | about 3 hours |
| 8 of all 94 keys | 52.4 | about 3.5 days |
| 10 of all 94 keys | 65.5 | about 85 years |
| 12 letters and digits | 71.5 | about 5,000 years |
| 12 of all 94 keys | 78.7 | about 750,000 years |
| 16 lowercase letters | 75.2 | about 69,000 years |
| 6 random words | 77.5 | about 350,000 years |

*Average time = half the possibilities ÷ 10 billion guesses a second. Slow, properly salted hashes make real-life numbers far longer. Fast, unsalted ones can be worse.*

Try it: [Password Entropy Calculator](https://toolcloak.pages.dev/tools/password-entropy-calculator). Type or paste a password to see its bits and the time it would take, all calculated in your browser.

## Where a six-word passphrase sits

**Bits of entropy and what they mean**

| Zone | Range |
| --- | --- |
| Weak | under 40 |
| Fair | 40 to 60 |
| Strong | 60 to 80 |
| Very strong | 80 and up |

*The needle shows a random six-word passphrase. A random 16-character mix of all keys is about 105 bits.*

## Why human-made passwords are weaker than the maths says

The formula assumes every password in the set is equally likely. People are not random. “Password1!” looks like 11 characters with four character types, but attackers try patterns like a word, a capital, a number and a symbol early. Its real strength is closer to a few thousand guesses than a few trillion.

**Looks strong, is not**

| Chosen by a person | Chosen by a generator |
| --- | --- |
| Summer2026! | random 6 words |
| Word + year + symbol is the first pattern tried | Every word is equally likely |
| Falls to a pattern-based attack quickly | Full entropy, easy to type and remember |

*The password generator makes both kinds. Random is what makes the entropy number true.*

## Practical rules

1. **Let a generator choose.** Random beats clever every time. See [Your password is not weak because it is short](https://toolcloak.pages.dev/articles/your-password-is-not-weak-because-it-is-short).
2. **Aim for 70 bits or more** for anything that matters: email, banking, your password manager.
3. **Use a long passphrase** where you must type it from memory, and a long random string everywhere else.
4. **Never reuse.** A strong password leaked from one site is a master key to all the others that share it.
5. **Turn on two-step login.** Entropy protects against guessing. It does not protect against phishing.

## Frequently asked questions

### What is password entropy?

A measure of how unpredictable a password is, in bits. Each bit doubles the number of guesses an attacker needs. It equals length times log₂ of the character set size, for randomly chosen passwords.

### How many bits of entropy is a good password?

At least 60 bits for ordinary accounts and 70 or more for important ones such as email, banking and a password manager.

### Is a longer password better than a complex one?

Usually yes. Twelve random lowercase letters (about 56 bits) beat eight characters of every type (about 52 bits), and length keeps paying off as you add more.

### Are passphrases secure?

Yes, if the words are chosen randomly. A six-word passphrase from a 7,776-word list is about 77.5 bits, about the same as a random 12-character mix of all keys (78.7 bits).

### Does the cracking time apply to my online accounts?

Not directly. Websites limit login attempts, so online guessing is far slower. The cracking-time figures apply when an attacker has a stolen database of hashed passwords.



---

# Up 50 percent, down 50 percent, and you are not back to zero

> A price rises by half and then falls by half. Surely you are where you started? You are not, and the reason catches out shop owners, investors and anyone reading a sale sign.

Published 2026-10-08 by Toolcloak. 3 minute read. Original: https://toolcloak.pages.dev/articles/up-50-percent-down-50-percent-and-you-are-not-back-to-zero
Related free tool: [Percentage Change Calculator](https://toolcloak.pages.dev/tools/percentage-change-calculator)

A share is priced at ₹100. It jumps 50 percent to ₹150. Next month it drops 50 percent. Where is it now? Most people say ₹100. The honest answer is **₹75**, and the missing ₹25 is the whole lesson of this article.

> **In short**
> - Percent changes multiply, they do not cancel: +50% then −50% leaves you 25% down.
> - The bigger the fall, the bigger the climb back: −50% needs +100% to break even.
> - Two discounts never add up: 30% off then 20% off is 44% off, not 50%.

Percentages feel like a fixed unit, like centimetres. They are not. A percentage is always a share **of something**, and the something changes every time the number moves. Once you see that, a handful of strange results stop being strange.

## Why the second 50 percent is bigger

The first 50 percent is taken from ₹100, so it is worth ₹50. The second 50 percent is taken from ₹150, so it is worth ₹75. You gained a little and then gave back a lot, because the thing you were taking half of had become larger.

**₹100 goes up 50 percent, then down 50 percent**

|  | Value |
| --- | --- |
| Start | ₹100 |
| After +50% | ₹150 |
| After −50% | ₹75 |

*The end point is 25 percent below where you started, even though the two moves look like mirror images.*

Mathematically it is a product, not a sum. Going up 50 percent multiplies by 1.5. Going down 50 percent multiplies by 0.5. And 1.5 × 0.5 is 0.75. Percent changes **multiply**, they do not cancel.

Try it: [Percentage Change Calculator](https://toolcloak.pages.dev/tools/percentage-change-calculator). Enter the old and new value to get the exact percentage change, up or down.

## How much you need to climb back

This is the one to remember. If something falls by a percentage, the gain needed to return to the start is always bigger than the fall. The formula is a fall of `x` needs a rise of `x ÷ (100 − x)`.

| You lose | Left with | Gain needed to recover |
| --- | --- | --- |
| 10% | ₹90 of ₹100 | 11.1% |
| 20% | ₹80 of ₹100 | 25% |
| 30% | ₹70 of ₹100 | 42.9% |
| 50% | ₹50 of ₹100 | 100% |
| 75% | ₹25 of ₹100 | 300% |
| 90% | ₹10 of ₹100 | 900% |

*A small fall is nearly symmetrical. A big fall is brutally one-sided. Halving your money means you must double what is left.*

**The gain you need to get back to even**

|  | Needed |
| --- | --- |
| After losing 10% | 11.1% |
| After losing 20% | 25% |
| After losing 30% | 42.9% |
| After losing 50% | 100% |

*The bars grow much faster than the losses do. This is why avoiding big drops matters more than chasing big rises.*

> **Why investors talk about drawdowns:** A fund that falls 50 percent has to **double** just to break even. That is the maths behind the advice to protect against large losses first and chase gains second.

## Percent versus percentage points

A bank raises its rate from 6 percent to 8 percent. News says “up 2 percent”. That is wrong in a way that matters. The rate went up by **2 percentage points**. In relative terms it rose by one third, which is 33 percent.

**Same move, two correct descriptions**

| Percentage points | Percent change |
| --- | --- |
| +2 points | +33% |
| Simple subtraction: 8 − 6 | (8 − 6) ÷ 6 × 100 |
| Use for rates, shares and polls | Use for prices, amounts and sizes |

*Both are true. Trouble starts when a headline says “percent” and means points, or the reverse.*

## Sale signs and stacked discounts

“Extra 20 percent off” on top of “30 percent off” does not make 50 percent off. Each discount applies to the price left by the one before. 100 becomes 70, then 70 × 0.8 = 56. You pay ₹56, which is 44 percent off.

| Offer | You pay on ₹1,000 | Real discount |
| --- | --- | --- |
| 50% off | ₹500 | 50% |
| 30% off, then 20% off | ₹560 | 44% |
| 20% off, then 30% off | ₹560 | 44% |
| 10% off, then 10% off | ₹810 | 19% |

*The order does not matter, because multiplication does not care. But the total is always less than the two numbers added together.*

The same trap works in the other direction. Raise a price by 20 percent and later cut it by 20 percent, and the item costs 4 percent less than it began: 100 × 1.2 × 0.8 = 96.

## The formula, once and for all

> **Percentage change = (new − old) ÷ old × 100:** Always divide by the **old** value, the starting point. A positive result is an increase, a negative result a decrease. If the old value is zero the change is undefined, because you cannot take a share of nothing.

- **Percent change** has a direction. From 80 to 100 is +25 percent. From 100 to 80 is −20 percent.
- **Percent difference** between two values with no “before” divides by their average, so it is the same either way round. Use it when neither number is the starting point.
- **Percent of a number** is simply multiplication. 18 percent of 2,500 is 2,500 × 0.18 = 450.

## Frequently asked questions

### If something rises 50 percent and then falls 50 percent, where does it end?

At 75 percent of the start. Percent changes multiply: 1.5 × 0.5 = 0.75, a net loss of 25 percent.

### How do I calculate percentage change?

Subtract the old value from the new one, divide by the old value, and multiply by 100. A negative answer means a decrease.

### What gain do I need to recover a 20 percent loss?

25 percent. After a 20 percent fall you hold 80 percent of the original, and 100 ÷ 80 is 1.25.

### What is the difference between percent and percentage points?

Percentage points are plain subtraction between two percentages, like 6 percent to 8 percent being 2 points. Percent change compares them proportionally, which here is 33 percent.

### Do two successive discounts add up?

No. They multiply. 30 percent off then 20 percent off leaves 0.7 × 0.8 = 0.56 of the price, which is 44 percent off, not 50.



---

# Percent vs percentage points: the difference behind half the misleading headlines

> The interest rate rose from 5 percent to 8 percent. Is that a 3 percent rise or a 60 percent rise? Both are true, and they mean very different things. Here is how to tell them apart in ten seconds.

Published 2026-10-08 by Toolcloak. 3 minute read. Original: https://toolcloak.pages.dev/articles/percent-vs-percentage-points-why-the-headline-is-often-wrong
Related free tool: [Percentage Calculator](https://toolcloak.pages.dev/tools/percentage-calculator)

“The loan rate rose by 3 percent.” Did your EMI go up 3 percent, or did the rate go from 5 to 8? Those are not the same story, and the word that separates them is two words long: **percentage points**.

> **In short**
> - A **percentage point** is the plain difference between two percentages. 5% to 8% is **+3 points**.
> - A **percent change** is that difference relative to the start. 5% to 8% is **+60 percent**.
> - Small starting numbers make big percent changes. Big starting numbers make small ones.
> - When a headline says “percent”, ask: of what? Percent of the old number, or points of the same scale?

## Same move, two numbers

**A rate that goes from 5% to 8%**

| Absolute change | Relative change |
| --- | --- |
| +3 percentage points | +60 percent |
| 8 − 5 = 3 | (8 − 5) ÷ 5 = 0.6 |
| Tells you the size of the move on the scale | Tells you how big the move is compared with where it started |

*Neither is wrong. They answer different questions.*

## The table to keep in your head

| From | To | Change in points | Change in percent |
| --- | --- | --- | --- |
| 2% | 3% | +1 point | +50% |
| 5% | 8% | +3 points | +60% |
| 6.5% | 6.0% | −0.5 points | −7.7% |
| 18% | 28% | +10 points | +55.6% |
| 40% | 50% | +10 points | +25% |
| 90% | 99% | +9 points | +10% |

*Same unit, two different sizes. Compare the 18 to 28 row with the 40 to 50 row: both +10 points, very different in percent.*

**The same +10 percentage points, as a percent change**

|  | Value |
| --- | --- |
| 5→15 | +200% |
| 18→28 | +55.6% |
| 40→50 | +25% |
| 80→90 | +12.5% |

*Each step is exactly 10 points. The lower the starting point, the bigger the percentage.*

Try it: [Percentage Calculator](https://toolcloak.pages.dev/tools/percentage-calculator). Work out a percent of a number, the percent difference between two numbers, and the percent change.

## Why headlines use the bigger number

“Risk doubles” sounds more dramatic than “risk goes from 1 in 1,000 to 2 in 1,000”. Both describe the same thing. The first is a 100 percent relative rise. The second is 0.1 points. If you want to know whether it matters to you, look for the starting number.

**“Risk doubles”, from 1 in 1,000 to 2 in 1,000**

|  | Value | Note |
| --- | --- | --- |
| Relative change | +100% | The headline number |
| Absolute change | +0.1 points | From 0.1% to 0.2% |
| Extra people | 1 in 1,000 | One more person per thousand |

*Same fact, three sizes. Only the last two tell you how much it matters to one person.*

## A quick test for any statistic

1. **Is it a share or a rate?** If the quantity is already a percentage (interest, tax, vote share), “percent” can mean two things.
2. **Find the starting number.** Without it, a percent change tells you almost nothing.
3. **Say it both ways.** “Up 3 points, which is 60 percent more” is a full sentence.
4. **Check which direction.** Down 50 percent then up 50 percent does not return you to the start. See [up 50, down 50](https://toolcloak.pages.dev/articles/up-50-percent-down-50-percent-and-you-are-not-back-to-zero).

> **In your own money:** A loan rate cut from 8.5% to 8.25% is **0.25 points**, or about **3 percent** relative. On a ₹50 lakh loan that is about ₹12,500 less interest in the first year. Small points, real rupees.

## Frequently asked questions

### What is the difference between percent and percentage points?

A percentage point is the simple difference between two percentages: 5% to 8% is 3 points. A percent change compares that difference to the starting value: 3 divided by 5 is 60 percent.

### If interest rates rise from 5% to 6%, how much did they rise?

By 1 percentage point, which is a 20 percent relative increase.

### Why do the media use percent instead of points?

Relative percentages often sound larger. Always check the starting value to see the real size of the change.

### Can a percentage point change be negative?

Yes. A fall from 6.5% to 6.0% is −0.5 percentage points, or about −7.7 percent.

### How do I calculate percent change?

Subtract the old value from the new one, divide by the old value, and multiply by 100.



---

# Permutations vs combinations: when order matters, and why a combination lock is misnamed

> Counting arrangements and selections sounds like school maths, but it decides lottery odds, password strength and how many handshakes happen in a room. This guide builds both formulas from scratch and shows the numbers in pictures.

Published 2026-10-08 by Toolcloak. 4 minute read. Original: https://toolcloak.pages.dev/articles/permutations-vs-combinations-when-order-matters-and-why-a-combination-lock-is-misnamed
Related free tool: [Permutation & Combination Calculator](https://toolcloak.pages.dev/tools/permutation-combination-calculator)

How many ways can five people sit in a row? How many different teams of three can be chosen from the same five? The numbers are 120 and 10. They sound like the same problem, but one of them cares about **order** and the other does not. That one question, “does order matter?”, is the whole difference between permutations and combinations.

> **In short**
> - **Permutation** = an arrangement where order matters. **Combination** = a selection where it does not.
> - Picking r from n: **nPr = n! ÷ (n−r)!** and **nCr = n! ÷ (r! × (n−r)!)**. Every combination has r! orderings, so nPr = nCr × r!.
> - A lottery that asks for 6 numbers from 49 has **1,39,83,816** combinations. A 4-digit PIN has **10,000**.
> - A “combination lock” is really a **permutation** lock, because 1-2-3 and 3-2-1 are different codes.

## Factorials: the building block

The factorial of n, written n!, is n × (n−1) × (n−2) × … × 1. It counts how many ways n different things can be lined up. With three letters A, B, C there are 3 × 2 × 1 = 6 lineups, because you have 3 choices for the first spot, 2 left for the second and 1 for the last.

**How fast n! grows**

|  | Value |
| --- | --- |
| 1! | 1 |
| 2! | 2 |
| 3! | 6 |
| 4! | 24 |
| 5! | 120 |
| 6! | 720 |
| 7! | 5040 |

*1, 2, 6, 24, 120, 720, 5,040. By 10! the number is already 3,628,800 and by 20! it passes two quintillion.*

**Factorials get big very fast**

|  | Value | Note |
| --- | --- | --- |
| 10! | 3,628,800 | Ways to order ten items |
| 25! | 1.6 × 10²⁵ | About 15 trillion trillion orderings |
| 52! | 8 × 10⁶⁷ | Orderings of a shuffled deck: almost certainly never repeated |

*Any properly shuffled deck of cards is very likely to be in an order that has never existed before in history.*

Try it: [Permutation & Combination Calculator](https://toolcloak.pages.dev/tools/permutation-combination-calculator). Enter n and r and get nPr and nCr, including the factorial steps.

## Permutations: order matters

Imagine a race with 5 runners where gold, silver and bronze go to different people. The first medal can go to 5 runners, the second to the 4 left and the third to the remaining 3: 5 × 4 × 3 = 60 ways. That is 5P3.

**Awarding 3 medals among 5 runners**

1. **Gold: 5 choices.** Any of the five runners can win.
2. **Silver: 4 choices.** The gold winner is out, four are left.
3. **Bronze: 3 choices.** 5 × 4 × 3 = 60 different podiums.

> **nPr = n! ÷ (n − r)!:** For 5P3: 5! ÷ 2! = 120 ÷ 2 = 60. Dividing by (n−r)! cancels the part of the factorial you did not use, here the last two runners.

## Combinations: order does not matter

Now pick a team of 3 from 5 people. The team {Asha, Ben, Chitra} is the same team no matter who was picked first. Each team appears in 3! = 6 different orders inside the 60 podium-style lists, so we divide by 6: 60 ÷ 6 = 10 teams.

**Same five people, three places**

| Permutations (5P3) | Combinations (5C3) |
| --- | --- |
| 60 | 10 |
| Gold, silver, bronze are different roles | Just a team of three, no roles |
| ABC and CBA count as different | ABC and CBA are the same team |
| Use when order or position matters | Use when you only choose a group |

*Every team of three can be arranged in 3! = 6 ways, so 60 ÷ 6 = 10.*

The formula is nCr = n! ÷ (r! × (n − r)!). A useful symmetry: choosing r items to take is the same as choosing n − r items to leave behind, so nCr = nC(n−r). That is why 5C3 = 5C2 = 10.

**Ways to choose k people from 20 (20Ck)**

|  | Value |
| --- | --- |
| 0 | 1 |
| 2 | 190 |
| 4 | 4845 |
| 6 | 38760 |
| 8 | 125970 |
| 10 | 184756 |
| 12 | 125970 |
| 14 | 38760 |
| 16 | 4845 |
| 18 | 190 |
| 20 | 1 |

*The counts rise to a peak at k = 10 (1,84,756) and fall symmetrically. The same shape appears in Pascal’s triangle and in the bell curve.*

| Situation | Order matters? | Formula | Example answer |
| --- | --- | --- | --- |
| Podium of 3 from 5 runners | Yes | 5P3 | 60 |
| Team of 3 from 5 people | No | 5C3 | 10 |
| 4-digit PIN, digits can repeat | Yes | 10⁴ | 10,000 |
| 4-digit code, no repeated digit | Yes | 10P4 | 5,040 |
| 5-card poker hand from 52 | No | 52C5 | 25,98,960 |
| Lottery, 6 numbers from 49 | No | 49C6 | 1,39,83,816 |

*Four quick questions to ask: does order matter, can items repeat, how many are available, and how many are chosen.*

## Four kinds of counting

**Which counting rule to use**

| Box | What it means |
| --- | --- |
| Order matters, no repeats | nPr = n! ÷ (n − r)!. Podiums, seating, racing positions. |
| Order matters, repeats allowed | nʳ. PINs, plates, passwords made from a small set of characters. |
| Order does not matter, no repeats | nCr = n! ÷ (r! (n − r)!). Teams, lotteries, card hands. |
| Order does not matter, repeats allowed | (n + r − 1)Cr. Choosing 3 scoops from 5 flavours when flavours can repeat. |

*Most real problems fall into the first three. The last is rarer but useful for “with replacement” selections.*

## Locks, lotteries and handshakes

A “combination lock” asks for an ordered code, so it is a permutation lock. With four wheels of ten digits and repeats allowed, there are 10 × 10 × 10 × 10 = 10,000 codes. A lottery really is a combination: only the set of numbers counts, not the order in which the balls come out.

**Choosing 6 numbers from 49**

|  | Value | Note |
| --- | --- | --- |
| Possible tickets | 1,39,83,816 | 49C6, order does not matter |
| Your chance with one ticket | 1 in 14 million | About 0.000007% |
| Orders of the same six balls | 720 | 6! orderings that all count as one ticket |

*If order mattered, there would be 49P6 = 10,068,347,520 tickets, which is 720 times more.*

Handshakes are combinations, because a handshake between A and B is the same handshake as between B and A. For n people the number of handshakes is nC2 = n × (n − 1) ÷ 2, which grows much faster than the head count.

**Handshakes if everyone shakes hands once**

|  | Handshakes |
| --- | --- |
| 5 people | 10 |
| 10 people | 45 |
| 20 people | 190 |
| 50 people | 1,225 |
| 100 people | 4,950 |

*Doubling the people from 50 to 100 does not double the handshakes. It multiplies them by about four, from 1,225 to 4,950.*

> **Why this matters for passwords:** An 8-character password using only lowercase letters has 26⁸ = 2,08,82,70,64,576 possibilities. Each extra character multiplies the count by 26, which is why length matters so much. See [password entropy](https://toolcloak.pages.dev/articles/password-entropy-bits-guesses-and-why-length-beats-symbols) for how that becomes “bits”.

## Quick practice

1. How many ways can 4 books be lined on a shelf? 4! = **24**.
2. How many 2-person committees from 6 people? 6C2 = 15, so **15**.
3. How many 3-letter codes from A–Z with no repeated letter? 26P3 = **15,600**.
4. How many ways to pick 4 from 4? 4C4 = **1**, because there is only one group.

Once you can tell whether order matters, the formula follows. Try several n and r values in the calculator and watch how quickly the numbers grow.

## Frequently asked questions

### What is the difference between a permutation and a combination?

In a permutation order matters (ABC differs from CBA). In a combination it does not (ABC and CBA are the same group).

### What is the nCr formula?

nCr = n! ÷ (r! × (n − r)!), the number of ways to choose r items from n when order does not matter.

### What is 0! equal to?

1. There is exactly one way to arrange nothing, and defining 0! = 1 keeps the formulas working.

### How many combinations in a 6/49 lottery?

1,39,83,816, which is 49C6. A single ticket has a 1 in 1,39,83,816 chance.

### Is a combination lock a permutation or a combination?

A permutation. The order of the digits matters, so 1-2-3 and 3-2-1 are different codes.

### How many ways can a deck of 52 cards be shuffled?

52!, about 8 × 10⁶⁷. That is so large that a thoroughly shuffled deck has almost certainly never occurred before.



---

# PPF in pictures: ₹1.5 lakh a year for 15 years, and why the last years do the heavy lifting

> Put the maximum into a PPF every year and you invest ₹22.5 lakh over 15 years. At 7.1 percent you end with about ₹40.7 lakh. More than ₹18 lakh of that is interest, and most of it arrives in the final years.

Published 2026-10-08 by Toolcloak. 3 minute read. Original: https://toolcloak.pages.dev/articles/ppf-15-years-of-1-5-lakh-a-year-what-you-end-up-with
Related free tool: [PPF Calculator](https://toolcloak.pages.dev/tools/ppf-calculator)

PPF is the most boring investment most families own, and that is its charm. A fixed lock-in, a government-set rate, no daily price to watch. What is less obvious is how the money builds: slowly at first, then suddenly.

> **In short**
> - ₹1.5 lakh a year for 15 years at 7.1 percent grows to about **₹40.7 lakh**. You put in ₹22.5 lakh.
> - Interest is about **₹18.2 lakh**, and about half of it is earned in the last four years.
> - The rate is reset by the government every quarter, so 7.1 percent is an assumption, not a promise.
> - Interest and maturity are currently tax-free, which makes the effective return higher than it looks.

## How the balance builds up

This assumes ₹1.5 lakh deposited at the start of each year, interest compounding yearly at 7.1 percent. Real PPF interest is calculated monthly on the lowest balance between the 5th and month end, so deposit timing matters. The picture is close enough to see the shape.

**PPF balance by year (lakh rupees)**

| Year | Balance with interest | What you deposited |
| --- | --- | --- |
| 0 | 0 | 0 |
| 3 | 5.17 | 4.5 |
| 6 | 11.52 | 9 |
| 9 | 19.32 | 13.5 |
| 12 | 28.91 | 18 |
| 15 | 40.68 | 22.5 |

*The gap between the lines is interest. It is thin at year 3 and wide at year 15.*

**Where the ₹40.7 lakh comes from**

|  | Share |
| --- | --- |
| Your deposits | ₹22.5 L |
| Interest earned | ₹18.2 L |

*Interest is about 45 percent of the final amount.*

Try it: [PPF Calculator](https://toolcloak.pages.dev/tools/ppf-calculator). Enter a yearly deposit, the rate and the years to see the maturity amount and the interest.

## Why the last years matter most

Interest is earned on the whole balance. In year 1 the balance is ₹1.5 lakh. In year 15 it is more than ₹36 lakh. The same 7.1 percent rate does very different work on the two.

**Interest earned in a single year**

|  | Value |
| --- | --- |
| Year 1 | ₹0.11 L |
| Year 5 | ₹0.61 L |
| Year 10 | ₹1.48 L |
| Year 15 | ₹2.70 L |

*Interest in year 15 is about 25 times that of year 1, from the same rate.*

## What if you stop at year 15?

At maturity you can withdraw everything, or extend in blocks of five years, with or without fresh deposits. If you extend, the balance keeps compounding. That is often where PPF turns from good to remarkable, because the interest each year is already larger than a full deposit.

**At year 15**

| Withdraw | Extend 5 years, no deposits |
| --- | --- |
| ₹40.7 lakh | about ₹57 lakh |
| Cash in hand, tax-free | Same 7.1% assumed |
| Compounding stops | Locked for 5 more years |

*₹40.68 lakh × 1.071⁵ ≈ ₹57.4 lakh. Rates and rules can change, so recheck before you rely on this.*

## Honest limits

- **The yearly cap.** You cannot put in more than ₹1.5 lakh a year, so PPF alone rarely builds a large corpus.
- **Liquidity.** Partial withdrawals are allowed only after year 7, with limits.
- **Inflation.** At 7.1 percent against 6 percent inflation your real gain is small. See [what will ₹1 lakh buy in 2046](https://toolcloak.pages.dev/articles/what-will-one-lakh-buy-in-2046-inflation-in-pictures).
- **Rules change.** Rate, cap and tax treatment are set by the government. Check the current scheme.

PPF works best as the steady part of a plan, next to something that can grow faster, such as a [monthly SIP](https://toolcloak.pages.dev/articles/5000-rupees-a-month-for-20-years-what-a-sip-calculator-can-tell-you).

## Frequently asked questions

### How much will ₹1.5 lakh a year in PPF become in 15 years?

At an assumed 7.1 percent, about ₹40.7 lakh on a total deposit of ₹22.5 lakh. The actual rate is reset every quarter by the government.

### Is PPF interest taxable?

Under the current rules the interest and the maturity amount are tax-free. Rules can change, so check the latest scheme.

### What is the maximum I can invest in PPF in a year?

₹1.5 lakh across all your PPF deposits in a financial year.

### Can I extend a PPF account after 15 years?

Yes, in blocks of five years, with or without further deposits. The balance continues to earn interest.

### Why does PPF seem slow at first?

Interest is earned on the balance, and the balance is small early on. Most of the interest arrives in the last few years.



---

# Prime numbers: why they thin out, how to test one, and why your bank depends on them

> There are 25 primes below 100 but only 168 below 1,000 and 78,498 below a million. They never run out, yet they get steadily rarer. This guide shows how to test a number by hand, why you only need to check up to its square root, and how a simple fact about primes protects online payments.

Published 2026-10-08 by Toolcloak. 7 minute read. Original: https://toolcloak.pages.dev/articles/prime-numbers-why-they-thin-out-and-why-your-bank-depends-on-them
Related free tool: [Prime Number Checker](https://toolcloak.pages.dev/tools/prime-number-checker)

Is 91 a prime number? Most people pause, think it looks prime, and say yes. It is not: 91 = 7 × 13. Primes are the atoms of arithmetic, and the surprising thing is how often our eyes get them wrong. This guide gives you a reliable way to test any number, a picture of how primes are spread, and a look at the real-world job they do every time you pay online.

> **In short**
> - A **prime** is a whole number greater than 1 whose only divisors are 1 and itself. **2** is the only even prime, and **1 is not prime**.
> - To test **n**, you only need to try dividing by primes up to **√n**. For 1,000,003 that is primes up to 1,000, just 168 of them.
> - Primes **never run out** (Euclid, about 2,300 years ago) but get **rarer**: 25% of numbers below 100 are prime, 7.85% below a million.
> - Online security relies on the fact that **multiplying two big primes is easy and undoing it is not**.

## What makes a number prime

A number is **prime** if it has exactly two divisors: 1 and itself. 7 is prime (only 1 × 7). 9 is not (3 × 3 is another way to build it). Numbers with more divisors are called **composite**. The number 1 has only one divisor, so it is neither prime nor composite, and 2 is the only even prime, since every larger even number is divisible by 2.

Why do mathematicians care? Because of a basic fact: every whole number above 1 can be written as a product of primes in exactly one way (ignoring order). 60 = 2 × 2 × 3 × 5, and no other combination of primes gives 60. Primes are the building blocks, and the multiplication table is built from them.

**The primes from 1 to 100 (dark cells are prime)**

| Numbers | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 |
| --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- |
| 0 + | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 |
| 10 + | 11 | 12 | 13 | 14 | 15 | 16 | 17 | 18 | 19 | 20 |
| 20 + | 21 | 22 | 23 | 24 | 25 | 26 | 27 | 28 | 29 | 30 |
| 30 + | 31 | 32 | 33 | 34 | 35 | 36 | 37 | 38 | 39 | 40 |
| 40 + | 41 | 42 | 43 | 44 | 45 | 46 | 47 | 48 | 49 | 50 |
| 50 + | 51 | 52 | 53 | 54 | 55 | 56 | 57 | 58 | 59 | 60 |
| 60 + | 61 | 62 | 63 | 64 | 65 | 66 | 67 | 68 | 69 | 70 |
| 70 + | 71 | 72 | 73 | 74 | 75 | 76 | 77 | 78 | 79 | 80 |
| 80 + | 81 | 82 | 83 | 84 | 85 | 86 | 87 | 88 | 89 | 90 |
| 90 + | 91 | 92 | 93 | 94 | 95 | 96 | 97 | 98 | 99 | 100 |

| Looks prime… | But it is | Quick tell |
| --- | --- | --- |
| 51 | 3 × 17 | Digits add to 6, so 3 divides it |
| 57 | 3 × 19 | Digits add to 12 |
| 87 | 3 × 29 | Digits add to 15 |
| 91 | 7 × 13 | Needs the 7 times table |
| 119 | 7 × 17 | Needs the 7 times table |

*The digit-sum rule for 3 catches many fakes. 7 and 13 are the usual troublemakers.*

Try it: [Prime Number Checker](https://toolcloak.pages.dev/tools/prime-number-checker). Type any number to find out if it is prime, and see its smallest divisor if it is not.

## How to test a number: stop at the square root

The obvious way to test n is to try dividing by every number up to n. That is slow. Luckily there is a shortcut. If n = a × b, then one of the two factors must be at most √n. (If both were bigger than √n, their product would be bigger than n.) So if no divisor appears up to √n, none will appear after it, and the number is prime.

**Testing whether 1,000,003 is prime**

1. **Find the square root.** √1,000,003 is about 1,000.0015. Only divisors up to 1,000 matter.
2. **Try only primes.** There are 168 primes below 1,000, so at most 168 divisions, not a million.
3. **Stop at the first hit.** If any prime divides evenly, the number is composite. None does here.
4. **Conclude.** 1,000,003 is prime. A shortcut turned a million checks into 168.

*The shortcut gets better with size: the work grows with the square root of the number, not the number itself.*

**Divisions needed by plain trial division (up to √n)**

|  | Largest divisor to try |
| --- | --- |
| n ≈ 1 million | 1,000 |
| n ≈ 1 billion | 31,623 |
| n ≈ 1 trillion | 1,000,000 |
| n ≈ 10¹⁵ | 31.6 million |

*Multiplying n by 1,000 multiplies the work by about 31.6. Even so, this method is hopeless for the 600-digit numbers used in cryptography, which use cleverer tests.*

> **Quick divisibility rules:** By **2**: ends in 0, 2, 4, 6, 8. By **3**: digits add to a multiple of 3. By **5**: ends in 0 or 5. By **11**: alternate digits and subtract; if the answer is 0 or a multiple of 11, it divides. These four cover most composites you will meet.

## Finding all the primes: the sieve

Suppose you want every prime up to 100 at once. An old method called the **Sieve of Eratosthenes**, over two thousand years old, does it without any division at all. Write down the numbers, then repeatedly cross out the multiples of each surviving number. What is left is prime.

**The Sieve of Eratosthenes, up to 100**

1. **List 2 to 100.** Start with every number except 1.
2. **Keep 2, cross out its multiples.** 4, 6, 8, 10 and so on disappear.
3. **Keep 3, cross out its multiples.** 9, 15, 21 and all other odd multiples go.
4. **Keep 5 and 7, repeat.** After 7 you can stop, because 11 × 11 is already bigger than 100.
5. **Read what survived.** The 25 numbers left are exactly the primes below 100.

A computer can run the same sieve on ten million numbers in a fraction of a second. That is how the prime counts in this article were checked.

## Primes never end, but they thin out

Euclid’s proof that there are infinitely many primes is only a few lines: assume there is a largest prime, multiply all primes together and add 1; the result is not divisible by any of them, so there must be a new prime. Beautiful, short and still airtight.

Yet primes become sparser as you climb. In the first hundred numbers, a quarter are prime. In each later hundred, there are fewer. Here is the count for each century up to 1,000.

**Primes in each block of 100 numbers (1 to 1,000)**

|  | Value |
| --- | --- |
| 1–100 | 25 |
| 101–200 | 21 |
| 201–300 | 16 |
| 301–400 | 16 |
| 401–500 | 17 |
| 501–600 | 14 |
| 601–700 | 16 |
| 701–800 | 14 |
| 801–900 | 15 |
| 901–1000 | 14 |

*The counts total 168. The trend is downward, but not smoothly: 16 and then 16 and then 17. Primes are irregular in the small and predictable in the large.*

**How many primes up to x, and what share of numbers that is**

| When | What |
| --- | --- |
| 10 | 4 primes. 2, 3, 5, 7. That is 40% of the numbers. |
| 100 | 25 primes. 25% of the numbers. |
| 1,000 | 168 primes. 16.8%. Roughly one in six. |
| 10,000 | 1,229 primes. 12.3%. Roughly one in eight. |
| 1 million | 78,498 primes. 7.85%. Roughly one in thirteen. |
| 10 million | 664,579 primes. 6.65%. Roughly one in fifteen. |

**Share of numbers that are prime: actual against the 1 ÷ ln x estimate (percent)**

| Numbers up to x | Actual share of primes | Estimate 1 ÷ ln x |
| --- | --- | --- |
| 10 | 40 | 43.43 |
| 100 | 25 | 21.71 |
| 1k | 16.8 | 14.48 |
| 10k | 12.29 | 10.86 |
| 100k | 9.59 | 8.69 |
| 1M | 7.85 | 7.24 |
| 10M | 6.65 | 6.2 |

*The prime number theorem says the share near x is about 1 divided by the natural logarithm of x. The estimate is close and gets relatively better as x grows. Primes thin out, but very slowly.*

## Odd patterns: twin primes and open questions

Primes that differ by 2 are called **twin primes**: (3, 5), (5, 7), (11, 13), (17, 19), (29, 31), (41, 43). Whether there are infinitely many of them is still unproven, as is the **Goldbach conjecture**, which says every even number above 2 is the sum of two primes (4 = 2 + 2, 10 = 3 + 7, 100 = 3 + 97). Computers have checked it for enormous ranges without a counterexample, yet nobody has a proof.

**A few famous prime facts**

| Box | What it means |
| --- | --- |
| Infinitely many primes | Proved by Euclid. There is no biggest prime. |
| Unique factorisation | Every whole number above 1 is a product of primes in exactly one way. |
| Twin primes | Pairs like 11 and 13. Infinitely many? Still an open question. |
| Goldbach | Every even number above 2 is a sum of two primes. Checked widely, unproven. |

*Primes are an area where a child can ask a question that stumps experts.*

## The job primes do for your bank

Here is the practical payoff. Multiplying two primes is easy: 101 × 103 = 10,403 takes a moment. Going the other way, finding the two primes that produce 10,403, takes real work. With small numbers it is just a puzzle. With numbers hundreds of digits long, the best known methods on ordinary computers would need an impractical amount of time.

That one-way difficulty is the heart of **RSA**, one of the systems that protects secure websites. A common key has 2,048 bits, a number of about 617 digits, built from two primes of roughly 309 digits each. Your browser and the server can use the public product freely. Only someone who knows the two primes can reverse the process.

**Easy one way, very hard the other**

| Multiply two big primes | Factor the product back |
| --- | --- |
| Instant | Impractical |
| Two 309-digit primes | No known fast method on classical computers |
| A fraction of a second | Brute force is far beyond practical computing power |
| Produces a 617-digit number | Quantum computers could change this, which is why new methods are being developed |

*Security does not come from hiding the method, only from this gap in effort. See [what a hash is](https://toolcloak.pages.dev/articles/what-a-hash-is-and-why-changing-one-letter-changes-everything) and [password entropy](https://toolcloak.pages.dev/articles/password-entropy-bits-guesses-and-why-length-beats-symbols) for other forms of the same idea.*

**A typical RSA key in numbers**

|  | Value | Note |
| --- | --- | --- |
| Size of the public number | 2,048 bits | About 617 decimal digits |
| Each secret prime | ~309 digits | Two of them, multiplied together |
| Odd candidates that are prime | 1 in ~355 | So about 355 random odd tries finds one |

*Finding the primes is quick. Recovering them from the product, without being told, is what nobody can do in practice.*

How do you find a 309-digit prime in the first place? By trial and error with a fast probabilistic test. Roughly 1 in 710 numbers of that size is prime (1 ÷ ln N), or 1 in about 355 if you skip even numbers. A computer simply picks random odd numbers and tests them until one passes, which takes a fraction of a second.

## Practice and take-aways

1. Is 221 prime? √221 is about 14.9. Try 2, 3, 5, 7, 11, 13. 13 × 17 = 221, so it is **not** prime.
2. Is 97 prime? √97 is about 9.8. Try 2, 3, 5, 7. None divides, so it **is** prime.
3. Write 84 as a product of primes: 84 = 2 × 2 × 3 × 7.
4. Which is the only even prime? **2**, because every other even number is divisible by 2.

Primes combine the simplest definition in mathematics with some of the hardest unsolved questions. They thin out gently, hide behind clever fakes like 91, and quietly keep your payments private. Paste any number in the checker above and see for yourself. If you like number systems, the [binary, hex and decimal](https://toolcloak.pages.dev/articles/binary-hex-and-decimal-counting-in-bases-2-10-and-16) guide is a natural next step.

## Frequently asked questions

### Is 1 a prime number?

No. A prime has exactly two divisors, 1 and itself, and 1 has only one divisor. This convention keeps the unique-factorisation rule true.

### Is 2 prime?

Yes, and it is the only even prime. Every other even number is divisible by 2, so it has at least three divisors.

### How do I check if a number is prime?

Divide it by every prime up to its square root. If none divides evenly, the number is prime.

### How many primes are there below 100?

25: 2, 3, 5, 7, 11, 13, 17, 19, 23, 29, 31, 37, 41, 43, 47, 53, 59, 61, 67, 71, 73, 79, 83, 89 and 97.

### Why are prime numbers used in encryption?

Multiplying two large primes is quick, but recovering them from the product is extremely hard with known methods. Systems such as RSA are built on that gap.

### Are there infinitely many primes?

Yes. Euclid proved it over 2,000 years ago: no list of primes can be complete, because multiplying them and adding 1 produces a number with a prime factor outside the list.



---

# px vs rem: why your font sizes should be allowed to scale

> Pixels are fixed. Rem units follow the reader's own text size. The switch takes one division by 16, and it quietly makes your site usable for everyone who has changed their browser's font size.

Published 2026-10-08 by Toolcloak. 3 minute read. Original: https://toolcloak.pages.dev/articles/px-vs-rem-why-your-font-sizes-should-scale
Related free tool: [PX to REM Converter](https://toolcloak.pages.dev/tools/px-to-rem-converter)

Someone with poor eyesight raises their browser's default text size to 24 pixels. Your page, with every font set in px, ignores it. A page built with rem does not. That is the whole argument, and it fits in one sentence.

> **In short**
> - **rem** = a multiple of the root font size. By default, 1rem = 16px.
> - px to rem: **divide by 16**. 24px is 1.5rem.
> - Use rem for font sizes and spacing so the page respects the reader's settings.
> - Use px for things that should never scale, such as 1px borders.

## What a rem is

A rem is relative to the font size set on the root element (the html tag). Browsers default that to 16px, so 1rem is 16px, 2rem is 32px and 0.5rem is 8px. If a reader changes their browser's text size, the root changes, and everything in rem moves with it.

> **rem = px ÷ root size:** With the default 16px root, 14px is 0.875rem, 20px is 1.25rem and 40px is 2.5rem. If you set the root to something else, divide by that number instead.

**Common pixel sizes in rem (16px root)**

|  | Value |
| --- | --- |
| 12px | 0.75 |
| 14px | 0.875 |
| 16px | 1 |
| 20px | 1.25 |
| 24px | 1.5 |
| 32px | 2 |

*16px, the browser default, is exactly 1rem. Body text usually sits there.*

Try it: [PX to REM Converter](https://toolcloak.pages.dev/tools/px-to-rem-converter). Type a pixel value and a root size and get the rem value, or go the other way.

## What happens when the reader changes their size

**A reader sets their browser text to 24px**

| Your text in px | Your text in rem |
| --- | --- |
| stays 16px | becomes 24px |
| Their setting is ignored | 1rem × 24 = 24px |
| Hard to read for people who set it on purpose | Layout, headings and spacing grow in step |

*Many people change their browser or phone text size on purpose. Respecting it costs you nothing.*

## rem, em and px side by side

| Unit | Relative to | Good for |
| --- | --- | --- |
| px | Nothing: a fixed size | Hairline borders, shadows, icons that must not scale |
| rem | The root font size | Font sizes, margins and padding, media-query widths |
| em | The parent's font size | Spacing that should follow its own element's text, such as button padding |

*The trap with em is nesting: an em inside an em compounds, so 1.2em in a 1.2em parent is 1.44 times the base.*

## The 62.5% trick, and why to skip it

Some developers set `html { font-size: 62.5% }` so that 1rem equals 10px, making the maths easier. It works, but it quietly overrides the user's default and can shrink text for people who raised it. Divide by 16 in your head, or let a tool do it, and leave the root alone.

1. **Leave the root at the default.** Do not set html to a px value.
2. **Write font sizes in rem.** `1rem` for body, `1.25rem` for lead text.
3. **Use px for borders and shadows.** A 1px line should stay 1px.
4. **Test with a bigger default.** Raise your browser's text size once and see what breaks.
5. **Think of the viewport as well.** For fluid headings, try `clamp()` with rem bounds.

## Frequently asked questions

### How do I convert px to rem?

Divide the pixel value by the root font size, which is 16 by default. 24px divided by 16 is 1.5rem.

### What is the difference between rem and em?

rem is relative to the root font size, so it is predictable everywhere. em is relative to the parent element's font size, so nested ems compound.

### Should I use px or rem for font size?

Use rem, so text respects the user's browser settings. Keep px for things that should not scale, like thin borders.

### Is 1rem always 16px?

Only when the root font size is the default 16px. If the user or your CSS changes the root size, 1rem changes with it.

### Does using rem help accessibility?

Yes. People who set a larger default font size get a page that scales with their choice, instead of one that ignores it.



---

# Recurring deposit: ₹5,000 a month for 5 years, and how the maturity is worked out

> A recurring deposit (RD) lets you save a fixed amount every month at a fixed rate. At 7% for five years, ₹5,000 a month becomes about ₹3.60 lakh. This guide shows where each rupee of that interest comes from, why the first deposit earns 70 times more than the last, and what tax and tenure do to the final number.

Published 2026-10-08 by Toolcloak. 6 minute read. Original: https://toolcloak.pages.dev/articles/recurring-deposit-5000-a-month-for-5-years-how-the-maturity-is-worked-out
Related free tool: [RD Calculator](https://toolcloak.pages.dev/tools/rd-calculator)

Many people cannot park a large sum in a fixed deposit, but can set aside ₹2,000 or ₹5,000 every month. A recurring deposit is built for exactly that. You pick an amount, a tenure and get a fixed rate from the bank, and at the end you receive everything back with interest. The surprise is usually in the maths: the total is not “deposit × months × rate”, because each instalment earns for a different length of time.

> **In short**
> - ₹5,000 a month for **5 years at 7%** puts in **₹3,00,000** and returns about **₹3,59,664**, an interest of **₹59,664**.
> - Most banks compound RD interest **quarterly**. The first instalment earns for 60 months, the last for just one.
> - Interest is **taxable** at your slab rate. At 30% it trims the gain by almost a third.
> - An RD is for **safe, fixed savings**. For long goals, compare it with equity options and with [step-up SIP](https://toolcloak.pages.dev/articles/step-up-sip-raise-your-sip-10-percent-a-year).

## What a recurring deposit is

An RD is a bank (or post office) product where you deposit a fixed amount on a fixed day every month for a chosen period, usually from 6 months to 10 years. The rate is fixed at the start, so the maturity amount is known on day one. It is closer to a fixed deposit that you build slowly than to a mutual fund. There is no market risk, and also no chance of a market-sized return.

**RD or FD?**

| Recurring deposit | Fixed deposit |
| --- | --- |
| Small amounts, monthly | One lump sum, once |
| You save from your salary every month | You already have the full amount |
| Each instalment earns from its own date | All of it earns from day one |
| Good for building a habit | Better if you have a bonus or windfall |

*Both are safe and fixed. The right one depends on how your money arrives, monthly or in a lump.*

## How the maturity amount is calculated

Banks usually compound RD interest every quarter. The standard formula for the maturity value M of a monthly deposit R at an annual rate r, with i = r ÷ 400 for the quarterly rate and n quarters, is:

> **M = R × [(1 + i)ⁿ − 1] ÷ [1 − (1 + i)^(−1/3)]:** Here R is the monthly deposit, i the quarterly rate (7% a year gives 1.75% a quarter) and n the number of quarters (20 for five years). You do not need to do this by hand: the calculator below does it instantly. But the idea inside it is simple.

Think of each instalment as its own small deposit. The first ₹5,000 goes in on day one and earns for 60 months. The second earns for 59 months, and so on until the last, which earns for one month. Add up what each instalment grows into and you have the maturity amount.

**What really happens to the 60 instalments**

1. **Month 1.** ₹5,000 goes in and will earn interest for all 60 months.
2. **Month 30.** The 30th instalment goes in with 31 months to go.
3. **Month 60.** The last ₹5,000 earns for only one month.
4. **Maturity.** Every instalment, plus its own interest, is paid out together.

Try it: [RD Calculator](https://toolcloak.pages.dev/tools/rd-calculator). Enter a monthly deposit, a rate and a tenure to see the maturity amount and the interest earned.

## The ₹5,000 a month example, in numbers

**₹5,000 a month, 7%, 5 years**

|  | Value | Note |
| --- | --- | --- |
| You deposit | ₹3,00,000 | 60 instalments of ₹5,000 |
| Interest earned | ₹59,664 | About 19.9% on top |
| Maturity amount | ₹3,59,664 | Paid at the end of year 5 |

*Interest is about one rupee for every five you deposit. The rest of the money is simply your own savings returned.*

**Where the ₹3,59,664 comes from**

|  | Share |
| --- | --- |
| Your deposits | ₹3,00,000 |
| Interest | ₹59,664 |

*83% of what you receive is your own money. The interest is a bonus, not a salary.*

## Why the early instalments do the heavy lifting

An instalment made in month 1 earns for five full years. One made in month 60 earns for a single month. At 7%, the first ₹5,000 earns about **₹2,074** of interest, the 30th about **₹982**, and the last only **₹29**. That is roughly 70 times more for the first than for the last.

**Interest earned by selected instalments (₹5,000 each, 7%, 5 years)**

|  | Interest |
| --- | --- |
| 1st instalment | ₹2,074 |
| 15th | ₹1,524 |
| 30th | ₹982 |
| 45th | ₹485 |
| 60th | ₹29 |

*Each bar is one ₹5,000 instalment compounded quarterly for the months it stays in. The numbers add up across all 60 instalments to the ₹59,664 total.*

This also explains a feature of RD that surprises people: the interest earned in the last year is nearly **ten times** the first year, even though the instalments are all the same. By then you have a big balance earning interest, and the money keeps compounding.

**How the balance is built, year by year (₹ thousand)**

|  | Your deposits in the year | Interest earned in the year |
| --- | --- | --- |
| Year 1 | ₹60k | ₹2.3k |
| Year 2 | ₹60k | ₹6.8k |
| Year 3 | ₹60k | ₹11.6k |
| Year 4 | ₹60k | ₹16.7k |
| Year 5 | ₹60k | ₹22.2k |

*You deposit ₹60,000 every year, but the interest column keeps growing: ₹2.3k, 6.8k, 11.6k, 16.7k and 22.2k (rounded). Staying the full tenure matters.*

**RD balance against what you have deposited (value in lakh rupees)**

| Year | Maturity value if closed that year | Total you deposited |
| --- | --- | --- |
| 0 | 0 | 0 |
| 1 | 0.62 | 0.6 |
| 2 | 1.29 | 1.2 |
| 3 | 2.01 | 1.8 |
| 4 | 2.77 | 2.4 |
| 5 | 3.6 | 3 |

*The two lines start together and slowly pull apart. The gap, which is the interest, widens faster each year.*

## What changes the result

### Tenure matters more than rate

Because interest compounds on a growing balance, a longer tenure improves the percentage gain much more than a slightly higher rate does. Ten years at the same 7% earns about 45 paise for every rupee you deposit. Five years earns about 20 paise.

**Interest as a share of what you deposit, by tenure at 7% (percent)**

| Tenure in years | Interest ÷ deposits |
| --- | --- |
| 1 | 3.9 |
| 2 | 7.6 |
| 3 | 11.5 |
| 5 | 19.9 |
| 10 | 44.8 |

*Doubling the tenure from 5 to 10 years more than doubles the percentage gain, from 19.9% to 44.8%. The x-axis is not evenly spaced in years, only in the steps shown.*

**Maturity amount at 7% (₹)**

| Monthly deposit | 1 year | 3 years | 5 years |
| --- | --- | --- | --- |
| ₹1,000 | ₹12,462 | ₹40,137 | ₹71,933 |
| ₹2,000 | ₹24,924 | ₹80,275 | ₹1,43,866 |
| ₹5,000 | ₹62,311 | ₹2,00,686 | ₹3,59,664 |
| ₹10,000 | ₹1,24,621 | ₹4,01,373 | ₹7,19,328 |

### A one percent change in rate

For ₹5,000 a month over 5 years, each additional percentage point in the rate adds roughly ₹9,000 to ₹9,600 to the maturity. That is useful, but small next to the ₹3 lakh you put in.

**Maturity of ₹5,000 a month over 5 years at different rates**

|  | Maturity |
| --- | --- |
| 5% | ₹3,41,261 |
| 6% | ₹3,50,318 |
| 7% | ₹3,59,664 |
| 8% | ₹3,69,309 |

*Moving from 5% to 8% adds about ₹28,000 over five years. The bars look similar because the base, your own deposits, dominates.*

## Tax on RD interest

RD interest is added to your income and taxed at your slab rate, whether or not you take the money out. Banks may deduct TDS if the interest in a year crosses a threshold. That limit has changed in recent Budgets, so check your bank for the current figure. If your total income is below the taxable limit, you can submit a declaration form to avoid TDS.

**Our ₹3 lakh RD for someone in the 30% tax slab (before cess)**

| Step | Value |
| --- | --- |
| Maturity on the statement | ₹3.60 lakh |
| After 30% tax on the interest | ₹3.42 lakh |
| What you deposited | ₹3.00 lakh |

*Tax of about ₹17,900 takes the effective gain from ₹59,664 to ₹41,765. The same effect is explained for fixed deposits in [FD after tax and inflation](https://toolcloak.pages.dev/articles/fd-at-7-percent-after-tax-and-inflation-what-you-actually-earn).*

> **Closing early costs you:** If you break an RD before maturity, banks pay a lower rate (often the rate for the shorter period actually run, minus a small penalty). Missing instalments can add a late fee too. Pick an amount you can keep paying, not the largest one you can imagine.

## RD versus other ways to save monthly

A recurring deposit is a good fit when the money has a date and cannot be at risk: a school fee in three years, a wedding next year, a gadget, an emergency buffer. For goals over seven or ten years, savers also consider equity mutual funds through a SIP. Here is how ₹5,000 a month for five years looks under three very different assumptions.

**₹5,000 a month for 5 years under three assumptions**

|  | Final value |
| --- | --- |
| Savings account at 3% | ₹3,24,042 |
| RD at 7% | ₹3,59,664 |
| SIP at an assumed 12% | ₹4,12,432 |

*The 12% figure is an assumption, not a promise. Markets can return far less, or lose money, over five years. The RD and savings numbers are fixed by contract. See [what a SIP can tell you](https://toolcloak.pages.dev/articles/5000-rupees-a-month-for-20-years-what-a-sip-calculator-can-tell-you) for the longer view.*

None of these is “best”. Match the tool to the time you have. For money you need within three years, the certainty of an RD is worth its modest return. For goals decades away, a mix of safe and growth assets is more common. If you are saving for a specific target, the [monthly saving for ₹10 lakh](https://toolcloak.pages.dev/articles/saving-for-a-goal-how-much-per-month-to-reach-10-lakh) guide shows how rate and time trade off.

## A checklist before you open one

1. **Pick an amount you will not miss.** Set up an auto-debit for the day after your salary arrives.
2. **Choose the tenure to match the goal.** Do not lock the money for 10 years if you need it in 3.
3. **Compare rates.** Small finance banks and some co-operative banks may offer more, but check the deposit insurance limit before putting in large sums.
4. **Check the compounding.** Quarterly is common. A lower rate with more frequent compounding may still beat a higher rate with simple interest.
5. **Remember the tax.** Plan with the after-tax figure if you are in a higher slab.
6. **Review at maturity.** Renew, withdraw, or move the money to the next goal on purpose, rather than leaving it idle.

A recurring deposit rewards patience more than cleverness. Put in ₹5,000 for 60 months and the arithmetic does the rest. Use the calculator above to try your own amount, rate and tenure, and see the ₹3,59,664 become a number for your goal.

## Frequently asked questions

### How is the RD maturity amount calculated?

Banks compound the interest quarterly. Each monthly instalment earns interest from its deposit date to maturity, and the maturity value is the sum of all instalments grown at the quarterly rate.

### How much will ₹5,000 a month become in 5 years at 7%?

About ₹3,59,664. You deposit ₹3,00,000 and earn about ₹59,664 as interest.

### Is RD interest taxable?

Yes. RD interest is added to your income and taxed at your slab rate. Banks may deduct TDS above a yearly interest limit, which has changed in recent Budgets, so check the current figure.

### Which is better, RD or FD?

An FD suits a lump sum you already have, because all of it earns from day one. An RD suits monthly savings. Both give a fixed, bank-guaranteed return.

### What happens if I miss an RD instalment?

Most banks charge a small late fee per missed instalment, and several missed payments can lead to the RD being closed early at a reduced rate.

### Can I withdraw my RD before maturity?

Usually yes, but at a lower interest rate and sometimes with a penalty. Check your bank’s terms before opening the deposit.



---

# Flesch reading ease: what your readability score really means

> A score of 60 to 70 is “plain English”. A score of 30 is a legal contract. The formula behind it looks at only two things, sentence length and word length, which makes it easy to improve once you see how it works.

Published 2026-10-08 by Toolcloak. 3 minute read. Original: https://toolcloak.pages.dev/articles/flesch-reading-ease-what-your-readability-score-means
Related free tool: [Readability Score Checker](https://toolcloak.pages.dev/tools/readability-score-calculator)

Why does one article feel effortless and another feel like wading through mud, even when both are about the same thing? Much of it comes down to two numbers: how long your sentences are, and how long your words are. The Flesch reading ease score turns those two numbers into one.

> **In short**
> - Score = 206.835 − 1.015 × (words ÷ sentences) − 84.6 × (syllables ÷ words). **Higher is easier.**
> - 60 to 70 is plain English. Most web writing aims for 60 or above.
> - The score ignores meaning, tone and layout. It measures length, not clarity.
> - To raise it: **shorten sentences first**, then swap long words for short ones.

## The formula in plain words

The score was devised by Rudolf Flesch in 1948. It has two levers. The first is the average number of words in a sentence. The second is the average number of syllables in a word. More of either lowers the score.

> **Reading ease = 206.835 − 1.015 × (words ÷ sentences) − 84.6 × (syllables ÷ words):** A page of short sentences and short words scores near 100. A page of long sentences and long words can score near zero, or even below it.

## What the number means

**Flesch reading ease bands**

| Zone | From | To |
| --- | --- | --- |
| Very hard | 0 | 30 |
| Hard | 30 | 50 |
| Fairly hard | 50 | 60 |
| Plain | 60 | 70 |
| Fairly easy | 70 | 80 |
| Easy | 80 | 90 |
| Very easy | 90 | 100 |

*The marker is the example sentence below. Bands follow the standard published table.*

## Two sentences, scored by hand

**Worked examples**

|  | Value | Note |
| --- | --- | --- |
| “The cat sat on the mat.” | 116 | 6 words, 6 syllables, 1 sentence: very easy |
| “We moved the meeting to Friday because the team needs more time to finish the report.” | 79.6 | 16 words, 21 syllables: fairly easy |

*Check them: 206.835 − 1.015 × 16 − 84.6 × (21 ÷ 16) = 79.6. The score can go above 100 for very short sentences.*

Try it: [Readability Score Checker](https://toolcloak.pages.dev/tools/readability-score-calculator). Paste your text and get Flesch reading ease, the grade level and the counts behind them.

## Which lever should you pull?

Sentence length usually moves the score fastest, because splitting one long sentence in two cuts the average without changing a single word. Word length helps more slowly: “use” instead of “utilise” saves a syllable each time.

**Four kinds of writing**

| Box | What it means |
| --- | --- |
| Short and simple | Short sentences, short words. Easy to read, can feel choppy if everything is this size. |
| Short but heavy | Brief sentences full of long words, as in “Remuneration comprises discretionary components.” |
| Long but light | Long sentences of plain words. Readable, but the reader has to hold a lot in mind. |
| Long and heavy | Long sentences, long words. Policy documents and legal text live here. |

*Aim for a mix. Vary the length, but keep most sentences short.*

## What the score cannot tell you

- **Whether it is true or useful.** Nonsense in short words scores very well.
- **Whether it is well organised.** Headings, order and examples matter just as much.
- **Technical words that are easy for your audience.** “Algorithm” is long, but fine for programmers.
- **Languages other than English.** The constants are tuned for English syllables.

## Five quick edits that raise the score

1. **Cut every sentence over 25 words in two.**
2. **Swap a long word for a short one** when the meaning is the same.
3. **Use the active voice.** “We moved the meeting” beats “The meeting was moved by us.”
4. **Delete filler openings** like “It should be noted that”.
5. **Read it aloud.** If you run out of breath, the sentence is too long.

Word and character limits are a different constraint. For those, see [how long is long enough](https://toolcloak.pages.dev/articles/how-long-is-long-enough-a-field-guide-to-word-and-character-limits).

## Frequently asked questions

### What is a good Flesch reading ease score?

60 to 70 is usually called plain English, and most writing for the general public aims for 60 or above. Technical or legal writing is often 30 to 50.

### How is the Flesch reading ease score calculated?

206.835 minus 1.015 times the average words per sentence, minus 84.6 times the average syllables per word.

### Can the score be above 100 or below 0?

Yes. Very short sentences of one-syllable words can score above 100, and very long sentences with long words can score below 0.

### Does a high score mean good writing?

No. It only measures sentence and word length. A text can score high and still be confusing, inaccurate or badly organised.

### How do I improve my readability score?

Shorten your sentences first, then replace long words with shorter ones. Using the active voice and cutting filler phrases both help.



---

# ROI: why a 40% return can beat a 50% return, and what to count

> Return on investment is the quickest way to judge whether money was well spent, and the easiest to misuse. This guide shows the formula, why time and costs change the answer, and the difference between ROI and the ad-spend ratio marketers quote.

Published 2026-10-08 by Toolcloak. 3 minute read. Original: https://toolcloak.pages.dev/articles/roi-why-a-40-percent-return-can-beat-a-50-percent-return-and-what-to-count
Related free tool: [ROI Calculator](https://toolcloak.pages.dev/tools/roi-calculator)

Someone says they made a 50% return and someone else says 40%. Who did better? It depends on a question ROI does not ask by itself: **how long did it take?** ROI is a fast, flexible measure, and that is exactly why it is easy to bend. A few extra lines of working make it honest.

> **In short**
> - **ROI = (final value − amount invested) ÷ amount invested.** Put ₹50,000 in and get ₹75,000 out: ROI is **50%**.
> - ROI ignores time. Turn it into a yearly rate: 50% over 5 years is **8.4% a year**, while 40% over 2 years is **18.3% a year**.
> - Include **every cost**: fees, repairs, taxes. A 30% price gain on a property became **22%** after costs.
> - Marketing numbers differ: **ROAS** is revenue ÷ ad spend. A 3.5× ROAS can still be only a **13% ROI** once product cost is counted.

## The formula

> **ROI = (gain − cost) ÷ cost × 100%:** If you spent ₹50,000 and ended with ₹75,000, the net gain is ₹25,000. Divide by the ₹50,000 spent: 25,000 ÷ 50,000 = 0.5, which is 50%.

**A ₹50,000 investment that returns ₹75,000**

| Step | Amount |
| --- | --- |
| Invested | ₹50,000 |
| Gain | +₹25,000 |
| Value now | ₹75,000 |

*ROI is the gain as a share of the amount put in: ₹25,000 ÷ ₹50,000 = 50%.*

Try it: [ROI Calculator](https://toolcloak.pages.dev/tools/roi-calculator). Enter the amount invested, the final value and the time to see ROI and the yearly rate.

## Time changes everything

Compare two options. Project A turns ₹50,000 into ₹75,000 in 5 years. Project B turns ₹20,000 into ₹28,000 in 2 years. A has the bigger ROI, but it is also much slower.

**Total ROI against the yearly rate for the same two projects (%)**

|  | Rate |
| --- | --- |
| A: total ROI over 5 years | 50% |
| A: per year | 8.4% |
| B: total ROI over 2 years | 40% |
| B: per year | 18.3% |

*A wins on total ROI (50% against 40%), but B earns more than twice as much per year. The yearly rate is (1 + ROI)^(1 ÷ years) − 1.*

**Turning ROI into a yearly rate**

1. **Compute total ROI.** For B: (28,000 − 20,000) ÷ 20,000 = 0.40.
2. **Add 1 and take the root.** 1.40 raised to 1/2 (the square root) is 1.1832.
3. **Subtract 1.** 0.1832, or 18.3% a year. This is the CAGR.

This yearly figure is the same one used in [CAGR](https://toolcloak.pages.dev/articles/average-return-vs-cagr-why-20-percent-is-not-what-you-earned). Money that works faster can be reinvested sooner, which is why speed matters as much as size.

## Count every cost

A flat bought for ₹40 lakh and sold for ₹52 lakh four years later looks like a 30% gain. But buying and selling costs, repairs and brokerage came to ₹3.2 lakh. The honest gain is ₹52 − ₹40 − ₹3.2 = ₹8.8 lakh.

**Property example (₹ lakh)**

| Step | Amount |
| --- | --- |
| Sale price | 52 |
| Purchase price | −40 |
| Fees and repairs | −3.2 |
| Net gain | 8.8 |

*The headline gain of ₹12 lakh shrinks to ₹8.8 lakh after costs.*

**Same sale, three ways to measure it**

|  | Rate |
| --- | --- |
| Headline ROI (no costs) | 30% |
| Net ROI (after costs) | 22% |
| Net, per year over 4 years | 5.1% |

*A 30% headline becomes 22% after costs and only about 5.1% a year, which is below 6% inflation.*

**From headline to buying power**

|  | Value | Note |
| --- | --- | --- |
| Headline gain | ₹12 lakh | Sale price minus purchase price |
| After costs | ₹8.8 lakh | Fees and repairs of ₹3.2 lakh taken off |
| In today’s money | ≈ −₹2 lakh | ₹52 lakh ÷ 1.06⁴ = ₹41.2 lakh, less ₹40 lakh and ₹3.2 lakh costs |

*With 6% yearly inflation the sale price buys what ₹41.2 lakh buys today. That is barely above the purchase price, so after costs the real result is a small loss. See the inflation guide for how this adds up.*

> **Ask what is missing:** Has the return included fees, tax, repairs, the cost of your own time, and the interest you could have earned elsewhere? The last one is the opportunity cost. A project that earns 5% when a safe option pays 7% has a positive ROI and still a bad result.

## ROI is not ROAS

In marketing, ROAS (return on ad spend) is revenue divided by what you spent on ads. It ignores the cost of making the product, so a high ROAS can hide a thin profit. ROI counts profit.

**One ₹1 lakh ad campaign (₹ lakh)**

| Step | Amount |
| --- | --- |
| Revenue | 3.5 |
| Cost of goods (60%) | −2.1 |
| Ad spend | −1.0 |
| Profit | 0.4 |

*ROAS = 3.5 ÷ 1 = 3.5×. ROI = profit ÷ cost = 0.4 ÷ (2.1 + 1.0) ≈ 13%, or 40% if you count only the ad spend as the investment. Always say which cost base you use.*

**Four return measures that sound alike**

| Box | What it means |
| --- | --- |
| ROI | Net profit ÷ total cost. Works for any project, any time length. |
| ROAS | Revenue ÷ ad spend. Ignores product cost, so it can look better than ROI. |
| CAGR | ROI turned into a yearly rate. Use it to compare different time lengths. |
| Margin | Profit ÷ revenue. A different base from ROI, which divides by cost. |

*Always name the measure and the cost base when you quote a percentage.*

## Quick practice

1. Buy for ₹80,000, sell for ₹92,000. ROI? 12,000 ÷ 80,000 = **15%**.
2. It took 3 years. Yearly rate? 1.15^(1/3) − 1 = **4.8%**.
3. Spend ₹10,000 on ads, get ₹40,000 revenue. ROAS? **4×**.
4. Which is better: 20% in 1 year or 50% in 3 years? 50% over 3 years is **14.5% a year**, so the 1-year 20% wins.

ROI is a start, not a verdict. Add time, count all the costs and compare against what the money could have earned elsewhere.

## Frequently asked questions

### What is the ROI formula?

ROI = (final value − amount invested) ÷ amount invested, shown as a percentage. Include all costs in the amount invested.

### What is a good ROI?

It depends on time and risk. Compare the yearly rate with safer alternatives such as fixed deposits and with inflation.

### How do you annualise ROI?

Use (1 + ROI)^(1 ÷ years) − 1. For a 50% ROI over 5 years that is 1.5^0.2 − 1 = 8.4% a year.

### What is the difference between ROI and ROAS?

ROAS divides revenue by ad spend. ROI divides profit by total cost. ROAS can be high while ROI is small.

### Can ROI be negative?

Yes. If you end with less than you put in, the ROI is negative. Losing half your money is an ROI of −50%.

### Why is ROI misleading?

It ignores how long the money was tied up, risk, and often hidden costs. Use the yearly rate and net figures for fair comparisons.



---

# Roman numerals: how MCMXCIV adds up, and why 49 is XLIX, not IL

> Roman numerals are seven letters and two simple rules, yet people get 49, 99 and 1994 wrong all the time. This guide shows the greedy method that never fails, the numbers with the longest and shortest forms, and why the system tops out at 3,999.

Published 2026-10-08 by Toolcloak. 4 minute read. Original: https://toolcloak.pages.dev/articles/roman-numerals-how-mcmxciv-adds-up-and-why-49-is-xlix-not-il
Related free tool: [Roman Numeral Converter](https://toolcloak.pages.dev/tools/roman-numeral-converter)

You see them on clock faces, film credits, book chapters, Super Bowl logos and the cornerstone of old buildings. Roman numerals have no zero, no place value and no tidy columns, yet a handful of letters can write any number up to 3,999. The system is easy once you know the one method that always works.

> **In short**
> - Seven letters: **I = 1, V = 5, X = 10, L = 50, C = 100, D = 500, M = 1000**.
> - Add letters from left to right, except a **smaller letter before a bigger one is subtracted**: IV = 4, IX = 9, XL = 40, XC = 90, CD = 400, CM = 900.
> - Use the **greedy method**: take the biggest value that fits, write it, subtract, repeat. 1994 becomes **MCMXCIV**.
> - No letter repeats more than three times in a row, so the standard system ends at **3,999 (MMMCMXCIX)**.

## The seven letters

**What each letter is worth**

|  | Value |
| --- | --- |
| I | 1 |
| V | 5 |
| X | 10 |
| L | 50 |
| C | 100 |
| D | 500 |
| M | 1,000 |

*Values rise in a pattern of 1, 5, 10, 50, 100, 500, 1000: alternately ×5 and ×2.*

Notice the rhythm: 1 → 5 is times five, 5 → 10 is times two, 10 → 50 is times five again. Early Romans probably used V as half of X, because V looks like the upper half of an X.

Try it: [Roman Numeral Converter](https://toolcloak.pages.dev/tools/roman-numeral-converter). Enter 2026 to see MMXXVI, or paste a Roman numeral to read it as a number.

## Add, but sometimes subtract

Read left to right and add the values: VIII is 5 + 1 + 1 + 1 = 8, and MMXXVI is 1000 + 1000 + 10 + 10 + 5 + 1 = 2026. The one twist is the subtractive pair. When a small letter sits **in front of** a bigger one, you subtract it, which saves writing four of the same letter in a row (IIII).

**The six subtractive pairs**

| Box | What it means |
| --- | --- |
| IV = 4 and IX = 9 | I before V or X. Only I can go in front of these two. |
| XL = 40 and XC = 90 | X before L or C. X is the only small letter allowed here. |
| CD = 400 and CM = 900 | C before D or M. C is the only letter that can go in front. |
| Never VX, LC or DM | V, L and D are never subtracted. 95 is XCV, not VC. |

*Each subtractive pair uses a power of ten (I, X, C) in front of one of the next two bigger letters.*

**Writing 49**

| IL | XLIX |
| --- | --- |
| Wrong | Right |
| I may only come before V or X | XL = 40, then IX = 9 |
| It tries to skip from 1 straight to 50 | Handle each digit separately |
| Many people write it this way | 40 + 9 = 49 |

*Treat tens and ones as separate blocks. 49 is “forty” then “nine”, so XL then IX.*

## The greedy method

The easiest way to convert any number is to work from the largest value to the smallest. Keep this list of thirteen values in order and you never need to think about the subtraction rule again, because it already includes the six pairs.

| Value | 1000 | 900 | 500 | 400 | 100 | 90 | 50 | 40 | 10 | 9 | 5 | 4 | 1 |
| --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- |
| Letters | M | CM | D | CD | C | XC | L | XL | X | IX | V | IV | I |

*Always take the largest value that still fits. Repeat until nothing is left.*

**Converting 1994 step by step**

| Step | Amount |
| --- | --- |
| 1994 | 1,994 |
| Take M | −1000 |
| Take CM | −900 |
| Take XC | −90 |
| Take IV | −4 |

*The pieces M + CM + XC + IV give MCMXCIV. After the last step nothing is left, so the number is done.*

**The conversion in three moves**

1. **Find the largest value that fits.** For 2026 that is 1000 (M). Write M and subtract: 1026 remains.
2. **Repeat with what is left.** M again (26 left), then X, X (6 left), then V (1 left), then I.
3. **Stop at zero.** MMXXVI. The letters were written in the same order they were found.

> **How to read one:** To read one, scan left to right. If a letter is smaller than the one after it, subtract it; otherwise add. For MCMXCIV: M (add 1000), C before M (subtract 100), M (add 1000), X before C (subtract 10), C (add 100), I before V (subtract 1), V (add 5). Total 1000 − 100 + 1000 − 10 + 100 − 1 + 5 = 1994.

## Short numbers and long ones

Roman numerals are not neatly proportional to size. Some small numbers are longer than much bigger ones, because 8 needs four letters while 10 needs one.

**Number of letters needed for 1 to 30**

| The number | Letters in the numeral |
| --- | --- |
| 1 | 1 |
| 2 | 2 |
| 3 | 3 |
| 4 | 2 |
| 5 | 1 |
| 6 | 2 |
| 7 | 3 |
| 8 | 4 |
| 9 | 2 |
| 10 | 1 |
| 11 | 2 |
| 12 | 3 |
| 13 | 4 |
| 14 | 3 |
| 15 | 2 |
| 16 | 3 |
| 17 | 4 |
| 18 | 5 |
| 19 | 3 |
| 20 | 2 |
| 21 | 3 |
| 22 | 4 |
| 23 | 5 |
| 24 | 4 |
| 25 | 3 |
| 26 | 4 |
| 27 | 5 |
| 28 | 6 |
| 29 | 4 |
| 30 | 3 |

*The pattern repeats every ten: 3 (III), 8 (VIII), 13 (XIII), 18 (XVIII) and 23 (XXIII) are all longer than the numbers just after them (IV, IX, XIV, XIX, XXIV).*

**Records between 1 and 3,999**

|  | Value | Note |
| --- | --- | --- |
| 3,888: the longest | MMMDCCCLXXXVIII | 15 letters |
| 3,999: the largest | MMMCMXCIX | 9 letters. It is the largest standard value |
| Average length | 7.5 | Letters across all 3,999 numbers |

*The longest numeral, 3,888, is MMMDCCCLXXXVIII. Every power of ten there uses its longest pattern.*

**Average numeral length in each hundred (letters)**

|  | Value |
| --- | --- |
| 1+ | 4 |
| 100+ | 5 |
| 200+ | 6 |
| 300+ | 7 |
| 400+ | 6 |
| 500+ | 5 |

*The 300s are the longest on average (about 7 letters, with CCC plus a long tail). The 400s get shorter again because CD replaces four Cs.*

## Why it stops at 3,999, and the missing zero

Roman numerals have no symbol for zero and no place value, so the same letter always means the same amount wherever it sits. The standard rule of no more than three of the same letter in a row means M can only appear three times, so 3,999 is the end. Larger numbers were written with a bar over a letter to multiply it by 1,000, but that is not used in everyday life.

> **Clocks cheat:** Many clock faces show 4 as IIII instead of IV. It balances the look of the dial against VIII on the other side. Old inscriptions also sometimes wrote 9 as VIIII. Treat the modern rule as the standard, and expect surprises on old buildings.

## Quick practice

1. What is 2026? **MMXXVI** (MM + XX + VI).
2. What is 444? **CDXLIV** (CD + XL + IV, three subtractive pairs).
3. What does XCIX mean? **99** (XC + IX). It is not IC.
4. What is 3,999? **MMMCMXCIX**, the largest standard numeral.

Once you stop trying to see “place value” and just match values from the list, Roman numerals become mechanical. For another number system that works in blocks, see [binary, hex and decimal](https://toolcloak.pages.dev/articles/binary-hex-and-decimal-counting-in-bases-2-10-and-16), where every position is a power of the base.

## Frequently asked questions

### What is 49 in Roman numerals?

XLIX, which is XL (40) followed by IX (9). IL is not allowed because I may only come before V or X.

### What is 2026 in Roman numerals?

MMXXVI: M + M + X + X + V + I = 1000 + 1000 + 10 + 10 + 5 + 1.

### What is the largest Roman numeral?

In the standard system it is MMMCMXCIX, which is 3,999. Larger numbers were written with a bar over a letter to multiply it by 1,000.

### Is there a zero in Roman numerals?

No. The system has no zero and no place value. Medieval writers sometimes used the Latin word nulla for none.

### Why do clocks show IIII instead of IV?

It is a design choice: IIII balances the dial visually against VIII and avoids repeating the IV-IX pattern. Both forms have been used for centuries.

### How do you write 99 in Roman numerals?

XCIX: XC (90) plus IX (9). IC is not valid because I can only be placed before V or X.



---

# The rule of 72: how long your money takes to double

> Divide 72 by the yearly return and you get the years to double. It takes ten seconds, works in your head, and quietly explains both why investing rewards patience and why inflation is such a thief.

Published 2026-10-08 by Toolcloak. 4 minute read. Original: https://toolcloak.pages.dev/articles/the-rule-of-72-how-long-money-takes-to-double
Related free tool: [Rule of 72 Calculator](https://toolcloak.pages.dev/tools/rule-of-72-calculator)

A friend says a fixed deposit pays 7 percent. Another says a fund gave 12. You do not need a calculator to know how different those are. Divide 72 by each number: about **10 years** to double the first, **6 years** to double the second. That is the whole rule.

> **In short**
> - Years to double ≈ 72 ÷ the yearly return. 6% takes about 12 years, 12% about 6.
> - It is accurate to within weeks between 6 and 12 percent, and looser outside that range.
> - Point it at inflation: at 6% your cash loses half its buying power in about 12 years.

It is called the rule of 72, and bankers have used it for centuries because it turns a scary compound-interest formula into a sum you can do while someone is still talking. Below is why it works, how far you can trust it, and the one place it is most useful for ordinary people: seeing what inflation does to cash.

## The rule in one line

> **Years to double ≈ 72 ÷ yearly return (in percent):** At 6 percent: 72 ÷ 6 = 12 years. At 9 percent: 72 ÷ 9 = 8 years. At 12 percent: 72 ÷ 12 = 6 years. It works the same way for a loan that grows against you, or a price that rises.

Notice it does not matter how much you start with. ₹10,000 or ₹10 crore, the time to double is the same. Only the **rate** and the **time** matter, which is exactly why compounding feels so unfair to people who start late.

## Why 72, and not some tidier number

The exact answer comes from logarithms: the years to double are `ln(2) ÷ ln(1 + r)`. The number `ln(2)` is about 0.693, which would suggest a “rule of 69.3”. People use 72 anyway for a practical reason: it divides cleanly by 2, 3, 4, 6, 8, 9 and 12, so the mental maths stays easy at the rates people actually meet.

It also happens to land closer to the true answer at the rates most of us deal with, roughly 6 to 10 percent. At very low or very high rates it drifts a little, which is why some people prefer 70 for low rates and 75 for high ones.

| Yearly return | Rule of 72 | Exact answer | Off by |
| --- | --- | --- | --- |
| 3% | 24 years | 23.4 years | 0.6 years |
| 6% | 12 years | 11.9 years | 0.1 years |
| 8% | 9 years | 9.0 years | 0.0 years |
| 12% | 6 years | 6.1 years | 0.1 years |
| 18% | 4 years | 4.2 years | 0.2 years |
| 24% | 3 years | 3.2 years | 0.2 years |

*Exact values are ln(2) ÷ ln(1 + rate), rounded to one decimal. For anything between 6 and 12 percent the shortcut is within a few weeks.*

**Years to double at different yearly returns**

|  | Value |
| --- | --- |
| 4% | 18 |
| 6% | 12 |
| 8% | 9 |
| 12% | 6 |
| 18% | 4 |

*Doubling the rate does not just halve the wait in a vague way. It is exactly inverse: twice the return, half the years.*

Try it: [Rule of 72 Calculator](https://toolcloak.pages.dev/tools/rule-of-72-calculator). Type a rate to see the years to double, or ask for triple and ten times. It shows the exact answer next to the shortcut.

## What doubling looks like over thirty years

Doubling once sounds modest. The trick is that it repeats. Put ₹1 lakh away at 12 percent and it doubles about every six years, so in thirty years it has doubled five times: 1, 2, 4, 8, 16, 32. You end up near ₹30 lakh from a single lakh, and you did nothing in between.

**What ₹1 lakh becomes (value in lakh rupees)**

| Year | At 12% a year | At 8% a year | At 6% a year |
| --- | --- | --- | --- |
| 0 | 1 | 1 | 1 |
| 5 | 1.76 | 1.47 | 1.34 |
| 10 | 3.11 | 2.16 | 1.79 |
| 15 | 5.47 | 3.17 | 2.4 |
| 20 | 9.65 | 4.66 | 3.21 |
| 25 | 17 | 6.85 | 4.29 |
| 30 | 29.96 | 10.06 | 5.74 |

*Move your finger along the curves. At year 15 the 12 percent line is at ₹5.5 lakh. Between year 25 and year 30 it gains more than it did in the first twenty years put together.*

Four percentage points sounds like a small gap between 8 and 12. After thirty years it is the gap between ₹10 lakh and ₹30 lakh. That is not a 50 percent difference. It is three times as much.

## The same rule, pointed at inflation

Here is the version that changes how people feel about idle cash. Prices rise, so each rupee buys less. If prices go up 6 percent a year, then 72 ÷ 6 = **12 years** for your money to lose half its buying power. The cash did not go anywhere. The things it buys just got more expensive.

**The same ₹1 lakh in a drawer, in 12 years at 6 percent inflation**

| What the note says | What it buys |
| --- | --- |
| ₹1,00,000 | about ₹50,000 |
| Same number on the paper | Half the groceries, half the fuel |
| Still “safe”, nothing was lost | Nothing visibly happened, which is the trap |

*Use the rule backwards to check what your savings need to earn just to stand still.*

> **A useful sanity check:** If a deposit pays 6 percent and prices rise 6 percent, your money is **not growing**. It is treading water. The return that matters is the rate **after** inflation, called the real rate.

## Where it breaks down

- **Taxes and fees.** The rule uses the rate you actually keep. A fund that returns 12 percent but costs 1.5 percent in fees is a 10.5 percent investment, which doubles in about 7 years, not 6.
- **Returns that are not steady.** Shares do not go up 12 percent every year. The rule describes a smooth average, not the bumps. See the CAGR article for why the two differ.
- **Very high or very low rates.** Below 2 percent or above 30 percent the shortcut gets noticeably loose. Use the exact formula there.
- **Regular deposits.** The rule is for one lump sum left alone. A monthly SIP is a different shape, because each deposit starts its own clock.

## Three quick uses

1. **Compare two offers.** 8 percent versus 10 percent sounds close. 9 years versus 7.2 years to double shows it is not.
2. **Check a promise.** “Double your money in 3 years” means about 24 percent a year. Be suspicious of anything that needs that.
3. **Size a goal.** Need your savings to double in 9 years? You need roughly 8 percent a year. Now you know what rate to look for.

## Frequently asked questions

### What is the rule of 72?

A shortcut for how long money takes to double. Divide 72 by the yearly return in percent. At 9 percent, 72 ÷ 9 gives about 8 years.

### How accurate is the rule of 72?

Very accurate between about 6 and 12 percent, usually within a few weeks. At very low or very high rates it drifts, and the exact formula, ln(2) ÷ ln(1 + rate), is better.

### Why 72 and not 69 or 70?

The exact constant is about 69.3, but 72 divides evenly by many common rates (2, 3, 4, 6, 8, 9, 12), which makes mental maths easier, and it fits the 6 to 10 percent range well.

### Can I use the rule of 72 for inflation?

Yes. Divide 72 by the inflation rate to find how many years it takes for money to lose half its buying power. At 6 percent inflation that is about 12 years.

### Does the rule work for tripling?

Not directly. The rule of 114 is the usual shortcut for tripling, and 72 is only for doubling. A calculator will give you any multiple exactly.



---

# Running pace explained: from min/km to 5K, 10K and marathon times

> Pace is the time you take for one kilometre. Hold 6:00 per km and you finish a 5K in 30:00, a 10K in 1:00:00, a half marathon in 2:06:35 and a marathon in 4:13:10. This guide shows the maths, a ready table for common paces and how to pick a pace you can actually keep.

Published 2026-10-08 by Toolcloak. 3 minute read. Original: https://toolcloak.pages.dev/articles/running-pace-explained-from-min-per-km-to-5k-10k-and-marathon-times
Related free tool: [Running Pace Calculator](https://toolcloak.pages.dev/tools/running-pace-calculator)

Your watch shows pace, your friend talks about speed, and the race website lists finish times. They are the same thing in different clothes. Once you can switch between them, you can pick a target for any race and know if it is realistic.

> **In short**
> - **Finish time = pace per km × distance in km.** At 6:00 per km, a 10K takes exactly 60 minutes.
> - **Speed in km/h = 60 ÷ pace in minutes.** 6:00 per km is **10 km/h**.
> - Marathon is **42.195 km** and a half marathon **21.0975 km**, so a small pace change adds minutes: **5:30 vs 6:00 per km is 21 minutes in a marathon**.
> - Start at a pace you can hold. Going out too fast is the most common racing mistake.

## Pace and speed are two views of one number

Pace tells you minutes per kilometre. Speed tells you kilometres per hour. They are reciprocals, so a faster pace means a smaller number of minutes and a bigger speed.

**Pace to speed**

|  | km/h |
| --- | --- |
| 7:00 per km | 8.6 km/h |
| 6:30 per km | 9.2 km/h |
| 6:00 per km | 10 km/h |
| 5:30 per km | 10.9 km/h |
| 5:00 per km | 12 km/h |

*To convert, divide 60 by the pace in minutes. 5:30 is 5.5 minutes, 60 ÷ 5.5 = 10.9 km/h.*

> **Finish time = pace × distance:** 6:00 per km over 10 km is 6 × 10 = 60 minutes. For a pace like 5:30, use 5.5 minutes: 5.5 × 10 = 55 minutes.

Try it: [Running Pace Calculator](https://toolcloak.pages.dev/tools/running-pace-calculator). Enter a distance and a time, or a pace, to get the other values.

## Finish times at common paces

**Finish time by pace per km**

| Pace | 5K | 10K | Half | Marathon |
| --- | --- | --- | --- | --- |
| 5:00 | 25:00 | 50:00 | 1:45:29 | 3:30:59 |
| 5:30 | 27:30 | 55:00 | 1:56:02 | 3:52:04 |
| 6:00 | 30:00 | 1:00:00 | 2:06:35 | 4:13:10 |
| 6:30 | 32:30 | 1:05:00 | 2:17:08 | 4:34:16 |
| 7:00 | 35:00 | 1:10:00 | 2:27:41 | 4:55:22 |

**Marathon time at each pace (hours)**

|  | Value |
| --- | --- |
| 5:00 | 3:31 |
| 5:30 | 3:52 |
| 6:00 | 4:13 |
| 6:30 | 4:34 |
| 7:00 | 4:55 |

*Each 30 seconds per km adds about 21 minutes over 42.195 km. Small pace gains pay off over long distances.*

## Choosing a pace you can hold

- **Use a recent race or hard time trial** as the starting point. A recent 5K tells you more than a guess.
- **Add time for longer races.** Your marathon pace is slower than your 10K pace, often by 30 to 60 seconds per km, depending on your training.
- **Allow for conditions.** Heat, hills and wind slow everyone down. Run by effort on those days.
- **Use the talk test for easy runs.** If you cannot speak in sentences, you are going faster than easy pace. See [heart rate zones](https://toolcloak.pages.dev/articles/heart-rate-zones-explained-what-each-of-the-five-zones-is-for).

## Pacing a race

**Two ways to run the same 10K**

| Fast start | Even or negative split |
| --- | --- |
| Slows in the second half | Steady or faster finish |
| First km far too quick | Start slightly easy |
| Heart rate climbs early | Hold the target pace in the middle |
| Last 3 km feel like a struggle | Pick up in the last km if you can |

*A negative split means the second half is faster than the first. It is easier to achieve if you start a little slower than you feel you could.*

## Splitting the distance

Break the race into pieces. For a 10K at 6:00 per km, check your watch at 2.5 km (15:00), 5 km (30:00) and 7.5 km (45:00). Small checks keep you on target without constantly watching the screen.

**A 10K at 6:00 per km**

| When | What |
| --- | --- |
| 2.5 km | 15:00. Settle in. If you are ahead of this, ease back. |
| 5 km | 30:00. Halfway. Hold the same effort. |
| 7.5 km | 45:00. The hard part starts. Focus on form and breathing. |
| 10 km | 1:00:00. Finish. If you have energy, use it in the last kilometre. |

## Training sensibly

1. Increase weekly distance gradually, not by large jumps.
2. Keep most runs easy and add one harder session a week as you build up.
3. Rest days are part of training. Our [sleep guide](https://toolcloak.pages.dev/articles/sleep-in-90-minute-cycles-when-to-go-to-bed) helps with recovery.
4. Stop and seek advice if pain is sharp, getting worse or changes the way you run.

> **Use pace as a guide, not a judge:** Some days are slower. A pace chart shows what is possible on a good day. On other days, effort is a better guide.

## Frequently asked questions

### How do I calculate running pace?

Pace per km = total time ÷ distance in km. A 50-minute 10K is 50 ÷ 10 = 5:00 per km.

### What is 6:00 per km in km/h?

10 km/h. Divide 60 by the pace in minutes.

### What is a 5K time at 6:00 per km?

30:00. A 10K takes 1:00:00, a half marathon 2:06:35 and a marathon 4:13:10.

### How long is a marathon?

42.195 km (26.2 miles). A half marathon is 21.0975 km.

### What is a negative split?

Running the second half of a race faster than the first half.



---

# Saving for a goal: how much per month to reach ₹10 lakh

> To have ₹10 lakh in 5 years you need about ₹16,667 a month if you earn nothing on it, ₹13,887 at 7% and ₹12,807 at 10%. The longer you give yourself, the more your money does the work. This guide shows the formula, the trade-off between time and monthly amount, and a simple plan.

Published 2026-10-08 by Toolcloak. 3 minute read. Original: https://toolcloak.pages.dev/articles/saving-for-a-goal-how-much-per-month-to-reach-10-lakh
Related free tool: [Savings Goal Calculator](https://toolcloak.pages.dev/tools/savings-goal-calculator)

A goal without a number is a wish. “I want to buy a car” becomes a plan only when it says how much, by when, and how much to put away each month. The answer depends on one more thing: how much your savings can earn while they wait.

> **In short**
> - **Monthly saving = goal ÷ growth factor.** With no return, it is just goal ÷ months. With returns, you need less.
> - For **₹10 lakh in 5 years**: **₹16,667** with 0% return, **₹13,887** at 7%, **₹12,807** at 10%.
> - Giving yourself **10 years** instead of 5 cuts the monthly amount at 7% from ₹13,887 to **₹5,744**.
> - Short goals (under 3 years) should use safe places. Higher returns mean higher risk.

## The formula

> **Monthly saving = goal × i ÷ [((1 + i)ⁿ − 1) × (1 + i)]:** i is the monthly rate (7% a year is 0.5833% a month) and n is the number of months, assuming you deposit at the start of each month. At a return of zero, divide the goal by the number of months.

You do not have to do this by hand. The point is the shape: each extra year and each extra point of return reduce the monthly figure.

**Monthly saving needed for ₹10 lakh in 5 years**

|  | Per month |
| --- | --- |
| Earn 0% | ₹16667 |
| Earn 7% | ₹13887 |
| Earn 10% | ₹12807 |

*Returns trim the monthly amount by ₹2,800 to ₹3,900 here. Over just 5 years, the amount you save matters more than the rate.*

**Where the ₹10 lakh comes from at 7% over 5 years**

|  | Your deposits | Growth |
| --- | --- | --- |
| ₹13,887 × 60 months | ₹8.33 L yours | ₹1.67 L |

*About 83% is your own money. On a short goal, you are mostly saving, not investing.*

Try it: [Savings Goal Calculator](https://toolcloak.pages.dev/tools/savings-goal-calculator). Enter a goal, the years and a return to get the monthly amount.

## Time is the cheapest lever

**Monthly saving for ₹10 lakh (₹)**

| Time | 0% | 7% | 10% |
| --- | --- | --- | --- |
| 3 years | ₹27,778 | ₹24,899 | ₹23,736 |
| 5 years | ₹16,667 | ₹13,887 | ₹12,807 |
| 7 years | ₹11,905 | ₹9,206 | ₹8,200 |
| 10 years | ₹8,333 | ₹5,744 | ₹4,841 |

A 10-year goal at 7% needs only 41% of the monthly amount of a 5-year goal. If the date is flexible, moving it is the easiest way to make the plan fit your budget.

## Match the place to the time

**How long until you need the money**

| When | What |
| --- | --- |
| Under 1 year | Savings account or liquid options. Stability matters more than return. |
| 1 to 3 years | Fixed or recurring deposits, short-term debt options. Predictable and low risk. Compare after-tax return with [FD after tax and inflation](/articles/fd-at-7-percent-after-tax-and-inflation-what-you-actually-earn). |
| 3 to 5 years | A mix of debt and some equity. Some growth, but you must be able to cope with a dip near the date. |
| 5 years or more | More equity can make sense. Time smooths out the ups and downs. Move to safer places as the date nears. |

## A plan that works

1. Write down the goal in today’s rupees and the date.
2. Add inflation if it is a big purchase. A ₹10 lakh car today may cost more in 5 years.
3. Find the monthly amount using a realistic return, such as the safe rate for short goals.
4. Set up an automatic transfer on salary day, to a separate account.
5. Raise the amount every year when your pay rises. A [step-up plan](https://toolcloak.pages.dev/articles/step-up-sip-raise-your-sip-10-percent-a-year) lowers the starting figure.

> **Do not forget the safety net:** Keep an emergency fund before you commit to goals. Otherwise a surprise expense forces you to break the goal. See [how big an emergency fund should be](https://toolcloak.pages.dev/articles/how-big-should-an-emergency-fund-be).

## What if you are behind?

- **Extend the date.** Even 6 months can change the monthly amount a lot.
- **Lower the target.** A cheaper car or a smaller wedding budget is a real option.
- **Add a lumpsum.** A bonus or a refund can fill part of the gap. See [what a lumpsum becomes](https://toolcloak.pages.dev/articles/lumpsum-investment-what-1-lakh-becomes-in-10-20-and-30-years).
- **Do not chase higher returns.** Taking big risk on a short deadline is how goals are lost.

## Frequently asked questions

### How much should I save per month for a goal?

Divide the goal by the number of months if it earns nothing. If it earns a return, use the future value of a monthly deposit formula, or a calculator.

### How much per month to reach ₹10 lakh in 5 years?

About ₹16,667 with no return, ₹13,887 at 7% a year and ₹12,807 at 10% a year, assuming deposits at the start of each month.

### Where should I keep money for a goal under 3 years?

In low-risk places such as savings accounts, fixed or recurring deposits and short-term debt options, because you cannot wait out a market fall.

### Does a higher return help a lot on short goals?

Less than you would think. Over 5 years, your own deposits make up most of the final amount.

### Should I include inflation?

Yes for big purchases. Estimate the future price and use that as your goal.



---

# Sleep in 90-minute cycles: when to go to bed so the alarm does not feel like an attack

> You do not wake up groggy only because you slept too little. Often it is because the alarm cut into the middle of a deep cycle. Counting back in blocks of about 90 minutes is a simple way to plan your night.

Published 2026-10-08 by Toolcloak. 3 minute read. Original: https://toolcloak.pages.dev/articles/sleep-in-90-minute-cycles-when-to-go-to-bed
Related free tool: [Sleep Cycle Calculator](https://toolcloak.pages.dev/tools/sleep-calculator)

Two people sleep seven and a half hours. One wakes up fine. The other feels like they were pulled out of a well. The difference is often not the hours but where in the night the alarm lands.

> **In short**
> - A night is a chain of sleep cycles of roughly **90 minutes** each (it varies from person to person and through the night).
> - Most adults need **7 or more hours**, which is about 5 cycles. Waking at the end of a cycle usually feels easier.
> - Add about **15 minutes** to fall asleep when you count back from your wake-up time.
> - These are rules of thumb. If you are tired every day despite long nights, speak to a doctor.

## What a sleep cycle is

Sleep is not a flat line. You move from light sleep into deep sleep, then into REM sleep, where most vivid dreaming happens, and then the pattern starts again. One trip through is a cycle. It lasts around 90 minutes on average, and the mix changes through the night: more deep sleep in the first half, more REM in the second.

**A typical night for a healthy adult (approximate)**

|  | Share |
| --- | --- |
| Light sleep | about 50% |
| Deep sleep | about 20% |
| REM sleep | about 25% |
| Brief wake-ups | about 5% |

*Shares differ with age and between people. These are round figures to show the idea, not a target to hit.*

## Counting back from your alarm

Pick the time you must wake up, then count backwards in 90-minute blocks, and add roughly 15 minutes for falling asleep. For a 6:30 am alarm:

**Bedtimes for a 6:30 am wake-up**

| When | What |
| --- | --- |
| 9:15 pm | 6 cycles, 9 hours. More than most adults need. Good after a short week. |
| 10:45 pm | 5 cycles, 7.5 hours. The sweet spot for many adults. |
| 12:15 am | 4 cycles, 6 hours. Enough once in a while, not as a habit. |
| 1:45 am | 3 cycles, 4.5 hours. An emergency night. Expect to pay for it in the afternoon. |

*Each time already includes about 15 minutes to fall asleep.*

Try it: [Sleep Cycle Calculator](https://toolcloak.pages.dev/tools/sleep-calculator). Enter your wake-up time or your bedtime and it lists the times that end a cycle.

## How many hours do you need?

**Hours of sleep for an adult**

| Zone | Range |
| --- | --- |
| Too little | under 6 |
| Borderline | 6 to 7 |
| Recommended | 7 to 9 |
| Check why | over 9 |

*Sleep and health bodies advise 7 or more hours for adults. Teenagers and children need more.*

## Why 90 minutes is only a guide

A single cycle is not exactly 90 minutes for everyone, and it is shorter early in the night than later. So the cycle method is a way to pick a sensible time, not a stopwatch. The bigger wins are simpler:

- **Keep a regular wake time**, even on weekends. It steadies the whole body clock.
- **Get bright light in the first hour** of the day, and dim lights before bed.
- **Stop caffeine by the early afternoon.** It stays in the body for hours.
- **Keep the room dark, cool and quiet.** Screens and phones in bed do the opposite.
- **Do not use a late-night alarm trick to fix a bad week.** A consistent routine beats a perfect calculation.

> **When this is not about timing:** If you sleep eight hours and still feel exhausted, snore loudly, stop breathing at night, or cannot fall asleep for weeks, it is worth seeing a doctor. A bedtime calculator cannot fix a sleep disorder.

## Frequently asked questions

### How long is a sleep cycle?

About 90 minutes on average, though it varies from person to person and across the night. A full night is usually four to six cycles.

### How many hours of sleep does an adult need?

Most adults need 7 or more hours a night. Teenagers and children need more.

### Does waking at the end of a cycle really help?

Many people find it easier to wake from light sleep than from deep sleep, which is the idea behind it. The effect varies, and total sleep and a regular schedule matter more.

### How long does it take to fall asleep?

A typical healthy adult falls asleep in about 10 to 20 minutes, which is why calculators add around 15 minutes.

### Is it better to sleep 6 hours or 7.5 hours?

7.5 hours is closer to what most adults need. Six hours can work occasionally, but regularly sleeping less builds up a sleep debt.



---

# Step-up SIP: raise your SIP 10 percent a year and almost double the result

> A flat ₹10,000 SIP and one that grows 10 percent a year start in exactly the same place. Twenty years later, one has ₹1 crore and the other nearly ₹2 crore. Here is why a small yearly raise does so much.

Published 2026-10-08 by Toolcloak. 3 minute read. Original: https://toolcloak.pages.dev/articles/step-up-sip-raise-your-sip-10-percent-a-year
Related free tool: [Step-Up SIP Calculator](https://toolcloak.pages.dev/tools/step-up-sip-calculator)

Your salary is not the same in year ten as it was in year one. Your SIP probably should not be either. A step-up SIP raises the monthly amount by a fixed percentage every year, and over twenty years that one habit does more than a better fund ever could.

> **In short**
> - ₹10,000 a month for 20 years at 12 percent ends near **₹1 crore**. The same SIP raised 10 percent a year ends near **₹2 crore**.
> - The step-up version put in about 2.9 times the money, but ended with 2 times the corpus. The raise is the cost, the compounding is the reward.
> - Match the step-up to your salary hike. Even 5 percent a year adds about 38 percent to the final amount.

## What a step-up SIP actually is

A normal SIP invests the same amount every month, forever. A step-up (or top-up) SIP keeps the same plan but adds a fixed percentage once a year. With a 10 percent step-up, ₹10,000 becomes ₹11,000 in year two, ₹12,100 in year three, and so on. You never notice any single raise. You notice the total.

| Year | Monthly SIP at 10% step-up |
| --- | --- |
| Year 1 | ₹10,000 |
| Year 5 | ₹14,641 |
| Year 10 | ₹23,579 |
| Year 15 | ₹37,975 |
| Year 20 | ₹61,159 |

*The amount grows 10 percent every year. By year 20 the monthly SIP is about six times the first one, which is roughly what a salary growing 10 percent a year does too.*

## Same start, very different finish

All three plans below start at ₹10,000 a month and assume 12 percent a year. The only difference is the yearly raise.

**Value of the SIP over 20 years (lakh rupees)**

| Year | 10% step-up | 5% step-up | Flat SIP |
| --- | --- | --- | --- |
| 0 | 0 | 0 | 0 |
| 5 | 9.8 | 9 | 8.2 |
| 10 | 33.7 | 27.9 | 23.2 |
| 15 | 86.8 | 65.3 | 50.5 |
| 20 | 198.9 | 137.4 | 99.9 |

*For the first five years the three lines are almost one line. The gap only opens later, which is why people give up on step-ups too early.*

**After 20 years**

|  | Value | Note |
| --- | --- | --- |
| Flat SIP | ₹1.0 Cr | You invested ₹24 lakh |
| 5% step-up | ₹1.37 Cr | You invested ₹39.7 lakh |
| 10% step-up | ₹1.99 Cr | You invested ₹68.7 lakh |

*Rounded from a monthly calculation at 12 percent a year. Real returns are never this smooth.*

Try it: [Step-Up SIP Calculator](https://toolcloak.pages.dev/tools/step-up-sip-calculator). Enter your monthly amount, your yearly raise and the years. It shows the corpus and how much of it you put in.

## Where the money comes from

A bigger corpus does not only come from putting in more. It also comes from putting the bigger amounts in while the earlier money is already growing. Look at the 10 percent plan:

**Where the ₹1.99 crore comes from (10% step-up)**

|  | Share |
| --- | --- |
| Money you put in | ₹68.7 L |
| Growth on top | ₹130.2 L |

*About two thirds of the final amount is growth. In the flat SIP it is about three quarters, because you put in less to begin with.*

## How big a raise is realistic

- **Tie it to your hike.** If your pay rises 8 to 10 percent, step up by about half of that. You will not feel it.
- **Start smaller than you think.** A 5 percent step-up is easier to keep than a 15 percent one you stop in year three.
- **Do it in the same month as the raise.** The new money has not become a habit yet, so it is the cheapest time to redirect it.
- **Review once a year, not every month.** Step-ups work because they run on their own.

> **A fair warning:** These numbers assume a steady 12 percent. Markets do not do that. Some years are negative. The point of the comparison is the **gap between plans**, which stays real even when the absolute numbers change.

## Step-up versus a bigger starting SIP

If you could afford more at the start, you should start higher. But most people cannot, and the step-up is how a modest start still reaches a large number. A flat ₹20,000 SIP reaches about ₹2 crore too, but it needs you to find that money on day one.

**Two ways to reach about ₹2 crore in 20 years**

| Flat SIP | Step-up SIP |
| --- | --- |
| ₹20,000 / month | ₹10,000 + 10% a year |
| Needs the full amount from month one | Starts at half the amount |
| Invests ₹48 lakh in total | Invests about ₹69 lakh in total |

*The step-up asks less at the start and more later, when your income is higher.*

## Frequently asked questions

### What is a step-up SIP?

A SIP whose monthly amount rises by a fixed percentage every year, for example 10 percent. It lets your investments grow along with your income.

### How much does a 10 percent step-up add?

In this example, ₹10,000 a month for 20 years at 12 percent gives about ₹1 crore flat, and about ₹1.99 crore with a 10 percent yearly step-up.

### Is a step-up SIP better than a normal SIP?

If your income rises over time, usually yes, because it keeps your investing in step with your earnings. It is not a way to earn a higher return, only a way to invest more over time.

### What step-up percentage should I choose?

A common choice is 5 to 10 percent, tied to your salary growth. Choose a number you can keep for many years rather than a big one you stop.

### Are these returns guaranteed?

No. The 12 percent a year is an assumption for illustration. Actual market returns vary and can be negative in some years.



---

# Averaging down: how buying more at a lower price changes your average, and what it cannot fix

> When a share you own falls, buying more lowers your average cost and cuts the rise needed to break even. This guide shows the exact arithmetic, why a fixed monthly amount buys at a lower average than the average price, and the risk that no formula removes.

Published 2026-10-08 by Toolcloak. 3 minute read. Original: https://toolcloak.pages.dev/articles/averaging-down-how-buying-more-at-a-lower-price-changes-your-average-and-what-it-cannot-fix
Related free tool: [Stock Average Calculator (Average Buy Price)](https://toolcloak.pages.dev/tools/stock-average-calculator)

You bought 100 shares at ₹500. The price has dropped to ₹400. Buying 100 more would bring your average cost down, but by how much, and does it actually help? The arithmetic is simple, and it is worth seeing once, because it shows both the benefit and the trap of “averaging down”.

> **In short**
> - **Average price = total money spent ÷ total shares bought.** It is a weighted average, not the simple mean of the prices.
> - Buying 100 at ₹500 and 100 at ₹400 gives an average of **₹450**. The price now only needs to rise **12.5%** to break even, instead of 25%.
> - A **fixed amount each month** buys more shares when the price is low, so the average cost (₹76.92) ends **below the average price** (₹82).
> - Averaging lowers your break-even price, but it also **puts more money into a falling share**. It does not make a bad business good.

## The formula

To find your average buying price, add up everything you paid and divide by everything you now hold. Do not average the prices: a purchase of 200 shares matters twice as much as one of 100.

> **Average price = (q₁ × p₁ + q₂ × p₂ + …) ÷ (q₁ + q₂ + …):** Each purchase has a quantity q and a price p. Multiply, add up the amounts, then divide by the total quantity. Brokerage and taxes can be added to the cost for a more complete figure.

Try it: [Stock Average Calculator (Average Buy Price)](https://toolcloak.pages.dev/tools/stock-average-calculator). Enter each purchase and get the average price, total cost and break-even price.

## One example, three purchases

You buy in three steps as a share keeps falling. Each purchase changes the average by a smaller amount, because the old holding grows larger.

| Step | Bought | Price | Spent so far | Shares held | Average price |
| --- | --- | --- | --- | --- | --- |
| 1 | 100 | ₹500 | ₹50,000 | 100 | ₹500 |
| 2 | 150 | ₹400 | ₹1,10,000 | 250 | ₹440 |
| 3 | 200 | ₹320 | ₹1,74,000 | 450 | ₹386.67 |

*After the third purchase you hold 450 shares that cost ₹1,74,000 in total, an average of about ₹386.67 each.*

**The price you pay against the average you hold**

| Purchase | Price at the time | Your average cost |
| --- | --- | --- |
| Buy 1 | 500 | 500 |
| Buy 2 | 400 | 440 |
| Buy 3 | 320 | 386.67 |

*The average always stays above the latest price while the price is falling, and it moves toward it more slowly than the price itself.*

**How far the price must rise to break even at ₹320 (%)**

|  | Rise needed |
| --- | --- |
| No averaging (cost ₹500) | 56.3% |
| After buy 2 (cost ₹440) | 37.5% |
| After buy 3 (cost ₹386.67) | 20.8% |

*Without averaging you would need a 56% rise from ₹320 to get back to ₹500. With the third purchase, only about 21%.*

That looks like an easy win, and it is the main attraction of averaging down. The catch is on the next line of the table.

**How much money is now riding on the share**

| Step | Value |
| --- | --- |
| After buy 1 | ₹50,000 |
| After buy 2 | ₹1,10,000 |
| After buy 3 | ₹1,74,000 |

*Your cost per share went down, but the total money at risk more than tripled. If the share keeps falling, you lose on a bigger base.*

> **A lower average is not a lower risk:** Averaging down only works if the share eventually recovers. If the business has a real problem, each purchase adds money to a falling asset. Many experienced investors set a total limit on how much they will put into one share before they start.

## Why a fixed amount buys at a lower average

There is a second kind of averaging, which is investing the same **amount** at regular times (as in a monthly SIP) instead of buying the same **number** of shares. Because a fixed amount buys more shares when the price is low, it automatically weights the cheaper prices more heavily.

| Month | Price | Amount invested | Shares bought |
| --- | --- | --- | --- |
| 1 | ₹100 | ₹10,000 | 100 |
| 2 | ₹80 | ₹10,000 | 125 |
| 3 | ₹50 | ₹10,000 | 200 |
| 4 | ₹80 | ₹10,000 | 125 |
| 5 | ₹100 | ₹10,000 | 100 |

*The price falls from 100 to 50 and recovers to 100. Total invested ₹50,000 for 650 shares.*

**Two ways to average the same prices**

| Simple average of the prices | What you actually paid |
| --- | --- |
| ₹82.00 | ₹76.92 |
| (100 + 80 + 50 + 80 + 100) ÷ 5 | ₹50,000 ÷ 650 shares |
| Ignores how many shares you got | Cheaper months bought more shares |
| Not what you actually paid | The harmonic mean of the prices |

*The gap is the benefit of buying a fixed amount: you paid about 6% less per share than the average price.*

**Result when the price returns to ₹100**

|  | Value | Note |
| --- | --- | --- |
| Invested | ₹50,000 | Five equal instalments |
| Value at ₹100 | ₹65,000 | 650 shares × ₹100 |
| Gain | +30% | Even though the price ended where it began |

*The price is flat from start to finish, but the investor gains 30%. Falling prices in the middle gave extra shares at a discount.*

This is the idea behind rupee-cost averaging. It does not guarantee a profit, and it works only if prices recover. See [the SIP calculator article](https://toolcloak.pages.dev/articles/5000-rupees-a-month-for-20-years-what-a-sip-calculator-can-tell-you) for the long-run view.

## Why losses are harder to recover than they look

A 50% fall needs a 100% rise to recover. This asymmetry is why a lower average cost feels so valuable. The same maths is explained in [up 50%, down 50%](https://toolcloak.pages.dev/articles/up-50-percent-down-50-percent-and-you-are-not-back-to-zero).

**Rise needed to recover from a fall (%)**

|  | Value |
| --- | --- |
| −10% | 11.1% |
| −20% | 25% |
| −33% | 49.3% |
| −50% | 100% |
| −75% | 300% |

*A 10% loss needs 11%. A 50% loss needs 100%. A 75% loss needs 300%.*

## Quick practice

1. You hold 40 shares at ₹250 and buy 60 more at ₹200. Average? (10,000 + 12,000) ÷ 100 = **₹220**.
2. The price is ₹180. What gain do you need to break even? 220 ÷ 180 − 1 = **22.2%**.
3. You invest ₹5,000 at ₹50, then ₹5,000 at ₹25. Average cost? 10,000 ÷ 300 shares = **₹33.33**, not ₹37.50.
4. Should you average down on every fall? **No.** Decide in advance how much you are willing to hold in one share.

This article explains arithmetic, not what to buy, and it is not investment advice. Past prices do not predict future ones, so make decisions based on your own situation or with a registered adviser.

## Frequently asked questions

### How do you calculate the average price of a stock?

Divide the total money you spent (quantity × price for each purchase, plus costs) by the total number of shares you hold.

### What is averaging down?

Buying more of a share after its price has fallen, so your average cost per share drops and the price needs a smaller rise to reach your break-even.

### Is averaging down a good idea?

It depends on why the price fell. It lowers your average cost, but it also puts more money into the same share. It can hurt if the share keeps falling.

### What is rupee-cost averaging?

Investing a fixed amount at regular intervals, as in a SIP. You buy more units when prices are low and fewer when they are high.

### Why is my average lower than the average of the prices?

Because a fixed amount buys more shares at the low prices. The result is the harmonic mean of the prices, which is always at or below the simple mean.

### Should brokerage be included in the average price?

For a true break-even figure, yes. Add buying and selling costs and taxes to the total cost.



---

# SWP: how long will ₹1 crore last when you withdraw every month?

> A systematic withdrawal plan pays you a fixed amount each month from an invested corpus. At an 8% return, ₹50,000 a month from ₹1 crore never runs out if the amount stays flat, but if your withdrawals rise 6% a year with inflation, the money lasts about 21 years. The difference is the whole lesson.

Published 2026-10-08 by Toolcloak. 3 minute read. Original: https://toolcloak.pages.dev/articles/swp-how-long-will-1-crore-last-when-you-withdraw-every-month
Related free tool: [SWP Calculator](https://toolcloak.pages.dev/tools/swp-calculator)

Building a corpus is half the job. The other half is turning it into a monthly income without running out. A systematic withdrawal plan (SWP) does that by selling a fixed amount from your investment each month. How long the money lasts depends on three things: the amount you take, the return the rest earns, and how fast you raise your withdrawal.

> **In short**
> - On a ₹1 crore corpus at **8% a year**, a **flat ₹50,000 a month** (a 6% withdrawal rate) does not run out and the balance keeps growing.
> - If the same ₹50,000 **rises 6% every year** to keep up with prices, the corpus lasts about **21 years**.
> - Starting at **₹40,000** with the same rise lasts about **29 years**, and **₹60,000** about **17 years**.
> - Returns are not smooth in real life. A bad market early in retirement hurts more than the same drop late. This is a model, not a promise.

## How an SWP works

You hold a lump sum in an investment, usually a mutual fund. Every month an amount you choose is sold and sent to your bank. The units that remain keep earning. If the return on the balance is higher than your withdrawals, the corpus grows. If not, it shrinks.

> **Each month: new balance = (old balance − withdrawal) × (1 + monthly return):** At 8% a year the monthly return is about 0.667%. On ₹1 crore that is roughly ₹66,700 of growth a month, which is more than a ₹50,000 withdrawal. That is why a flat ₹50,000 never drains the pot.

## The trap: a flat withdrawal feels safe

A flat ₹50,000 looks sustainable because it is smaller than the monthly growth. But prices rise. ₹50,000 today buys much less in 15 years. To keep your lifestyle you must raise the withdrawal, and the SWP then has to carry bigger and bigger amounts.

**₹1 crore at 8%: balance over time when withdrawals rise 6% a year (lakh rupees)**

| Years | Start ₹40,000 a month | Start ₹50,000 a month | Start ₹60,000 a month |
| --- | --- | --- | --- |
| 0 | 100 | 100 | 100 |
| 5 | 116 | 108 | 99 |
| 10 | 129 | 105 | 82 |
| 15 | 132 | 83 | 33 |
| 20 | 118 | 24 | 0 |
| 25 | 70 | 0 | 0 |
| 30 | 0 | 0 | 0 |

*Each line rises at first because returns exceed withdrawals, then bends down as the yearly rise in the withdrawal catches up. Higher starting amounts bend down sooner.*

Try it: [SWP Calculator](https://toolcloak.pages.dev/tools/swp-calculator). Enter your corpus, monthly withdrawal, return and yearly rise to see the balance and how long it lasts.

## How long does it last?

**Years the ₹1 crore lasts at 8% when withdrawals rise 6% a year**

|  | Years |
| --- | --- |
| Start ₹30,000 | 44 yrs |
| Start ₹40,000 | 29 yrs |
| Start ₹50,000 | 21 yrs |
| Start ₹60,000 | 17 yrs |

*Adding just ₹10,000 to the starting withdrawal can cut years from the plan. At ₹30,000 the corpus runs for 44 years.*

## The three levers

**Years ₹1 crore lasts at ₹50,000 a month, by return and yearly rise in withdrawal**

| Return | Rise 0% | Rise 3% | Rise 6% | Rise 8% |
| --- | --- | --- | --- | --- |
| 6% | 89 yrs | 24 yrs | 17 yrs | 15 yrs |
| 8% | Never ends | 37 yrs | 21 yrs | 18 yrs |
| 10% | Never ends | Never ends | 30 yrs | 22 yrs |

Note that the table is a simplified model with a smooth return and fixed yearly rises. It shows direction, not forecast.

## Sequence risk: the order of returns matters

A smooth 8% every year is easy to calculate and impossible to find. In real markets, returns come in uneven years. If the early years are bad while you are also withdrawing, you sell more units at low prices, and those units are not there to recover later. This is called sequence risk.

**Same average return, different order**

| Bad years first | Good years first |
| --- | --- |
| Runs out sooner | Lasts longer |
| You withdraw while the balance is low | Balance grows early, giving a cushion |
| More units sold at cheap prices | Fewer units sold per rupee withdrawn |
| Less left to recover when markets rise | Later falls hurt less |

*This is why many planners suggest keeping 1 to 2 years of withdrawals in a safer place such as a short-term debt fund or FD, so you do not have to sell in a crash.*

## A rough withdrawal-rate check

A quick rule is the withdrawal rate: yearly withdrawal ÷ corpus. In our example ₹6 lakh ÷ ₹1 crore is 6%. The old “4% rule” comes from studies on long retirements and is a starting point, not a law, and was built on a different market and currency. See [the 25x rule](https://toolcloak.pages.dev/articles/the-25x-rule-how-much-is-enough-to-stop-working) for how it links to your target corpus.

**Starting withdrawal rate (a rough guide, not an official band)**

| Zone | From | To |
| --- | --- | --- |
| Cautious | 2 | 4 |
| Stretch | 4 | 6 |
| Risky | 6 | 9 |

*For a retirement that may last 30 years or more, many planners prefer to start near the low end and adjust each year.*

## SWP versus fixed deposit interest

- **SWP sells units, so some of each payout is your own capital.** Tax is usually on the gain portion only, but the rules change, so check the current treatment.
- **FD interest is fully taxable each year** at your slab, even if you reinvest it. Compare after-tax numbers with our [FD after tax and inflation guide](https://toolcloak.pages.dev/articles/fd-at-7-percent-after-tax-and-inflation-what-you-actually-earn).
- **SWP returns are not guaranteed.** An FD’s rate is fixed. An SWP from an equity fund can go up or down.
- **You can change an SWP any time.** You can raise, cut or stop the payout, which helps in a bad year.

> **A safer way to set it up:** Pick a starting amount that works at a lower return than you expect, such as 6% or 7%. Review once a year. In a bad year, take less or hold the rise. Small adjustments early are worth far more than big ones late.

## Frequently asked questions

### What is a systematic withdrawal plan?

An SWP is an instruction to sell a fixed amount from your investment every month and send it to your bank, while the rest stays invested.

### How long will ₹1 crore last with ₹50,000 a month?

At 8% a year with a flat withdrawal it does not run out. If the withdrawal rises 6% a year, it lasts about 21 years in this model.

### What is a safe withdrawal rate?

Many planners start near 3% to 4% for long retirements, but there is no single safe figure. It depends on returns, inflation and how long you need the money.

### Is SWP better than FD interest?

It depends on tax, risk and flexibility. An SWP from a market-linked fund can grow and is flexible but not guaranteed. An FD gives a fixed rate that is taxed each year.

### What is sequence of returns risk?

It is the risk that poor returns early in your withdrawal period shrink the corpus more than the same returns later, because you sell more units while prices are low.



---

# Celsius, Fahrenheit and Kelvin: the conversion, a mental shortcut, and the one place they agree

> 35°C is a hot day, 95°F is the same hot day, and 308 K is the same day again. This guide gives you the exact formulas, a shortcut that is good enough for the weather, a cooking chart for oven settings, and the surprising temperature at which Celsius and Fahrenheit show the same number.

Published 2026-10-08 by Toolcloak. 5 minute read. Original: https://toolcloak.pages.dev/articles/celsius-fahrenheit-and-kelvin-the-conversion-the-shortcut-and-the-one-place-they-agree
Related free tool: [Temperature Converter](https://toolcloak.pages.dev/tools/temperature-converter)

A recipe from abroad says to bake at 350°F. Your oven only shows Celsius. A weather app in another country says 95°F and you wonder whether to pack a jacket. Three temperature scales are in everyday use, and moving between them needs one small formula and a little feel for the numbers. Here is everything in one place, with a way to do it in your head.

> **In short**
> - **°F = °C × 9/5 + 32** and **°C = (°F − 32) × 5/9**. **K = °C + 273.15**.
> - Water freezes at **0°C = 32°F = 273.15 K** and boils at **100°C = 212°F = 373.15 K** (at sea level).
> - **Shortcut:** to go from °C to °F, **double it and add 30**. It is exact at 10°C and off by only a few degrees in everyday weather.
> - At **−40**, Celsius and Fahrenheit show the same number.

## Three scales, two fixed points

Every temperature scale needs two reference points and a decision about how many steps lie between them. **Celsius** puts the freezing point of water at 0 and its boiling point at 100, so there are 100 steps. **Fahrenheit** uses 32 and 212, so there are 180 steps between the same two events. **Kelvin** uses the same step size as Celsius but starts from absolute zero, the coldest temperature possible, so it never goes negative.

**Same water, three labels**

| Water freezes | Water boils (sea level) |
| --- | --- |
| 0°C = 32°F = 273.15 K | 100°C = 212°F = 373.15 K |
| Celsius: 0 | Celsius: 100 |
| Fahrenheit: 32 | Fahrenheit: 212 |
| Kelvin: 273.15 | Kelvin: 373.15 |

*The boiling point drops as you go up in altitude, so 100°C holds only at sea level pressure.*

Because Fahrenheit packs 180 steps into the span where Celsius has 100, one Fahrenheit degree is smaller: **1°C = 1.8°F**. A change of 10°C in the weather is a change of 18°F. Kelvin steps are the same size as Celsius steps, so a change of 10 K is a change of 10°C.

Try it: [Temperature Converter](https://toolcloak.pages.dev/tools/temperature-converter). Type a temperature in any scale and see it in the other two at once.

## The formulas

> **°F = °C × 9 ÷ 5 + 32     °C = (°F − 32) × 5 ÷ 9     K = °C + 273.15:** For the first, multiply by 1.8 and then add 32. For the second, subtract 32 first, then multiply by 5/9 (about 0.556). The order matters: the 32 is added after multiplying going up and subtracted before multiplying going down.

**Converting 35°C to Fahrenheit**

1. **Multiply by 1.8.** 35 × 1.8 = 63.
2. **Add 32.** 63 + 32 = 95.
3. **Say it.** 35°C is 95°F, a hot day.

**Converting 350°F to Celsius**

1. **Subtract 32.** 350 − 32 = 318.
2. **Multiply by 5/9.** 318 × 5 ÷ 9 = 176.7.
3. **Round for the oven.** About 177°C, which most Indian ovens set as 180°C.

## A shortcut for your head

For weather, you rarely need exactness. Take the Celsius value, **double it, and add 30** to get a quick Fahrenheit guess. It works because the exact rule is “× 1.8 + 32”, and doubling and adding 30 is a close, easier cousin. It is perfect at 10°C and drifts as you move away.

**Exact Fahrenheit against the double-and-add-30 shortcut**

| Degrees Celsius | Exact °F | Shortcut: 2 × °C + 30 |
| --- | --- | --- |
| −40 | -40 | -50 |
| −30 | -22 | -30 |
| −20 | -4 | -10 |
| −10 | 14 | 10 |
| 0 | 32 | 30 |
| 10 | 50 | 50 |
| 20 | 68 | 70 |
| 30 | 86 | 90 |
| 40 | 104 | 110 |

*The two lines cross at 10°C. Between 0°C and 30°C, the weather range for most people, the shortcut is off by at most 4°F.*

**How far the shortcut is off (size of the error, °F)**

|  | Error |
| --- | --- |
| 0°C | 2 too low |
| 10°C | 0 |
| 20°C | 2 too high |
| 30°C | 4 too high |
| 40°C | 6 too high |

*The error is 0.2 × (°C − 10) degrees: tiny near 10°C and larger on very hot or very cold days. On hot days, take a few degrees off the guess.*

> **Going back: subtract 30, then halve:** To guess Celsius from Fahrenheit, subtract 30 and halve. 80°F gives 25°C (exact: 26.7). It is a rough guide, fine for weather and not for medicine, cooking or science.

## Landmarks to memorise

A few anchor points make conversions feel natural, because you can place any new number between two landmarks you already know.

**Temperatures worth knowing**

| When | What |
| --- | --- |
| −273.15°C | Absolute zero (0 K, −459.67°F). The lowest temperature possible. Nothing can be colder. |
| −40°C | The scales meet. −40°C and −40°F are the same temperature. |
| 0°C | Water freezes (32°F). Ice at the edge of melting. |
| 20°C | A comfortable room (68°F). Many people set air conditioners between 24 and 26°C (75–79°F). |
| 37°C | Typical body temperature (98.6°F). Normal varies by person and time of day, roughly 36.1 to 37.2°C. |
| 100°C | Water boils (212°F). At sea level. In the mountains it boils at a lower temperature. |

**Everyday weather on the Celsius scale**

| Zone | From | To |
| --- | --- | --- |
| Freezing | -10 | 0 |
| Cool | 0 | 15 |
| Mild | 15 | 25 |
| Hot | 25 | 38 |
| Extreme | 38 | 50 |

*These bands are casual descriptions, not official categories. Heat warnings depend on the local climate and on humidity.*

## Why −40 is special

If a number is the same in both scales, then C = F. Put that in the formula: C = C × 1.8 + 32. Move terms: C − 1.8C = 32, so −0.8C = 32, and C = −40. At −40 the Celsius and Fahrenheit thermometers show the same number. Winter in parts of Canada and Siberia does reach it, and weather reporters get to say “minus forty, either way”.

**Fahrenheit against Celsius: the lines meet at −40**

| Degrees Celsius | °F for each °C |
| --- | --- |
| −40 | -40 |
| −20 | -4 |
| 0 | 32 |
| 20 | 68 |
| 40 | 104 |
| 60 | 140 |
| 80 | 176 |
| 100 | 212 |

*A straight line: every extra °C adds 1.8°F. The line passes through (−40, −40), (0, 32) and (100, 212).*

## Kelvin and why scientists use it

Kelvin starts at absolute zero, so there are no negative numbers, and a temperature of 200 K is genuinely twice as hot as 100 K in terms of thermal energy. That does not work with Celsius: 20°C is not “twice as hot” as 10°C. Science uses Kelvin for gas laws, colour temperatures and anything where ratios matter. The unit is written as K, without a degree sign.

| Event | °C | °F | K |
| --- | --- | --- | --- |
| Absolute zero | −273.15 | −459.67 | 0 |
| Water freezes | 0 | 32 | 273.15 |
| Room temperature | 20 | 68 | 293.15 |
| Body temperature | 37 | 98.6 | 310.15 |
| Water boils | 100 | 212 | 373.15 |

*To get Kelvin, just add 273.15 to the Celsius value.*

## Oven temperatures

Recipes use round numbers in whichever scale the author prefers. The usual conversions are rounded to the nearest 5 or 10 degrees, since ovens are not precise to the last degree.

**Common oven settings**

| Style | °C | °F |
| --- | --- | --- |
| Low and slow | 150 | 302 |
| Moderate | 180 | 356 |
| Hot (roasting) | 200 | 392 |
| Very hot (pizza, bread) | 220 | 428 |

For measuring ingredients as well, the [cooking measurement converter](https://toolcloak.pages.dev/tools/cooking-measurement-converter) turns cups and spoons into grams and millilitres.

## Fever and thermometers

For health use, be exact. 98.6°F is 37°C, but 100.4°F, a common fever threshold, is 38°C. Different thermometers (mouth, ear, forehead) read a little differently, so compare like with like and follow a doctor’s advice for decisions about treatment. A shortcut guess is fine for the weather, never for a child’s fever.

1. **Weather:** use double-and-add-30 and enjoy the quick answer.
2. **Cooking:** use the exact formula and round to the nearest 5 or 10.
3. **Health:** use the converter and the exact figure.
4. **Science:** use Kelvin so that ratios and gas laws work.

The relationship is linear, which means you can always check a result by plausibility: 100°C should be 212°F, 0°C should be 32°F, and anything in between lies on the line. If you are scaling recipes or measurements too, see [aspect ratios](https://toolcloak.pages.dev/articles/16-9-4-3-1-1-aspect-ratios-without-cropping-surprises) for another everyday ‘conversion’ with the same straight-line logic. Use the converter above, and the next time a foreign weather app says 95°F, you will know to leave the jacket at home.

## Frequently asked questions

### How do I convert Celsius to Fahrenheit?

Multiply the Celsius value by 1.8 (9/5) and add 32. For example 35°C × 1.8 + 32 = 95°F.

### How do I convert Fahrenheit to Celsius?

Subtract 32 and multiply by 5/9. For example (350 − 32) × 5 ÷ 9 = 176.7°C.

### At what temperature are Celsius and Fahrenheit equal?

At −40. Setting C = 1.8C + 32 gives C = −40, so −40°C and −40°F are the same temperature.

### What is 98.6°F in Celsius?

37°C, the traditional figure for normal body temperature. Normal varies from person to person.

### How do I convert Celsius to Kelvin?

Add 273.15. Water freezes at 273.15 K and boils at 373.15 K at sea level.

### What is a quick way to convert Celsius to Fahrenheit?

Double the Celsius number and add 30. It is exact at 10°C and good enough for weather, but not for cooking, medicine or science.



---

# How the 6-digit code in your authenticator app works (TOTP explained)

> Your phone shows a six-digit code that changes every 30 seconds and the website somehow knows it, with no internet connection between the two. The trick is a shared secret and a shared clock. This guide walks through exactly what happens, how safe it is, and where it still fails.

Published 2026-10-08 by Toolcloak. 6 minute read. Original: https://toolcloak.pages.dev/articles/how-the-6-digit-code-in-your-authenticator-app-works-totp-explained
Related free tool: [TOTP Code Generator](https://toolcloak.pages.dev/tools/totp-generator)

When you turn on two-factor authentication, a site shows you a QR code. You scan it once with an authenticator app, and from then on the app displays a six-digit number that changes twice a minute. You type it in, and you are let through. The app never talks to the site. So how do both sides arrive at the same number? The answer is a small, elegant recipe called TOTP.

> **In short**
> - **TOTP** (time-based one-time password) mixes a **shared secret** with the **current time** to produce a short code.
> - The secret is sent **once**, in the QR code. After that, your phone and the server each compute the code on their own.
> - The time is cut into **30-second steps**. The code only changes when the step number changes.
> - A random guess succeeds about **1 time in a million** per try, so rate limits matter. TOTP is far better than a password alone, but it **can still be phished**.

## The two ingredients

TOTP, defined in an open standard (RFC 6238), needs only two things that both sides already have: a **secret key** and the **clock**. The secret is a random string chosen by the server when you set up 2FA. The QR code you scan is just that secret, plus a label for the account, written in a way the app can read. The app stores it and the server stores it. That is the only moment the secret travels.

The clock is the second ingredient. Phones and servers keep time closely synchronised through the internet, so both can agree on “what time is it now” to within a second or two without talking to each other.

**What each side holds**

| Your phone | The website |
| --- | --- |
| Secret + clock | Secret + clock |
| Stored the secret when you scanned the QR | Stored the same secret for your account |
| Reads the current time from the phone | Reads the current time from the server |
| Works with no signal at all | Computes the code it expects and compares |

*Same inputs on both sides give the same output. That is why the app works in airplane mode.*

Try it: [TOTP Code Generator](https://toolcloak.pages.dev/tools/totp-generator). Paste a secret and see the live code and the seconds left. Use test secrets only, never your real account secret on a site you do not trust.

## From time to a counter

The first step is to turn the time into a whole number that changes only every 30 seconds. Take the Unix time (the seconds since 1 January 1970; see [Unix time explained](https://toolcloak.pages.dev/articles/unix-time-the-number-that-counts-seconds-since-1970)) and divide by 30, dropping the fraction. The result is the **counter**.

> **counter = floor(Unix time ÷ 30):** At Unix time 1,700,000,010 the counter is 56,666,667. It stays 56,666,667 for the next 30 seconds and becomes 56,666,668 at 1,700,000,040.

**Three moments and the counter they produce**

| Unix time | Counter | Same code as the first? |
| --- | --- | --- |
| 1,700,000,010 | 56,666,667 | Yes (start) |
| 1,700,000,039 | 56,666,667 | Yes (29 s later) |
| 1,700,000,040 | 56,666,668 | No, a new code |

**One 30-second step**

| When | What |
| --- | --- |
| 0 s | A new code appears. The counter just ticked up, so the app computes a fresh code. |
| 15 s | Halfway. The same code is shown. You have comfortable time to type it. |
| 25 s | Time is short. Many apps show the ring turning red. Wait for the next code if you are slow. |
| 30 s | Next step. The counter becomes N + 1 and the old code stops being the current one. |

## From counter and secret to six digits

Now the counter and the secret are mixed using **HMAC**, a keyed hash (the idea is explained in [what a hash is](https://toolcloak.pages.dev/articles/what-a-hash-is-and-why-changing-one-letter-changes-everything)). A hash scrambles its input so thoroughly that even a tiny change gives a completely different output, and HMAC adds the secret so that only someone who knows it can produce the same result. The output is a long block of bytes, much too long to type.

So the standard shrinks it with a step called **dynamic truncation**: look at the last byte, use its low four bits as a starting position, take four bytes from that position, drop the top bit, and read them as a number. Finally take that number modulo 1,000,000 (the remainder after dividing) to get exactly six digits, padding with zeros at the front if needed.

**The whole recipe**

1. **Read the clock.** Unix time, for example 1,700,000,010.
2. **Make the counter.** Divide by 30 and drop the fraction: 56,666,667.
3. **Mix with the secret.** Run HMAC over the counter using the shared secret (SHA-1 by default in most apps).
4. **Truncate.** Pick four bytes from a position chosen by the last byte of the result.
5. **Keep six digits.** Take the number modulo 1,000,000 and pad with leading zeros.

*You never need to do this by hand. The point is that every step is public and repeatable, and only the secret is private.*

> **Why the secret looks like “JBSWY3DPEHPK3PXP”:** Secrets are usually shown in Base32, a text format with 32 symbols (capital letters and digits 2 to 7), chosen to avoid confusing characters. Each Base32 character carries 5 bits, so a 16-character secret holds 80 bits and a 32-character one holds 160 bits. Compare the encoding step with the [Base32 encoder](https://toolcloak.pages.dev/tools/base32-encoder).

## How safe is a six-digit code?

There are only 1,000,000 possible six-digit codes, so a blind guess has a 1 in a million chance. That sounds weak, and by itself it would be, but the code is never the only barrier. It works together with a password, and servers limit attempts. Because clocks drift slightly, servers usually accept the previous, current and next step, so three codes are valid at once: the odds of a lucky guess become about 3 in a million.

**Odds of guessing blindly**

|  | Value | Note |
| --- | --- | --- |
| One try, one valid code | 1 in 1,000,000 | Six digits, 10⁶ possibilities |
| One try, three valid codes | 1 in 333,333 | With a ±1 step grace window |
| Five tries in a row | 1 in 66,667 | Without a lockout, 5 × 3 in a million |

*Real services lock the account or slow down after a handful of wrong codes, so an attacker cannot simply try hundreds of thousands of codes.*

**What TOTP protects against, and what it does not**

| Box | What it means |
| --- | --- |
| Stolen password | A leaked password alone is no longer enough. The attacker also needs the code. |
| Reused password in a breach | Other sites’ leaks cannot unlock this account without your phone. |
| Real-time phishing | A fake site can ask for your code and pass it on within 30 seconds. TOTP does not stop this. |
| Lost or wiped phone | Without a backup, you may be locked out yourself. Save the recovery codes. |

*TOTP is a strong everyday layer, not a perfect one. Hardware keys and passkeys, which are tied to the real site address, resist phishing better.*

## TOTP, SMS codes and passkeys

Text-message codes are easy to use but depend on your phone number, which can be hijacked through SIM-swap fraud or intercepted, and on the network being available. TOTP codes are generated on your device and need no signal. Passkeys and hardware security keys go further by cryptographically tying the login to the site, so a copied page cannot trick you.

**Rough comparison of second factors**

| Method | Works offline | Hard to phish | Easy to set up |
| --- | --- | --- | --- |
| SMS code | No | No | Very easy |
| Authenticator app (TOTP) | Yes | No | Easy |
| Passkey or hardware key | Yes | Yes | Moderate |

None of these is useless. Any second factor is far better than none, because most account takeovers rely on a stolen or guessed password. If you can choose, use an authenticator app over SMS and a passkey over an app. For the password side of the equation, see [password entropy](https://toolcloak.pages.dev/articles/password-entropy-bits-guesses-and-why-length-beats-symbols).

## Practical habits

1. **Save the recovery codes** when you set up 2FA, and keep them somewhere other than your phone.
2. **Back up the secret** if your app offers encrypted backup, or note the setup key in a password manager.
3. **Keep your phone’s clock automatic.** If the codes are rejected, a wrong clock is the first thing to check.
4. **Never share the code**, even with someone who says they are from support. A valid code is a key to your account for 30 seconds.
5. **Check the address bar** before typing a code into any site.

The cleverness of TOTP is how little it needs: one secret, shared once, and the clock that is already on every device. Try the generator above with a test secret and watch the code change at the 30-second mark. It is the same maths protecting accounts all over the internet.

## Frequently asked questions

### What does TOTP stand for?

Time-based One-Time Password. It generates a short code from a shared secret and the current time, defined in RFC 6238.

### How does the app know the code without internet?

The app and the server share a secret and both use the clock. They each compute the same code independently, so no connection is needed at login time.

### Why does the code change every 30 seconds?

The time is divided into 30-second steps and the code is derived from the step number. When the step number changes, so does the code.

### Why does my code say it is invalid?

The most common cause is a wrong phone clock. Turn on automatic date and time. Also check that you typed the current code, not the one that just expired.

### Is an authenticator app safer than SMS?

Generally yes. App codes are generated on your device and cannot be intercepted through your phone number. Neither one stops a real-time phishing page.

### What happens if I lose my phone?

You need your recovery codes or a backup of the secret to get back in. Save these when you enable 2FA, and keep them away from the phone.



---

# Unix time: the number that counts seconds since 1970

> Developers store dates as a single number like 1700000000. It looks cryptic, but it is just a stopwatch started at midnight on 1 January 1970. Here is how to read it, why it has 10 or 13 digits, and what happens in 2038.

Published 2026-10-08 by Toolcloak. 3 minute read. Original: https://toolcloak.pages.dev/articles/unix-time-the-number-that-counts-seconds-since-1970
Related free tool: [Timestamp Converter](https://toolcloak.pages.dev/tools/timestamp-converter)

You open a database and a column says `1700000000`. It is a date. It means **14 November 2023, 22:13:20 UTC**. Once you know the trick, reading these takes no effort, and the 13-digit versions stop being scary too.

> **In short**
> - Unix time counts seconds since 1 January 1970, 00:00:00 UTC. No time zone is stored.
> - 10 digits means seconds, 13 digits means milliseconds.
> - A signed 32-bit counter overflows on 19 January 2038. Modern 64-bit systems are fine.

Unix time, also called epoch time, is the way most software counts time. It is the number of **seconds** that have passed since 00:00:00 UTC on 1 January 1970, a moment called the epoch. There are no months, no time zones and no daylight saving in it. Just a number that goes up by one every second.

## Why a single number is so handy

- **Easy to compare.** The bigger number is later. No need to parse “3rd Oct” against “10/03/26”.
- **Easy to subtract.** The gap between two moments is just one number minus the other, in seconds.
- **Same everywhere.** A timestamp means the same instant in Mumbai, London and Tokyo. Only how you display it changes.
- **Tiny to store.** One integer instead of a text string with a format everyone has to agree on.

Try it: [Timestamp Converter](https://toolcloak.pages.dev/tools/timestamp-converter). Paste a timestamp to see the date in UTC, ISO format and your own time zone, or pick a date to get its timestamp.

## A few famous numbers

**Moments on the Unix clock**

| When | What |
| --- | --- |
| 0 | 1 January 1970, 00:00:00 UTC. The epoch. Every Unix timestamp counts from here. |
| 1,000,000,000 | 9 September 2001, 01:46:40 UTC. The clock’s billionth second. Programmers celebrated. |
| 1,700,000,000 | 14 November 2023, 22:13:20 UTC. In India this is already 15 November, 03:43:20. |
| 2,147,483,647 | 19 January 2038, 03:14:07 UTC. The largest value a signed 32-bit number can hold. |
| 4,294,967,295 | 7 February 2106, 06:28:15 UTC. The limit if the number is unsigned 32-bit instead. |

*Every number on the left is a count of seconds. The gap from 1970 to 2038 is a little over 68 years.*

## Seconds, or milliseconds?

The classic Unix timestamp is in seconds and has **10 digits** for any date between 2001 and 2286. JavaScript and many modern systems use **milliseconds**, which makes the number 1000 times larger, so it has **13 digits**.

| Unit | Example for the same moment | Digits |
| --- | --- | --- |
| Seconds | `1700000000` | 10 |
| Milliseconds | `1700000000000` | 13 |
| Microseconds | `1700000000000000` | 16 |

*Same instant, same date: 14 November 2023, 22:13:20 UTC. If a converter shows a date in January 1970 or a year past 50,000, you picked the wrong unit.*

> **Quick check:** Count the digits. **10 digits** means seconds. **13 digits** means milliseconds. If you feed a millisecond value to a tool expecting seconds you get a date tens of thousands of years in the future.

## The time zone is only in the display

The timestamp is the same for everyone on Earth. When you convert it into a readable date you choose a time zone. The same `1700000000` is 22:13 in London and Reykjavik (UTC), 03:43 the next morning in India (UTC+5:30), and 07:13 the next morning in Tokyo (UTC+9).

This is also why bugs appear around midnight and around daylight saving changes: the stored number is perfect, but a program displays it with the wrong zone. When something looks off by a few hours, suspect the display, not the data.

## A day is always 86,400 seconds

**Handy constants**

|  | Value | Note |
| --- | --- | --- |
| Seconds in a minute | 60 |  |
| Seconds in an hour | 3,600 | 60 × 60 |
| Seconds in a day | 86,400 | 24 × 3,600 |
| Seconds in a normal year | 31,536,000 | 365 × 86,400 |

*To get tomorrow at this time, add 86,400 to the timestamp. Unix time ignores leap seconds and treats every day as exactly this long.*

That makes simple jobs easy. “One week from now” is the current timestamp plus 604,800. “Thirty days ago” is minus 2,592,000. The tricky parts, like calendar months with different lengths, are where a real date library earns its keep.

## The year 2038 problem

Older systems stored the timestamp in a signed 32-bit integer. The biggest value that fits is 2,147,483,647, and one second later it wraps around to a large negative number, which a computer would read as December 1901. The moment of that overflow is **19 January 2038 at 03:14:07 UTC**.

Most modern software uses 64 bits, which lasts for hundreds of billions of years, so the problem is mostly solved on new systems. It still matters for old embedded devices and code that stores dates in 32-bit fields, and it is the reason people test their systems with dates past 2038.

## Frequently asked questions

### What is a Unix timestamp?

The number of seconds since 00:00:00 UTC on 1 January 1970. It identifies one exact moment independent of any time zone.

### How can I tell if a timestamp is in seconds or milliseconds?

Count digits. Ten digits is seconds, thirteen is milliseconds, for dates in the current era.

### Does a Unix timestamp include a time zone?

No. It always refers to UTC. A time zone only comes into play when you convert the number into a readable local date.

### What is the year 2038 problem?

Systems storing Unix time in a signed 32-bit number overflow on 19 January 2038 at 03:14:07 UTC. Modern 64-bit systems are not affected.

### How do I convert a date to a timestamp?

Count the seconds from 1 January 1970 UTC to your date, or use a converter. The timestamp for 8 October 2026 at 00:00 UTC is 1791417600.



---

# Can two UUIDs ever be the same? The odds, in pictures

> A random UUID has 122 random bits, a number with 37 digits. You could generate a billion of them every second for 86 years before a clash became as likely as a coin flip. Here is what that means for real systems.

Published 2026-10-08 by Toolcloak. 3 minute read. Original: https://toolcloak.pages.dev/articles/can-two-uuids-ever-be-the-same-the-odds-in-pictures
Related free tool: [UUID Generator](https://toolcloak.pages.dev/tools/uuid-generator)

Developers use UUIDs because they can create an ID on any machine without asking a central database whether it is taken. That only works if two machines never pick the same one. Is that a promise, a hope or a statistic?

> **In short**
> - A version 4 UUID has **122 random bits**, about 5.3 × 10³⁶ possible values.
> - Generate **1 billion** and the chance of any two matching is about **1 in 10 quintillion**.
> - You need about **2.7 quintillion** UUIDs for a 50 percent chance of a collision.
> - In practice, collisions come from **bad random number generators**, not from luck.

## What is inside a UUID

A UUID is 128 bits. In version 4, six of those bits are fixed to say “I am version 4” and “I follow the standard layout”. The other 122 are random. That is where the huge number comes from.

**The 128 bits of a version 4 UUID**

| Group | Count |
| --- | --- |
| Random bits | 122 |
| Version marker | 4 |
| Variant marker | 2 |

*Each dot is one bit. Only the six at the end are not random.*

## The odds of a collision

This is the birthday problem again: collisions appear much sooner than you would think, but not nearly soon enough to matter at normal sizes. The chance that any two among n random UUIDs match is about n² divided by twice the number of possible values.

| UUIDs generated | Chance of any duplicate | In words |
| --- | --- | --- |
| 1 billion (10⁹) | 9.4 × 10⁻²⁰ | about 1 in 10 quintillion |
| 1 trillion (10¹²) | 9.4 × 10⁻¹⁴ | about 1 in 10 trillion |
| 1 quadrillion (10¹⁵) | 9.4 × 10⁻⁸ | about 1 in 10 million |
| 1 quintillion (10¹⁸) | about 9.4% | about 1 in 11 |
| 2.7 quintillion | 50% | a coin flip |

*The 6-from-49 lottery jackpot is about 1 in 14 million. A trillion UUIDs is roughly 700,000 times less likely to clash than that.*

**Getting to a coin-flip chance**

|  | Value | Note |
| --- | --- | --- |
| UUIDs needed | 2.7 × 10¹⁸ | About 2.7 quintillion |
| At 1 billion a second | 86 years | Nonstop |
| Storage just for the IDs | 43 EB | 16 bytes each, about 43 exabytes |

*You would run out of disk space long before you ran out of luck.*

Try it: [UUID Generator](https://toolcloak.pages.dev/tools/uuid-generator). Generate one UUID or a whole batch of version 4 or other versions in one click.

## Where collisions actually come from

- **A weak random generator.** Old or badly seeded generators repeat. The maths above only holds for a good cryptographic source.
- **Cloned machines.** Two virtual machines started from the same snapshot may share a seed.
- **Copy and paste.** The most common “collision” is somebody pasting the same ID twice.
- **Truncating.** Cutting a UUID to 8 characters leaves 32 bits, and collisions then appear at tens of thousands of IDs.

**A random ID of two different lengths**

| Shortened to 8 characters | Full UUID v4 |
| --- | --- |
| 32 bits | 122 random bits |
| About 77,000 IDs for a 50% clash | About 2.7 quintillion IDs for a 50% clash |
| Fine for a small demo, risky at scale | Safe at any realistic size |

*Length is what protects you. See the same idea for passwords in [password entropy](https://toolcloak.pages.dev/articles/password-entropy-bits-guesses-and-why-length-beats-symbols).*

## Which version should you use?

**The versions you will meet**

| When | What |
| --- | --- |
| v1 | Time and machine. Built from the clock and the network address. Unique, but it can reveal when and where it was made. |
| v4 | Fully random. The usual choice. No information leaks, but IDs have no order, which can slow some databases. |
| v7 | Time-ordered. A timestamp followed by random bits. Newer IDs sort after older ones, which is kinder to database indexes. Defined in RFC 9562 (2024). |

*For database primary keys, v7 is increasingly popular. For a plain unique ID, v4 is still fine.*

A hash is a different tool altogether: it summarises content rather than naming it. See [what a hash is](https://toolcloak.pages.dev/articles/what-a-hash-is-and-why-changing-one-letter-changes-everything).

## Frequently asked questions

### Can two UUIDs be the same?

In theory yes, in practice almost never. A version 4 UUID has 122 random bits, so you would need about 2.7 quintillion of them for a 50 percent chance of a single duplicate.

### What is the probability of a UUID v4 collision?

About 1 in 10 quintillion after generating a billion UUIDs, and about 1 in 10 trillion after a trillion. The chance rises with the square of how many you generate.

### Is UUID v4 or v7 better?

v4 is fully random and fine for most uses. v7 puts a timestamp first so IDs sort by time, which helps database indexes. For primary keys, v7 is often preferred.

### Is a UUID the same as a GUID?

In practice yes. GUID is Microsoft's name for the same 128-bit identifier format.

### Are UUIDs secure to use as secrets?

Not by themselves. Random v4 UUIDs are hard to guess, but they were designed to be unique, not secret. Use a dedicated cryptographic token for secrets.



---

# Base64 is not encryption, and it makes files a third bigger

> It looks scrambled, so people assume it is secret. It is not. Here is what Base64 really does to your bytes, why the output is always a third larger, and the few places it is the right tool.

Published 2026-10-07 by Toolcloak. 3 minute read. Original: https://toolcloak.pages.dev/articles/base64-is-not-encryption-and-it-makes-files-a-third-bigger
Related free tool: [Base64 Encoder](https://toolcloak.pages.dev/tools/base64-encoder)

You open a config file and see `dXNlcjpzZWNyZXQ=` next to a field called password. It looks like gibberish, so surely it is protected. It is not. Anyone can turn it back into `user:secret` in about a second, and so can you.

Base64 is one of those things everyone has seen and few have thought about. Understanding it takes five minutes and clears up two common mistakes: trusting it for secrecy, and using it where it quietly bloats your page.

## The problem it solves

Some systems can only carry plain text safely: email bodies, JSON, URLs, HTML. Raw binary data, like an image or a PDF, contains bytes that those systems would mangle. Base64 is a way to write **any bytes using only 64 safe characters**: A to Z, a to z, 0 to 9, plus and slash.

## How it works

Take three bytes. That is 24 bits. Cut those 24 bits into four groups of six bits. Each group is a number from 0 to 63, and each number is looked up in the 64-character alphabet. Three bytes in, four characters out.

| Text | As Base64 | Why |
| --- | --- | --- |
| Man | TWFu | 3 bytes, exactly 4 characters |
| Hi | SGk= | 2 bytes, so one = of padding |
| Hello | SGVsbG8= | 5 bytes, so one = of padding |

*The equals sign pads the end when the input is not a multiple of three bytes. It carries no data.*

## The cost: about 33 percent more

Three bytes become four characters, so the output is always about **four thirds** the size of the input, plus a little padding. There is no way around it, because you are spending 8 bits of storage to carry 6 bits of data per character.

**A 100 KB image, before and after Base64**

|  | Size |
| --- | --- |
| Original file | 100 KB |
| As Base64 text | 133.3 KB |

*100 KB is 102,400 bytes, which becomes 136,536 characters. A 1 MB file turns into about 1.33 MB. Compression on the server can win some of it back, but never all.*

Try it: [Base64 Encoder](https://toolcloak.pages.dev/tools/base64-encoder). Paste text to encode it, or paste Base64 to decode it, right in your browser.

## It is encoding, not encryption

Encryption needs a key. Without the key, the result cannot be read. Base64 has no key: the alphabet is public and the method is fixed, so decoding is trivial. It is a **format**, like writing a number in hexadecimal.

> **Never hide secrets in Base64:** An HTTP Basic login header is just `username:password` in Base64. It is only safe because it travels over HTTPS. If a password or token is “protected” only by Base64, treat it as plain text.

## When it is the right tool

- **Tiny inline images.** A very small icon as a data URI saves one network request. Keep it to a few kilobytes at most.
- **Email attachments.** The mail format uses Base64 to carry files through text-only channels.
- **Binary inside JSON.** When an API has to return a small file in a JSON field, Base64 is the usual answer.
- **Tokens.** The parts of a JWT use a URL-safe variant, which swaps plus and slash for minus and underscore and drops the padding.

## When to avoid it

Putting a large photo in your HTML as Base64 makes the page a third heavier, stops the browser from caching the image on its own, and delays the first paint while the text is parsed. For anything bigger than a small icon, serve a normal image file.

## One trap with non-English text

Base64 encodes bytes, not letters. Text must first be turned into bytes, normally with UTF-8. The browser function `btoa` skips that step and throws an error on characters outside basic Latin, such as Hindi or emoji. A good tool encodes to UTF-8 first, so “नमस्ते” round trips correctly.

## Frequently asked questions

### Is Base64 encryption?

No. It is a reversible encoding with no key, so anyone can decode it instantly. Do not use it to protect passwords or private data.

### Why is Base64 larger than the original?

Every 3 bytes become 4 characters, so the output is about 33 percent bigger, plus up to two padding characters at the end.

### What does the = at the end mean?

It is padding. When the input length is not a multiple of three bytes, one or two equals signs fill the last group. It holds no data.

### What is Base64url?

A variant that replaces plus with minus and slash with underscore, and usually drops the padding, so the text is safe inside URLs and file names. JWTs use it.

### Why does decoding give garbled text?

The original bytes may not be text, for example an image, or they may use a different character encoding than the one used to decode. Decode as UTF-8 for text.



---

# Is your text readable? The contrast ratio, explained

> Light grey on white looks elegant on your monitor and vanishes on a phone in sunlight. Here is the number that settles it, the thresholds that matter, and why a popular indigo fails by a hair.

Published 2026-10-07 by Toolcloak. 3 minute read. Original: https://toolcloak.pages.dev/articles/is-your-text-readable-the-contrast-ratio-explained
Related free tool: [Color Contrast Checker](https://toolcloak.pages.dev/tools/contrast-checker)

A designer picks a soft grey for the body text. It looks calm and modern on a big, bright monitor. Then someone opens the page on a phone, outside, at noon, and the paragraph simply disappears.

Whether text is readable is not a matter of taste. There is a number for it, it is easy to compute, and it is the same number that accessibility rules and audits use.

## The ratio, in plain words

Contrast ratio compares how bright two colours are. It runs from **1:1**, meaning identical colours, up to **21:1**, which is pure black on pure white. Bigger is easier to read.

The brightness is called relative luminance. Each of red, green and blue is first converted to a linear value, then combined with weights that reflect how the eye sees:

`L = 0.2126 × R + 0.7152 × G + 0.0722 × B`

Green carries most of the weight and blue almost none, which is why a pure blue looks much darker than a pure green of the same “strength”. Then the ratio is:

`ratio = (L lighter + 0.05) ÷ (L darker + 0.05)`

The 0.05 stops the maths from blowing up near black. You never need to do this by hand, but it helps to know there is no opinion hidden in it.

## The thresholds that matter

| What you are checking | Level AA (the usual target) | Level AAA (stricter) |
| --- | --- | --- |
| Normal text | 4.5 : 1 | 7 : 1 |
| Large text (about 24 px, or 19 px bold, and up) | 3 : 1 | 4.5 : 1 |
| Icons, input borders, chart lines | 3 : 1 | No separate level |

*These are the WCAG 2.x values, which most laws and audits refer to today.*

## Some real colours on white

Numbers are easier to feel with examples. Here are a few text colours on a plain white background:

**Contrast ratio against white**

|  | Ratio |
| --- | --- |
| Black  #000000 | 21 : 1 |
| Indigo  #4f46e5 | 6.29 : 1 |
| Grey  #767676 | 4.54 : 1 |
| Grey  #777777 | 4.48 : 1 |
| Indigo  #6366f1 | 4.47 : 1 |
| Grey  #959595 | 3.0 : 1 |
| Grey  #cccccc | 1.61 : 1 |

*The 4.5 line is the pass mark for normal text. #767676 is famously the lightest grey that passes on white. The very next shade, #777777, fails.*

Look at the two shades of grey. They are one step apart and you cannot tell them apart by eye, yet one passes and one does not. The same is true of the indigo `#6366f1`, a very popular button colour. As white text on that colour it lands at 4.47, just under 4.5. Moving to `#4f46e5` fixes it with a change most people would never notice.

Try it: [Color Contrast Checker](https://toolcloak.pages.dev/tools/contrast-checker). Enter a text colour and a background and see the ratio with an AA and AAA verdict.

## How to fix a failing pair

1. Darken the text (or lighten it on a dark background) a little at a time. Small steps are often enough.
2. Change the colour that does less work. If a brand colour is the problem, keep it for large or decorative things and use a darker version for text.
3. Make the text bigger or bolder when you cannot change the colour. Large text only needs 3 : 1.
4. Re-check after every change, and check light and dark themes separately.

> **A pass is not the whole story:** Contrast does not fix thin fonts, tiny sizes or busy backgrounds. Also avoid using colour alone to carry meaning, since around 8 percent of men have some form of colour vision deficiency. Add an icon, a label or a pattern.

Newer methods for judging readability are being developed, and they may change what future guidelines say. For now, the 4.5 and 3 targets are the ones you will be measured against, and meeting them is a solid, cheap first step.

## Frequently asked questions

### What is a good colour contrast ratio?

For normal text, at least 4.5 : 1 to meet WCAG level AA, and 7 : 1 for AAA. Large text needs 3 : 1 for AA.

### What counts as large text?

Roughly 24 pixels and above, or about 19 pixels and above when bold. Smaller text counts as normal text and needs the higher ratio.

### Does contrast apply to buttons and icons?

Yes. Parts of the interface that you need to see to use it, such as input borders, icons and chart lines, should have at least 3 : 1 against their surroundings.

### Is black on white always best?

It has the highest ratio, 21 : 1, but very high contrast can feel harsh for long reading, especially on dark themes. Slightly softer pairs that still pass 7 : 1 are comfortable.

### Why does the same colour pass on one background and fail on another?

Contrast is a relationship between two colours. Change either one and the ratio changes, so every text and background pair needs its own check.



---

# Your EMI is mostly interest for the first decade. Here is the proof.

> Take a 50 lakh home loan. After five years of paying ₹43,391 every month, you will have cleared less than 6 lakh of what you borrowed. The numbers behind that, why tenure is the expensive lever, and what an extra ₹5,000 a month really does.

Published 2026-10-07 by Toolcloak. 5 minute read. Original: https://toolcloak.pages.dev/articles/your-emi-is-mostly-interest-for-the-first-decade
Related free tool: [EMI Calculator](https://toolcloak.pages.dev/tools/emi-calculator)

Here is a loan most Indian families would recognise. ₹50 lakh, 8.5 percent, 20 years. The EMI comes to **₹43,391** a month. You set up the auto-debit, feel relieved, and get on with life. Then, five years later, you check the outstanding balance and it says ₹44.06 lakh.

You have paid ₹26 lakh in five years. The loan went down by ₹5.94 lakh. Nobody was cheating you. This is simply how the sum works, and once you see it laid out, a few decisions about tenure, prepayment and “flat” rates stop being mysterious.

## What an EMI actually is

EMI stands for equated monthly instalment. The word that matters is **equated**: the amount stays the same every month, but what it is made of keeps changing.

Each month the bank charges interest on whatever you still owe. That is the first slice of your EMI. Whatever is left over reduces the loan. Early on you owe a lot, so the interest slice is fat and the principal slice is thin. Every payment shrinks the balance a little, which shrinks next month's interest, which leaves a little more for principal. It tilts slowly, then faster.

| Month | Interest part | Principal part | Still owed after |
| --- | --- | --- | --- |
| 1 | ₹35,417 | ₹7,974 | ₹49,92,026 |
| 60 (year 5) | ₹31,297 | ₹12,094 | ₹44,06,359 |
| 120 (year 10) | ₹24,920 | ₹18,471 | ₹34,99,691 |
| 180 (year 15) | ₹15,181 | ₹28,211 | ₹21,14,936 |
| 240 (last) | ₹305 | ₹43,086 | ₹0 |

*₹50,00,000 at 8.5% for 20 years. In month 1, 82 paise of every rupee you pay is interest. The two slices only cross over in month 143, nearly twelve years in.*

That is the picture on the cover of this article. Twenty bars, one per year. The amber top is interest, the coloured bottom is principal. Look how the amber starts as most of the bar and thins to nothing.

## The formula, once, so it stops being magic

Every calculator uses the same expression:

`EMI = P × r × (1 + r)^n ÷ ((1 + r)^n − 1)`

P is the loan amount. **r is the monthly rate**, meaning the yearly rate divided by 12 (and by 100). n is the number of months. That is it. A common slip is plugging 8.5 where 0.00708 belongs, or using years where months belong, and getting an EMI that is nonsense.

Try it: [EMI Calculator](https://toolcloak.pages.dev/tools/emi-calculator). Enter amount, rate and tenure. It shows the EMI, total interest and total payment.

## Tenure is the expensive lever

When an EMI looks too high, the instinct is to stretch the tenure. It works, in the sense that the monthly number falls. Look at what it costs.

**Same ₹50 lakh at 8.5%, different tenures**

|  | Principal | Interest |
| --- | --- | --- |
| 15 years, EMI ₹49,237 | 50 | 38.63 |
| 20 years, EMI ₹43,391 | 50 | 54.14 |
| 25 years, EMI ₹40,261 | 50 | 70.78 |
| 30 years, EMI ₹38,446 | 50 | 88.4 |

*Going from 20 to 30 years lowers the EMI by ₹4,945 a month. It adds ₹34.3 lakh of interest.*

Read that again. Ten extra years buys you five thousand rupees of monthly breathing room and costs a little over 34 lakh. Sometimes that is exactly the right trade. If the alternative is missing payments, take the longer tenure and plan to prepay later. But it should be a decision, not a default.

## What an extra ₹5,000 a month does

Back to the 20-year loan. Suppose you add ₹5,000 to every EMI from day one, paying ₹48,391 instead of ₹43,391.

- The loan closes in **187 months** instead of 240. That is 15 years and 7 months, so 4 years and 5 months early.
- Total paid drops from about ₹1.041 crore to about ₹90.5 lakh. You keep roughly **₹13.6 lakh**.
- You never feel it as a lump sum. It is a small extra transfer every month.

The reason it works so well early is the same reason the loan is front-loaded with interest. Every rupee you prepay in year one removes interest for the remaining 19 years. A rupee prepaid in year 18 saves almost nothing. If you are going to prepay, **sooner beats bigger**.

> **Check your loan terms first:** For floating-rate home loans to individuals, RBI rules generally bar lenders from charging a prepayment penalty. Fixed-rate loans, business loans and some other products can differ, so read the line in your own agreement before you plan around it.

## The “flat rate” trap

Car loans, two-wheeler loans and some personal loans are sometimes quoted at a flat rate. “Only 7 percent flat!” sounds cheaper than a home loan. It is not.

A flat rate charges interest on the **original** amount for the whole tenure, even though you are repaying it every month. Take ₹1,00,000 for 3 years at 7 percent flat. Interest is ₹21,000, so you repay ₹1,21,000 in 36 instalments of about ₹3,361. Now work out what yearly rate on a normal reducing balance would produce that same EMI. It is about **12.8 percent**.

Rule of thumb: on a loan of a few years, the real rate is often close to double the flat rate. Whenever you see “flat”, ask for the effective rate, or just use the calculator with the EMI they quote and work backwards.

## How much EMI is too much?

There is no universal answer, and we are not financial advisers. What can be said is how lenders think about it. Many look at the share of your monthly take-home that goes to all loan EMIs together, and commonly want it under roughly 40 to 50 percent. That is the lender's comfort, not yours.

A calmer way to test your own limit is to ask what happens in a bad month. If an EMI plus rent plus school fees leaves nothing for a medical bill or a few months without salary, the loan is too big for you even if the bank approves it.

## Before you sign anything

1. Ask for the **amortisation schedule**: the month-by-month split of interest and principal. A good lender hands it over without fuss.
2. Add up **processing fees and insurance**. They are not in the EMI but they are part of what the loan costs.
3. Compare loans by **total interest**, not by EMI alone. A lower EMI on a longer tenure is often the more expensive loan.
4. Check whether the rate is **floating** and what it is linked to. Your EMI or your tenure can change later when the rate moves.
5. Decide your **prepayment plan** now, even a small one, and put it in your calendar.

> **This is an explanation, not advice:** The figures here come from the standard reducing-balance formula and the example loan above. Your actual loan will depend on your lender's terms, rate changes, fees and rounding.

## Frequently asked questions

### How is EMI calculated?

With the reducing-balance formula `EMI = P × r × (1 + r)^n ÷ ((1 + r)^n − 1)`, where P is the loan amount, r is the monthly interest rate (yearly rate ÷ 12 ÷ 100) and n is the number of months. The EMI stays fixed while the interest and principal parts inside it change every month.

### Why is so much of my early EMI interest?

Interest is charged on the outstanding balance, and in the early years the balance is at its largest. As you repay, the balance falls, interest falls with it, and a bigger share of each EMI goes to principal.

### Is a longer tenure better?

It lowers your monthly payment but raises total interest, sometimes by a lot. On a ₹50 lakh loan at 8.5 percent, 30 years instead of 20 adds about ₹34 lakh of interest. Longer tenure makes sense only when you need the lower EMI to stay safe, ideally with a plan to prepay.

### What is the difference between flat and reducing rate?

A flat rate charges interest on the original loan amount for the entire tenure. A reducing rate charges interest only on what you still owe. A flat rate of 7 percent over 3 years works out to roughly 12.8 percent on a reducing balance.

### Does paying extra every month really help?

Yes, and the earlier the better. An extra ₹5,000 a month on a ₹50 lakh, 8.5 percent, 20-year loan closes it about 4 years 5 months early and saves roughly ₹13.6 lakh in interest.



---

# How to add or remove GST without getting it wrong

> A bill says 11,800 including 18 percent GST. How much of that is tax? Almost everyone subtracts 18 percent and gets it wrong. Here is the one-line fix, the CGST, SGST and IGST split, and the invoice mistakes that cost small businesses money.

Published 2026-10-07 by Toolcloak. 3 minute read. Original: https://toolcloak.pages.dev/articles/how-to-add-or-remove-gst-without-getting-it-wrong
Related free tool: [GST Calculator](https://toolcloak.pages.dev/tools/gst-calculator)

You are handed a bill for ₹11,800 and told it includes 18 percent GST. Your accountant, or your own head, says: take 18 percent off. That gives ₹2,124 of tax and a base price of ₹9,676. It feels right. It is wrong, and the mistake repeats on thousands of invoices every day.

The reason is small but stubborn. GST is charged on the **price before tax**, not on the final amount. So the 18 percent has to be taken from a number you do not know yet, and subtracting it from the total uses the wrong base.

## Adding GST is easy

When you start from the price without tax, one multiplication does everything:

`Total = Price × (1 + rate ÷ 100)`

A ₹10,000 service at 18 percent becomes ₹10,000 × 1.18 = **₹11,800**. The tax is the difference, ₹1,800. Nothing surprising here, and nobody gets this one wrong.

## Removing GST is where people slip

Now go backwards. You know ₹11,800 and the 18 percent rate. The base price is the number that, multiplied by 1.18, gives 11,800, so you **divide** by 1.18:

`Price before tax = Total ÷ (1 + rate ÷ 100)`

11,800 ÷ 1.18 = ₹10,000. The GST inside the bill is 11,800 − 10,000 = ₹1,800. If you only want the tax part directly, use a shortcut that works for any rate:

`GST inside = Total × rate ÷ (100 + rate)`

That is 11,800 × 18 ÷ 118 = ₹1,800 again. Notice the 118 on the bottom. It is 100 plus the rate, and it is the whole secret.

**Same ₹11,800 bill, two ways of finding the GST**

|  | GST amount |
| --- | --- |
| Divide by 1.18 (right) | ₹1,800 |
| Take 18% off (wrong) | ₹2,124 |

*The wrong method overstates the tax by ₹324 on a single bill. If you claim or pay on that figure, the error compounds across every invoice in the month.*

Try it: [GST Calculator](https://toolcloak.pages.dev/tools/gst-calculator). Switch between adding and removing GST and see the base, the tax and the CGST, SGST or IGST split together.

## Quick check with two rates

The rates below are examples to show the arithmetic. The rate for your product or service is whatever the current GST schedule says, so confirm it before you invoice.

| You know | Rate | Price before tax | GST | Total |
| --- | --- | --- | --- | --- |
| ₹10,000 before tax | 5% | ₹10,000 | ₹500 | ₹10,500 |
| ₹10,000 before tax | 18% | ₹10,000 | ₹1,800 | ₹11,800 |
| ₹10,500 final | 5% | ₹10,000 | ₹500 | ₹10,500 |
| ₹11,800 final | 18% | ₹10,000 | ₹1,800 | ₹11,800 |

> **A habit that prevents most errors:** Say the question out loud: does this number already include tax? If yes, you divide. If no, you multiply. Most wrong invoices come from skipping that one question.

## CGST, SGST and IGST: who gets the tax

The rate you see is the total. Who receives it depends on where the sale happens. Inside one state, the 18 percent is split evenly between the Centre and the state. Between two states, the whole thing goes under a single head.

| Sale | Tax heads | On ₹10,000 at 18% |
| --- | --- | --- |
| Within one state | CGST 9% + SGST 9% | ₹900 + ₹900 |
| Between two states | IGST 18% | ₹1,800 |

*The total is the same either way. Only the labels on the invoice change.*

This matters because the invoice must show the right heads. A buyer in another state who receives a CGST and SGST invoice, or the other way round, can run into trouble when claiming credit.

## Discounts change the base

If you give a discount that is shown on the invoice, tax is generally worked out on the **discounted** value. A ₹10,000 item with a ₹1,000 discount has a taxable value of ₹9,000, so 18 percent GST is ₹1,620 and the customer pays ₹10,620, not ₹11,800 minus something.

- Work out the discount first, then the tax on what is left.
- Do not add a discount after the tax line, or the invoice will not add up.
- Rounding: software may differ by a paisa or two on long invoices. That is normal, but be consistent.

> **This is arithmetic, not tax advice:** Rules about credit, special cases and the rate for a given product change. For anything that affects a filing, check the current GST schedule or ask a chartered accountant.

## A quick routine for any invoice

1. Find out whether the amount you hold is before or after tax.
2. Confirm the rate for that item from the current schedule.
3. Multiply by 1 plus the rate to add tax, or divide by the same number to remove it.
4. Choose the heads: CGST and SGST inside a state, IGST across states.
5. Round once, at the end, and keep the unrounded figures if you will reuse them.

## Frequently asked questions

### How do I remove 18% GST from a price?

Divide the final price by 1.18. For ₹11,800 that gives ₹10,000 before tax, and the GST is the difference, ₹1,800. Subtracting 18 percent of 11,800 gives the wrong answer.

### What is the formula for the GST amount inside a price?

GST inside = total × rate ÷ (100 + rate). For 18 percent that is total × 18 ÷ 118, which is about 15.25 percent of the final price, not 18 percent.

### What is the difference between CGST, SGST and IGST?

CGST and SGST are charged together on a sale within one state, each usually half of the rate. IGST is charged as one tax on a sale between states. The total rate is the same.

### Is the GST rate the same for everything?

No. Different goods and services fall under different rates, and the schedule is revised from time to time. Always confirm the current rate for your item before invoicing.

### Can I trust a calculator for invoices?

A calculator gives the correct arithmetic for the rate you enter. Whether that rate and treatment apply to your sale is a separate question that depends on current rules.



---

# Why your 4 MB photo is really a 400 KB photo

> Every job portal, KYC form and email has a size limit, and every phone camera ignores it. Here is what is inside a JPEG, what you can throw away safely, and how to hit 100 KB without the picture turning to mush.

Published 2026-10-07 by Toolcloak. 6 minute read. Original: https://toolcloak.pages.dev/articles/why-your-4mb-photo-is-really-a-400kb-photo
Related free tool: [Image Compressor](https://toolcloak.pages.dev/tools/image-compressor)

You are filling in a form that needs a passport photo under 100 KB. Your phone just took one that is 3.8 MB. The form does not care that it is a lovely photo. It says “file too large” in red and that is the end of the conversation.

Most of us fix this by trial and error. Screenshot it, WhatsApp it to ourselves, try some random site that wants an email address. It works about half the time and you never find out why. So let's find out why, because once you see what is inside an image file, the size problem mostly solves itself.

## Where 4 MB comes from

A modern phone takes a picture that is about 4000 pixels wide and 3000 tall. That is twelve million tiny dots. Each dot needs three numbers (how much red, green and blue), one byte each. Do the sum: 12 million times 3 is **36 MB** for the raw picture.

Your phone saves it as 4 MB. So the camera has already squeezed the picture to about a ninth of its natural size before you ever touched it. Compression is not some strange extra step. It is how every photo you own got small enough to exist.

The only question is how much more you can squeeze before someone notices.

## What JPEG throws away

JPEG is a lossy format, which is a polite way of saying it deletes things permanently. The clever part is what it chooses to delete: the stuff your eyes are bad at noticing.

- **Colour detail.** Your eyes are far more sensitive to brightness than to colour. JPEG keeps the brightness sharp and stores the colour at lower resolution. You do not see the difference.
- **Tiny textures.** It cuts the picture into 8 by 8 pixel blocks and drops the finest patterns inside each block, the grain in a wall or the fuzz on a jumper.
- **Repetition.** Whatever is left gets packed tightly, the way a zip file packs text.

The “quality” slider in any compressor is simply how hard it pushes on the second step. Low quality means big blocks of detail get wiped. At the extreme you see the ugly squares that make old WhatsApp forwards look like they were drawn in Minecraft.

## Pixels first, quality second

Here is the thing most people miss. File size follows the number of pixels more than it follows the quality slider. A website that shows your photo 800 pixels wide gains nothing from a 4000 pixel original. It is carrying five times the width and twenty-five times the dots, for no visible benefit.

**How many pixels you are carrying**

|  | Millions of pixels |
| --- | --- |
| Phone original, 4000 x 3000 | 12 M |
| 2000 x 1500 | 3 M |
| 1280 x 960 | 1.2 M |
| 1024 x 768 | 0.8 M |

*Halve the width and height and you keep only a quarter of the pixels. The file shrinks by roughly the same proportion, before quality is even touched.*

So the order that works is: resize to the size you actually need (our [Image Resizer](https://toolcloak.pages.dev/tools/image-resizer) does it in a few taps), then compress. If a form wants 100 KB, a 600 by 800 photo at 80 percent quality gets there easily. The same photo at 4000 by 3000 would need to be crushed to 15 percent quality to reach 100 KB, and it would look awful.

Try it: [Image Compressor](https://toolcloak.pages.dev/tools/image-compressor). Drop in one photo or fifty. Pick a quality, see the before and after size, download.

## The 75 percent rule

Quality is not linear. Going from 100 to 85 removes a huge amount of data and you cannot see it. Going from 50 to 35 removes much less data and you absolutely can. So the sweet spot sits in the middle: **70 to 85 percent**. Start at 75. Only go lower if you are still over the limit.

> **Aim under the limit, not on it:** If a portal says 200 KB, aim for 170 or 180. Some forms count in kilobytes of 1000, some in 1024, and nobody tells you which. A little room saves you a second round.

## JPG, PNG or WEBP?

Three formats, three personalities. Pick by what is in the picture, not by habit.

| Format | Good at | Watch out |
| --- | --- | --- |
| **JPG** | Photos of people, places, food. Anything with smooth gradients. | Text and sharp lines get fuzzy edges at low quality. No transparency. |
| **PNG** | Screenshots, logos, charts, anything with crisp text or a transparent background. | Photos saved as PNG are huge. It does not throw detail away, so there is less to squeeze. |
| **WEBP** | Websites. Handles both photos and graphics and supports transparency. | Some older apps and printers still do not open it. Google's own comparisons claim roughly a quarter to a third smaller than JPG at similar quality. |

A rule that is hard to get wrong: **photo in, JPG or WEBP out. Screenshot in, PNG out.** The most common mistake is saving a screenshot as JPG and wondering why the text looks dirty.

## Never compress the same file twice

Each time you open a JPG, edit it and save it again, the format re-applies its deleting step on top of what was already deleted. Photographers call this generation loss. Do it five times and the damage shows, especially around edges and text.

The habit that fixes it is simple. Keep the original in a folder you never touch. Always compress from that, once, into a new file. If the result is wrong, go back to the original and try a different setting.

## A word about WhatsApp

WhatsApp compresses any picture you send as a normal photo. That is why the one your cousin forwards looks worse every time it travels. If you send it as a **document** instead, it arrives untouched. This is useful when you are sending an ID scan to someone who needs to read small print.

It is also why a screenshot of a screenshot of a screenshot looks like a ransom note by the third generation.

## Where should a compressor run?

This one matters more than people think. Think about what you compress: passport photos, ID cards, signed forms, medical reports. The kind of images you would never post publicly.

Plenty of online compressors upload your file to their server, process it there and send it back. Some delete it afterwards. Some say they do. You cannot check either way. Toolcloak's compressor does the work inside your browser, so the picture never leaves your device. You do not need to take our word for it: open your browser's Network tab, compress something, and watch that nothing gets sent.

## Quick recipes

| You need | Resize to | Format, quality |
| --- | --- | --- |
| Form upload under 100 KB | About 600 x 800 | JPG, 70 to 80 |
| Email attachment, many photos | About 1600 wide | JPG, 75 |
| Website hero image | 1600 to 2000 wide | WEBP, 75 to 80 |
| Article or social post | 1200 wide | JPG or WEBP, 80 |
| Logo or screenshot | Keep original size | PNG, then check the file size |
| Printing at home | Do not shrink | Keep the original |

*Starting points, not laws. Look at the result before you upload it.*

And if the file is a PDF rather than an image, the same logic applies with a different tool. [PDF Compressor](https://toolcloak.pages.dev/tools/pdf-compressor) works in the browser the same way, and you can chain the two in [Tool Chains](https://toolcloak.pages.dev/chains) if you have a whole pile to get through.

## Frequently asked questions

### Will compressing make my photo blurry?

Not at 70 to 85 percent quality. The deleted detail is the kind your eyes cannot pick out at normal viewing size. Blur appears when you push quality below about 50 or when you shrink the picture far below the size it will be shown.

### How do I reduce a photo to exactly 100 KB?

Resize it first to roughly 600 by 800 pixels, then compress at 70 to 80 percent. If it is still over, lower the quality in small steps of 5 and check the size each time. Aim for a few KB under the limit.

### Is JPG or PNG better for a photo?

JPG, almost always. PNG keeps every pixel exactly, which makes photos several times larger for no visible gain. Use PNG for screenshots, logos and images with transparent backgrounds.

### Can I get the original quality back after compressing?

No. Lossy compression deletes data permanently, which is why you should always keep your original and compress a copy.

### Is it safe to compress ID photos online?

It depends on the tool. If the site uploads your file, a copy exists on someone else's server. A browser-based compressor processes it on your own device and sends nothing anywhere. You can verify this in your browser's Network tab.



---

# Why your JSON will not parse, and how to find the bad comma

> “Unexpected token at position 214” tells you almost nothing. Here is what a parser is really doing, the handful of mistakes behind nearly every error, and why the real problem is often just before the place it points to.

Published 2026-10-07 by Toolcloak. 3 minute read. Original: https://toolcloak.pages.dev/articles/why-your-json-wont-parse-and-how-to-find-the-bad-comma
Related free tool: [JSON Formatter](https://toolcloak.pages.dev/tools/json-formatter)

It is late, the deploy is blocked, and the log says one thing: unexpected token } in JSON at position 214. You open the file. It is 300 lines long and looks perfectly fine. Welcome to the most common small frustration in software.

The good news is that JSON is a tiny, strict language. There are only a few ways to break it, and once you know them you can spot most errors by eye in seconds.

## What JSON allows, in one paragraph

A JSON value is one of six things: a **string** in double quotes, a **number**, `true` or `false`, `null`, an **array** in square brackets, or an **object** in curly braces whose keys are strings in double quotes. Items are separated by commas. That is the entire language. No comments, no extras, no shortcuts.

## Looks fine in JavaScript, breaks in JSON

Most errors come from people writing JavaScript-style data and assuming it is valid. It is close, and close is not enough.

| You wrote | Why it fails | Write this instead |
| --- | --- | --- |
| {'name': 'Asha'} | Single quotes are not allowed | {"name": "Asha"} |
| [1, 2, 3,] | Trailing comma after the last item | [1, 2, 3] |
| {name: "Asha"} | Keys must be in double quotes | {"name": "Asha"} |
| // note | JSON has no comments | Remove it, or add a "note" field |
| {"a": undefined} | undefined is not a JSON value | Use null, or leave the key out |
| {"id": 007} | Numbers cannot have leading zeros | 7, or the string "007" |
| {"x": NaN} | NaN and Infinity do not exist | null, or a string |

*If you copy an object out of code or a log and paste it somewhere that wants JSON, check this list first.*

## Why the error points to the wrong place

A parser reads characters from left to right and stops at the first one that cannot be right. Crucially, it reports where it **gave up**, not where you made the mistake. In most parsers, position counts characters from zero at the start of the text.

Say you forget a comma between two fields. The parser happily reads the first value, then meets the next key, which should have been preceded by a comma. It complains about that key. The real mistake is the empty space right before it. The same happens with a missing closing quote or bracket: the error appears further down, where the structure finally stops making sense.

**What a parser does with your text**

1. **Read.** Go through the characters one by one from the start.
2. **Split into tokens.** Recognise strings, numbers, brackets, commas and colons.
3. **Check the grammar.** Each token must be allowed at that spot. A comma before a closing bracket is not.
4. **Build or stop.** If everything fits you get an object. At the first misfit it stops and reports it.

> **Rule of thumb:** When the error points at a token, look at the character just before it first. A missing comma, an extra comma or an unclosed quote is usually sitting there.

Try it: [JSON Formatter](https://toolcloak.pages.dev/tools/json-formatter). Paste your JSON to validate, indent and minify it, with the problem line shown.

## Formatting and minifying

A formatter only changes whitespace. It adds line breaks and indentation so a human can read the structure, and a minifier removes them again so the data is as small as possible. The data itself does not change.

Here is one small object, with an indent of two spaces, four spaces, and none at all:

**The same object, three layouts (bytes)**

|  | Bytes |
| --- | --- |
| Minified | 82 B |
| Indented 2 spaces | 121 B |
| Indented 4 spaces | 145 B |

*Whitespace added roughly half again to this tiny object. On large API responses, sending minified JSON is a cheap way to save bandwidth, while indented JSON is better for files people read.*

## Three traps that are valid but still bite

- **Big numbers.** JavaScript stores numbers as 64-bit floats, so integers above 9,007,199,254,740,991 can silently change. Send long IDs as strings.
- **Dates.** JSON has no date type. Use a text format like `2026-10-07T09:30:00Z` and agree on it with whoever reads it.
- **Duplicate keys.** The format says keys should be unique, but parsers disagree on what to do. Many keep the last one. Do not rely on it.

> **Be careful what you paste:** JSON from real systems often holds tokens, emails or customer records. Use a formatter that runs in your browser and does not upload the text, and avoid pasting secrets into sites you do not trust.

## Frequently asked questions

### What does “unexpected token in JSON” mean?

The parser reached a character that is not allowed at that point. The real mistake is often just before it, such as a missing or extra comma, a single quote, or an unclosed bracket.

### Can JSON have comments?

No. Standard JSON does not allow comments. Remove them, or store the note as a normal field. Some tools accept a relaxed variant, but you cannot count on it.

### Are single quotes allowed in JSON?

No. Strings and keys must use double quotes. Single quotes are valid in JavaScript but not in JSON.

### What is the difference between formatting and minifying JSON?

Formatting adds line breaks and indentation to make JSON readable. Minifying removes all unneeded whitespace to make it smaller. Both keep the data exactly the same.

### Why does my large ID change after parsing?

Numbers above 9,007,199,254,740,991 cannot be stored exactly as a JavaScript number. Store such IDs as strings to keep every digit.



---

# Merging PDFs without posting your documents to a stranger's server

> Resume, ID proof, marksheets, address proof: the bundle every application wants, and exactly the documents you should not be feeding to a random website. How merging works, how to do it on your own device, and how to keep the result small enough to upload.

Published 2026-10-07 by Toolcloak. 5 minute read. Original: https://toolcloak.pages.dev/articles/merging-pdfs-without-posting-your-documents-to-a-stranger
Related free tool: [Merge PDF](https://toolcloak.pages.dev/tools/merge-pdf)

A recruiter's email says: “Please send your resume, ID proof, degree certificate and last three payslips as a single PDF.” You have four files, two of them scans, one saved in a folder called New Folder (2). It is 11 at night and the deadline is morning.

So you search “merge pdf online”, click the first result, and upload your Aadhaar, your salary slips and your signature to a company you have never heard of. Most people do exactly this, and it is worth pausing on it for a moment, because there is no need.

## What happens when you upload

A typical online merger works like this: your files travel to its server, the server joins them, you download the result. For a short while, and often longer, a copy of your documents sits on a machine you do not control. A good service deletes it after an hour. A careless one keeps it, logs it, or gets breached.

You cannot check which kind you picked. The privacy policy tells you what the company says it does, not what its servers actually do. For a funny meme that does not matter. For identity documents, bank statements and payslips, it matters a lot.

## How a PDF is actually built

Here is the surprising bit: merging PDFs is not heavy work. A PDF is basically a container with a list of pages, plus the fonts and images those pages use. To merge two files, a program copies the page objects from one into the other and fixes up the table of contents that says where everything is. There is no rendering, no OCR, no magic. A modern phone can do it before you have finished blinking.

Which means the upload is unnecessary. The browser can do the whole job locally. When you open [Merge PDF](https://toolcloak.pages.dev/tools/merge-pdf), the page loads a small open-source PDF library (pdf-lib) and uses it right there on your device. Your files are read from your storage into the browser's memory, joined, and offered back as a download. They are never sent anywhere.

**What happens inside the tab**

1. **Add your files.** They are read from your device into the browser. Nothing leaves it.
2. **Set the order.** Move files up or down with the arrows until the sequence is right.
3. **Merge.** Pages are copied from each file into one new document, in your order.
4. **Download.** The finished PDF is created on your device and saved like any other download.

> **Verify it yourself:** Open your browser's developer tools, go to the Network tab, and merge two files. You will see the library load, and you will see no file being uploaded. If any site claims privacy, this takes thirty seconds to check.

Try it: [Merge PDF](https://toolcloak.pages.dev/tools/merge-pdf). Add your PDFs, arrange them with the arrows, download a single file.

## Order is the whole job

The merge itself is easy. What people get wrong is the order, and discover it only after the file is sent. A few rules that avoid embarrassment:

1. Put the document the reader cares about **first**. For a job application that is your resume, not your Class 10 certificate.
2. Then group the rest the way the reader will check them: identity, then education, then work.
3. Name files with numbers (`01-resume.pdf`, `02-id.pdf`) so the order survives being dropped into the tool in a different sequence.
4. After merging, scroll through the result. Count the pages: the total should equal the sum of your files. A missing page is far easier to find now than after the recruiter replies.

If one page is upside down, which happens constantly with phone scans, fix it with [Rotate PDF](https://toolcloak.pages.dev/tools/rotate-pdf) before merging. If a document has pages you do not want to send, [Delete PDF Pages](https://toolcloak.pages.dev/tools/delete-pdf-pages) or [PDF Page Organizer](https://toolcloak.pages.dev/tools/pdf-page-organizer) will tidy it first.

## Photos and scans

Documents on your phone are often photos, not PDFs. [Image to PDF](https://toolcloak.pages.dev/tools/image-to-pdf-converter) turns them into pages first, and then they can be merged with everything else. Keep one thing in mind: a phone photo of a page can be 3 or 4 MB on its own. Ten of them and you have a 35 MB PDF that no portal will take.

## Why the merged file is so big

Merging does not make files smaller. The result is roughly the sum of the originals, sometimes a touch less when fonts are shared, never dramatically so. If your four files were 1.8, 2.4, 3.1 and 0.9 MB, expect a bit over 8 MB.

Most application portals cap uploads at a few MB. So the real workflow is **merge, then compress**. Run the merged file through [PDF Compressor](https://toolcloak.pages.dev/tools/pdf-compressor), which shrinks the images inside the pages, and check the size again. Aim for a few hundred KB under the limit.

> **Do both in one go:** Toolcloak has a ready-made chain for exactly this: [Merge PDFs, then compress](https://toolcloak.pages.dev/chains/templates/pdf-merge-then-compress). You choose the files once and it runs the two tools in order.

## Things that can behave differently

- **Password-protected PDFs.** A locked file generally cannot be merged until it has been unlocked. If you have the password, remove the protection first, with whichever tool you have.
- **Forms and links.** Interactive form fields, bookmarks and internal links may not carry over exactly as they were. If a form must stay fillable, fill it first and then merge.
- **Page sizes.** If you combine A4 pages with a letter-size page or a tiny scan, the page sizes stay as they are. Nothing is stretched. That is usually what you want, but it can look odd when printed.
- **Very large files.** Everything happens in your device's memory, so a huge pile of big PDFs can strain an old phone. Merge in two batches if the browser struggles.

## A small checklist before you send

- Page count matches the total of the originals.
- Every page is upright and readable, including the stamps and signatures.
- The file is under the limit with a little to spare.
- The file name is something a human would want to find again, such as `Rahul-Sharma-Application.pdf`, not `merged (3) final.pdf`.
- The merged copy lives in a folder you can find next time. Applications come in waves.

## Frequently asked questions

### How do I merge PDF files without uploading them?

Use a browser-based merger that does the work on your own device. In Toolcloak's Merge PDF the files are read into your browser, combined there and downloaded straight back, with nothing sent to a server.

### Does merging reduce quality or file size?

It does not change quality, because pages are copied as they are. It also does not reduce size, so the merged file is about the sum of the originals. Compress the result afterwards if it is over a portal's limit.

### In what order will the pages appear?

In the order you set in the list, with each file's pages kept in their own order. Move files up or down before merging, and check the page count and the order afterwards.

### Why can I not merge a protected PDF?

A file that is locked with a password cannot be read by the merger. Unlock it first with the password you have, then merge.

### How many PDFs can I merge?

There is no fixed limit in the tool, but everything runs in your device's memory, so very large files add up. If the page slows down, merge in smaller groups and then merge the results.



---

# Your password is not weak because it is short. It is weak because it is yours.

> “Summer@2024” has a capital, a number and a symbol, and it still falls in seconds. The maths of guessing, why length beats cleverness, and the one habit that protects you more than any password rule.

Published 2026-10-07 by Toolcloak. 5 minute read. Original: https://toolcloak.pages.dev/articles/your-password-is-not-weak-because-it-is-short
Related free tool: [Password Generator](https://toolcloak.pages.dev/tools/password-generator)

Think of the last password you were proud of. It probably had a name or a place in it, a year, a capital letter at the start and an exclamation mark at the end. The site gave you a green bar that said “strong”. You felt safe.

The trouble is that password cracking programs were written by people who have also been proud of that exact password. They try “Priya@1998” and “Mumbai#2024” early, because humans are wonderfully predictable. So the first thing to understand is that password strength is not about how it looks. It is about how many other things it could have been.

## How guessing really works

Nobody sits at a login page typing guesses. The realistic attack starts after a company is breached and its list of password fingerprints (hashes) leaks. The attacker then guesses offline, on their own machines, with no lockouts and no alarms. Against a fast, badly protected hash, modern graphics hardware can try **billions of guesses per second**. We will use 10 billion a second below, a deliberately generous figure for the attacker.

Two kinds of guessing happen:

- **Dictionary attacks** try real words, names, leaked passwords and common patterns first, together with the usual tricks: capital at the front, a number at the end, “a” turned into “@”.
- **Brute force** tries every combination, shortest first. It only matters for passwords that are truly random, which is the only kind worth having.

A password that a human invented lives almost entirely in the first group. A password that a computer invented lives in the second. The gap between them is enormous.

## Counting the possibilities

A random password drawn from lowercase letters only has 26 options per position. Eight characters gives 26 to the power 8, about **208 billion** combinations. That sounds big until you divide by 10 billion guesses a second. On average the attacker finds it in roughly ten seconds.

Mix in capitals, digits and symbols and you have about 94 options per position. Eight characters is now 6.1 quadrillion combinations, which holds for around three and a half days on average. Better. Still a weekend project.

Now keep the same 94 symbols and stretch it. Twelve characters is 94 to the 12th power. At the same guessing speed, the average time to crack is about **750,000 years**. Same alphabet, four more characters. Length does what cleverness cannot.

**Strength in bits (each extra bit doubles the work)**

|  | Bits of randomness |
| --- | --- |
| 8 lowercase letters, random | 37.6 bits |
| 8 letters, digits, symbols | 52.4 bits |
| 4 random dictionary words | 51.7 bits |
| 6 random dictionary words | 77.5 bits |
| 12 random characters | 78.7 bits |
| 16 random characters | 104.9 bits |

*Bits measure how many guesses a random password takes: n bits means 2 to the power n possibilities. The word-based rows assume words picked at random from a list of 7,776, as in the diceware method. These are random passwords. Anything you invented yourself is far weaker than its length suggests.*

Notice something odd in that chart. A phrase of **four random words** matches an eight-character jumble of symbols, and six random words matches twelve random characters. The reason is that words come from a big list, and a random choice from 7,776 words is worth almost 13 bits all by itself.

This is why “correct horse battery staple” became a famous example. But note the word **random**. A line of a song or a quote from a film is not random, and it is in the attacker's word list before you finish typing it.

Try it: [Password Generator](https://toolcloak.pages.dev/tools/password-generator). Length from 8 to 128, four character types, one-click copy. Built on your browser's secure random generator.

## The real danger is reuse

Here is the part that matters more than the maths. Suppose you have a perfect twenty character password. You use it on your email, your shopping app and a small hobby forum. The forum gets breached, because small sites often have weak security, and the attacker now has your email and your password.

They do not crack anything. They try that pair on every popular service in the world, automatically, in minutes. This is called **credential stuffing**, and it is how most accounts are actually taken over. Your strong password was useless the moment you used it twice.

> **The rule that beats all other rules:** One site, one password. If you only fix one thing after reading this, make it your email account, because every “forgot password” link in the world lands there. Give that one a password that exists nowhere else.

## So how do I remember forty of them?

You do not. That is the point of a password manager: it remembers, you do not, and so the passwords can be fully random and different everywhere. You memorise exactly **one** strong passphrase, the one that opens the manager, and nothing else.

Your browser's built-in manager, or a dedicated one, will fill them in for you. Either is vastly better than a notebook of variations on the same name, which is what most of us have in our heads right now.

For that single master passphrase, use the word method. Pick four to six words completely at random (dice and a word list are best, if you can be bothered) and string them together. Do not choose words that mean something to you.

## Where the random comes from

Not every “random” is equal. Programmers have a casual function, `Math.random()`, that is fine for shuffling a playlist and **not** fine for secrets, because its output can be predicted. A proper generator uses the browser's cryptographic source, `crypto.getRandomValues()`, which draws on unpredictable data from your operating system.

Toolcloak's generator uses that secure source, runs on your device, and does not save or send what it makes. This is also how you should judge any online generator: if it needs a server to give you a password, the password has already been somewhere you did not choose.

## A thirty minute fix for this weekend

1. Open your **email account** settings and change the password to a long random one. Turn on two-step verification at the same time.
2. Do the same for your **bank, UPI and shopping apps**. These are the ones where a stranger can spend your money.
3. Install or turn on a **password manager** and save what you just created.
4. Work through the rest over the next few weeks, one account when you next log in. Use a length of 16 or more and let it fill all four character types.
5. Where you can, add an **authenticator app** or a passkey. A stolen password then stops being enough.

None of this is exciting. That is exactly why it works: boring habits repeated are what keeps accounts safe, far more than a clever password ever did.

> **Check what you already have:** Our [Password Strength Checker](https://toolcloak.pages.dev/tools/password-strength-checker) is a quick way to see how weak a human-made password really is. Do not type a password you currently use into any website that is not yours. This one runs only in your browser, but the habit is worth keeping.

## Frequently asked questions

### How long should a password be?

For anything important, 16 characters or more when generated randomly. A random 12 character password is already very strong, and the extra four give you a wide margin. For a master password you must remember, use four to six random words.

### Are symbols necessary?

They help a little, but length helps far more. Going from 8 to 12 random characters makes a password millions of times harder to guess, while adding symbols to an 8 character one multiplies the difficulty by a few thousand. If a site demands symbols, a password generator will include them for you.

### Is it safe to use an online password generator?

Only if it generates the password on your own device. A generator that runs in your browser using the secure random API never needs to send anything out. One that works through a server has seen your password before you did.

### What is a passphrase?

A password made of several random words, such as four to six picked from a large list. It is easy to remember and, when the words are truly random, very hard to guess.

### Should I change my passwords regularly?

Frequent forced changes mostly push people towards predictable patterns. Change a password when there is a reason: a breach, a shared device, or suspicious activity. Otherwise a long, unique password plus two-step verification is the better investment.



---

# What is actually inside the black squares of a QR code?

> You can scratch a QR code, cover a corner with a thumb, even stick a logo on it, and it still scans. That is not luck. It is a 1990s car factory trick and some very clever maths.

Published 2026-10-07 by Toolcloak. 6 minute read. Original: https://toolcloak.pages.dev/articles/whats-inside-the-black-squares-of-a-qr-code
Related free tool: [QR Code Generator](https://toolcloak.pages.dev/tools/qr-generator)

Look at a QR code on a restaurant table. It is torn at one corner, there is a coffee ring across it, and your phone reads it in half a second anyway. Try that with a barcode on a packet of biscuits and the till will beep angrily.

The difference is that a QR code was designed to be **wrong and still work**. Once you know how, you also know how to make ones that never fail at the worst moment, which is usually the moment after the printer has finished a thousand flyers.

## Born in a car factory

QR stands for Quick Response. It was created in 1994 at Denso Wave, a company in the Toyota group, to track car parts as they moved through factories. Ordinary barcodes held a handful of digits and had to be scanned straight on, one line at a time. The factory needed to hold more and to scan quickly from any angle, on dirty, scuffed parts.

Denso Wave holds the patent but chose not to enforce it and published the standard openly, which is a big reason you now see these squares on everything from bus tickets to wedding invitations.

## Reading the square

Every QR code is a grid of tiny squares called **modules**, each one either dark or light. Here is what your phone does with it in a blink:

**From camera to answer**

1. **Find the three big squares.** These corner marks are the same in every code. They tell the phone where the code is and which way up it is.
2. **Straighten it.** A smaller square pattern helps correct for tilt, curves and perspective.
3. **Read the grid.** Dark and light modules become ones and zeros, in a fixed zig-zag order.
4. **Repair the damage.** Extra error-correction data is used to fix any modules that were unreadable.
5. **Hand over the text.** A link opens, a phone number dials, plain text appears.

Notice that there is no fourth big square. Three corners are enough, and their layout is what tells the phone which way is up, so you can scan a code upside down or sideways and it still reads.

## The part that makes it tough: error correction

Some of the modules in every code are not your data. They are **repair information**, calculated with a type of maths called Reed-Solomon coding (the same family used in CDs and in space communications). If part of the code is damaged, the phone uses the repair information to rebuild what was lost.

When you make a code you choose how much repair information to include:

| Level | Can lose up to | Best for |
| --- | --- | --- |
| **L**, Low | about 7% | Clean screens, short codes you want as simple as possible |
| **M**, Medium | about 15% | A sensible default for most printing and screens |
| **Q**, Quartile | about 25% | Posters, packaging, anything that will be handled |
| **H**, High | about 30% | Codes with a logo in the middle, or outdoor and rough surfaces |

*The percentage is the share of the code that can be unreadable while still decoding. Higher levels cost you extra modules.*

There is a trade. More repair data means more squares, which means each square must be smaller, which makes the code harder to scan from far away. That is the whole game: **robust or simple, pick based on where it will live**.

The logo-in-the-middle codes you see on brands are using that built-in repair capacity on purpose. They cover the centre and trust the error correction to fill it in. That is why they are normally made at level H, and why putting a big logo on a level L code makes it fail.

Try it: [QR Code Generator](https://toolcloak.pages.dev/tools/qr-generator). Paste a link, text, email or phone number, pick a size, download the PNG.

## Shorter content, better code

Codes come in 40 sizes called versions. Version 1 is 21 by 21 modules and each step up adds four to each side, up to 177 by 177 at version 40. The more you put inside, the bigger the version, and the smaller each module is for a given printed size.

**Modules along one side of a QR code**

|  | Modules per side |
| --- | --- |
| Version 1, a few digits | 21 |
| Version 5 | 37 |
| Version 10, a longer web link | 57 |
| Version 25 | 117 |
| Version 40, the maximum | 177 |

*At the top end a code can hold around 7,000 digits or about 3,000 bytes of text. Almost nobody needs that, and a code that dense is hard for a phone to read from a table.*

The practical lesson: **use a short link**. A link of 25 characters makes a code with big, forgiving squares. A link of 250 characters, full of tracking parameters, makes a dense code that needs a steady hand and good light. If your address is long, put a short redirect page on your own site and encode that.

## Static or dynamic?

This distinction decides whether your code will work in five years.

- A **static** code contains the information itself. The link, the text or the phone number is written into the squares. It works forever, needs no account and nobody can switch it off. You also cannot change it afterwards.
- A **dynamic** code contains a short address that belongs to a company, which then redirects the visitor to your real link. You can edit the destination and see scan counts. If you stop paying, the code can die, and so can every poster you printed.

The codes you make here are static. For a menu, a Wi-Fi password, a contact number, a link on a business card, that is usually the right choice. If you truly need scan analytics or a changing target, a dynamic service is the honest answer, but you are renting the code.

## Print rules that save you a reprint

- **Quiet zone.** Leave an empty margin at least four modules wide around the code. Text or a border pressed up against it confuses the scanner.
- **Contrast.** Dark code on a light background. Avoid pale colours and be careful with inverted (light on dark) codes, which some scanners cannot read.
- **Size.** Around 2 cm square is a sensible minimum for things held in the hand. A common rule of thumb for distance is about ten times the code's width, so a 10 cm code works from roughly a metre.
- **Surface.** Glossy, curved or wrinkled surfaces bend the grid. Use higher error correction and test on the real object.
- **Test before the print run.** Scan with two different phones, an older one included, in poor light, from the distance a real person will use.

## A neat trick: Wi-Fi in a code

Phone cameras understand a special text format for Wi-Fi networks. Create a code from plain text in this form:

`WIFI:T:WPA;S:YourNetworkName;P:YourPassword;;`

Replace the name and password, generate the code, print it and stick it near the router. Guests point a camera at it and join without typing a single letter. Our [Wi-Fi QR Code Generator](https://toolcloak.pages.dev/tools/wifi-qr-code-generator) builds that text for you from two boxes. Two cautions: anyone who can see the code can read the password, so this suits a guest network and not your main one. And these online generators, ours included, draw the image through an outside QR service, so a password you care about deserves a generator that works fully offline.

## One last warning

> **Look before you open:** A QR code is just a link in disguise, and you cannot read it with your eyes. Scammers stick fake codes over real ones on parking meters and posters. Your phone shows the web address before it opens. Read it, and be extra careful whenever a code asks for a payment or a login.

## Frequently asked questions

### Do QR codes expire?

A static QR code, which holds the link or text inside the squares, never expires. A dynamic code can stop working if the service that redirects it is cancelled or you stop paying.

### What error correction level should I choose?

Level M is a good default. Use Q or H for printed material that may get scuffed, and H if you plan to place a logo in the middle. Level L suits clean, on-screen codes where you want the fewest modules.

### What size should a printed QR code be?

At least about 2 cm square for things held in the hand, and larger for posters. A common rule of thumb is a distance of roughly ten times the code's width. Always leave a blank margin around it and test the real print.

### Why will my QR code not scan?

The usual causes are too much data (a very long link), low contrast, no blank margin around the code, a code printed too small, or a logo that covers more than the error correction can repair. Shorten the link, raise the error correction and enlarge it.

### Can I put a QR code on a business card?

Yes. Keep the link short, use at least level M, print it at 2 cm or more, and encode a page you control so you can update what visitors see.



---

# ₹5,000 a month for 20 years: what a SIP calculator can and cannot tell you

> Put in ₹12 lakh over two decades and the screen says ₹50 lakh. That number is real arithmetic and also a bit of a fairy tale. Here is how to read it without fooling yourself.

Published 2026-10-07 by Toolcloak. 5 minute read. Original: https://toolcloak.pages.dev/articles/5000-rupees-a-month-for-20-years-what-a-sip-calculator-can-tell-you
Related free tool: [SIP Calculator](https://toolcloak.pages.dev/tools/sip-calculator)

Type ₹5,000, 12 percent, 20 years into a SIP calculator and it answers **₹49,95,740**. You put in ₹12,00,000. The rest, nearly ₹38 lakh, is growth. It is a lovely number and it makes you want to start tomorrow.

It deserves a little suspicion, not because the arithmetic is wrong (it is perfectly right) but because of what you feed into it. A calculator multiplies your assumptions. It does not know whether they will come true. So the useful skill is not operating the calculator. It is knowing which of the numbers you can trust and which are a guess.

## First, what a SIP is not

A SIP, systematic investment plan, is not an investment. It is a **way of paying**: a fixed amount moves from your bank account into a mutual fund on a fixed date each month. What the money is invested in depends on the fund you choose. A SIP into an equity fund, a debt fund and a gold fund will give you three very different journeys.

This matters because the calculator asks for an “expected return”, and that number belongs to the fund, not to the SIP. Nobody can promise it.

## What the calculator is doing

Each month your money earns a month's worth of return, then your next ₹5,000 lands, then the whole pile earns again. That is compounding: growth earning growth. For a 12 percent yearly return, the calculator uses 1 percent per month.

**₹5,000 a month, two decades (value in lakh rupees)**

|  | Value if the fund grows 12% a year | Value if it grows 8% a year | Money you put in |
| --- | --- | --- | --- |
| 0 | 0 | 0 | 0 |
| 2 | 1.36 | 1.31 | 1.2 |
| 4 | 3.09 | 2.84 | 2.4 |
| 6 | 5.29 | 4.63 | 3.6 |
| 8 | 8.08 | 6.74 | 4.8 |
| 10 | 11.62 | 9.21 | 6 |
| 12 | 16.11 | 12.11 | 7.2 |
| 14 | 21.82 | 15.5 | 8.4 |
| 16 | 29.07 | 19.49 | 9.6 |
| 18 | 38.27 | 24.16 | 10.8 |
| 20 | 49.96 | 29.65 | 12 |

*Years on the bottom. The grey dashed line is simply what you deposited. The gap above it is growth. Notice how late the curve takes off.*

Look at the shape. For the first ten years the curve hugs the line of your own deposits. At year 10 you have put in ₹6 lakh and have ₹11.6 lakh. Then it bends. Between year 15 and year 20, the value goes from ₹25.2 lakh to ₹49.96 lakh. **The last five years add almost as much as the first fifteen put together.**

That is the whole secret, and it is also the most boring advice there is: the biggest part of the result comes from the part where nothing seems to be happening. People who stop at year six, because it feels slow, quit just before it gets interesting.

Try it: [SIP Calculator](https://toolcloak.pages.dev/tools/sip-calculator). Monthly amount, expected return, years. Add a yearly step-up to see what raising your SIP does.

## Time beats amount

| Years | You put in | Value at 12% a year | Value at 8% a year |
| --- | --- | --- | --- |
| 5 | ₹3,00,000 | ₹4.12 lakh | ₹3.70 lakh |
| 10 | ₹6,00,000 | ₹11.62 lakh | ₹9.21 lakh |
| 20 | ₹12,00,000 | ₹49.96 lakh | ₹29.65 lakh |
| 30 | ₹18,00,000 | ₹1.76 crore | ₹75.0 lakh |

*Same ₹5,000 a month in every row. Only the number of years changes. The 8% column uses the same formula with a lower assumed return.*

Compare 20 years with 30. You deposit 50 percent more money (₹18 lakh instead of ₹12 lakh) and end up with **3.5 times** as much. Starting ten years later and doubling the SIP does not catch you up. Time is the one input you cannot buy back.

## Three honest corrections

### 1. The return is an assumption

Equity funds have, over long periods, delivered returns in the ballpark people quote. They do it unevenly: a year of 30 percent, a year of minus 15. The calculator draws a smooth line through all that. Your real journey will be bumpy, and the end value at any particular date can be well above or below the smooth line.

That is why you should always run the calculator **twice**: once with the number you hope for, once with a number that would disappoint you. If the disappointing version still gets you where you need to be, you have a plan. If only the hopeful version works, you have a wish.

### 2. Inflation eats the final number

₹49.96 lakh in 20 years is not ₹49.96 lakh of today's buying power. At 6 percent yearly inflation, it is worth about **₹15.6 lakh** in today's money. Still good. Just not what the headline implies. When you set a goal, decide whether it is “₹50 lakh” or “what ₹50 lakh buys today” and calculate accordingly.

### 3. Costs and tax are not in there

Fund expenses are deducted from the fund's returns, and gains are taxed when you sell. The calculator shows neither. Treat its answer as the number before costs and before tax.

> **Calculators are not guarantees:** Mutual fund investments are subject to market risk, and past returns do not predict future ones. This article explains how the arithmetic works. It is not investment advice, and we are not licensed financial advisers.

## The step-up trick

Most people's income grows. Their SIP often does not, because nobody remembers to change it. The calculator has a yearly step-up option for exactly this.

Start with the same ₹5,000 but raise it by 10 percent every year (₹5,500 in year two, ₹6,050 in year three, and so on). At 12 percent for 20 years you deposit about **₹34.4 lakh** in total, and the value comes to about **₹99.4 lakh**. That is double the flat SIP result, for a monthly increase you would barely notice in any single year.

The easy way to do it: every time you get a raise, move a slice of the raise into the SIP before you get used to spending it.

## Why monthly, and why bother

When the market falls, your fixed ₹5,000 buys more units. When it rises, it buys fewer. Over time your average purchase price ends up lower than the average price, a mechanical effect often called rupee-cost averaging. It does not remove risk. What it does is remove the temptation to wait for “the right time”, which nobody has ever reliably found.

The deeper benefit is behavioural. A SIP that runs automatically on the 5th is one fewer decision you can talk yourself out of when the news is bad.

## How to use the calculator well

1. Start with a **goal and a date**, not an amount. “₹25 lakh for a down payment in 12 years” beats “whatever I can spare”.
2. Run two returns: one you hope for and one that would hurt a little.
3. Subtract inflation from your goal in your head, or inflate the target upward.
4. Turn on the **step-up** and pick a number you can honestly commit to.
5. Set it up, then leave it alone. Look at it once or twice a year.

## Frequently asked questions

### How is SIP return calculated?

The calculator treats each monthly instalment as earning the expected return for the remaining months. For a 12 percent yearly return it uses 1 percent a month, adds each instalment as it is paid, and lets the whole balance compound. The answer is the total value at the end of the period.

### How much will ₹5,000 a month grow to in 20 years?

At an assumed 12 percent a year, about ₹49.96 lakh on ₹12 lakh invested. At 8 percent, about ₹29.65 lakh. These are estimates from the formula, not promises, and real returns vary year to year.

### Is 12 percent a safe number to assume?

It is an assumption, not a guarantee. It is sensible to run the calculator with a lower return as well, so your plan still works if markets disappoint.

### What is a step-up SIP?

A SIP whose amount increases by a set percentage every year. With a 10 percent step-up starting at ₹5,000, you invest about ₹34.4 lakh over 20 years instead of ₹12 lakh, and at 12 percent the estimated value roughly doubles to about ₹99 lakh.

### Does the calculator include tax and fund charges?

No. It shows the growth before expenses and before tax, so your actual result will be somewhat lower.



---

# What counts as fast typing? (It is not 100 WPM.)

> You typed 38 words per minute and felt slow. Then someone told you their friend does 120. Here is what the number really measures, what normal looks like, and why your accuracy is quietly costing you more speed than your fingers are.

Published 2026-10-07 by Toolcloak. 4 minute read. Original: https://toolcloak.pages.dev/articles/what-counts-as-fast-typing
Related free tool: [Typing Speed Test (WPM & Accuracy)](https://toolcloak.pages.dev/tools/typing-speed-test)

Somewhere there is a person who types 140 words a minute and mentions it at parties. Ignore them. For almost every job, and almost every day of your life, the number that matters is much smaller and much more boring than that: how fast can you type **without mistakes you have to fix**?

That is a very different question from the one most typing tests seem to ask, and it is the one that decides whether you finish your report at five or at seven.

## What a “word” is, in typing

Here is a small surprise. In a typing test a word is **not** a word. It is five characters, spaces included. “Elephant” is a bit under two words. “It is” is, well, one.

The reason is fairness. If tests counted real words, someone typing short ones like “a cat sat” would score far higher than someone typing “international”, even though they pressed the same number of keys. Fixing a word at five characters makes the score about keystrokes.

So the formula is nothing mysterious:

`WPM = (characters typed ÷ 5) ÷ minutes`

Type 300 characters in one minute and you scored 60 WPM. Type 150 characters in 30 seconds and you also scored 60 WPM, because the clock was halved too.

## What is normal?

**Rough ranges, words per minute**

|  | WPM |
| --- | --- |
| Hunt and peck | 10 to 25 |
| Typical adult | 40 to 50 |
| Office professional | 60 to 70 |
| Quick typist | 80 to 100 |
| Competitive typist | 120 and up |

*Loose bands, not official categories. Bars are drawn at the middle of each band. A large 2018 study of many thousands of volunteers, run by researchers at Aalto University, reported an average of roughly 50 WPM.*

Where do you fit? If you land between 40 and 60, you are completely ordinary and perfectly fast enough for nearly every job that involves a keyboard. Beyond about 70 to 80, the extra speed saves you very little in practice, because typing is rarely what limits your work. Thinking is.

Try it: [Typing Speed Test (WPM & Accuracy)](https://toolcloak.pages.dev/tools/typing-speed-test). Choose 30, 60 or 120 seconds. See every mistake in red as you type, and your accuracy live.

## The accuracy trap

Let us do some arithmetic with 60 WPM, which is 300 characters a minute. Suppose your accuracy is 95 percent. That sounds good, right? It means **15 wrong characters every minute**. Each one needs you to notice, stop, delete, retype, and usually re-find your place. A fix takes a second or two at least.

Fifteen corrections is 20 to 30 seconds out of every minute spent repairing. Your real, finished speed is nowhere near 60. Meanwhile someone typing at 50 WPM with 99 percent accuracy makes three errors a minute and keeps flowing.

> **The takeaway:** Speed you have to correct is not speed. Slow down until your accuracy is about 97 to 98 percent or better, then push the pace up gradually. The fast people you envy mostly got there by being accurate first.

This is why serious tests report a **net** speed, which subtracts uncorrected errors from the raw number. Watching the red letters during a test is not about shame. It is the most useful feedback you will get.

## Why the three test lengths exist

| Length | What it really measures | Good for |
| --- | --- | --- |
| 30 seconds | Short burst speed. Little fatigue, so your best number. | A quick warm-up, a first-ever benchmark |
| 60 seconds | The standard. Long enough to settle into a rhythm. | Comparing yourself over weeks |
| 120 seconds | Stamina and consistency. Mistakes and tired fingers show up. | A realistic score for a working day |

*Expect your two-minute number to be a little lower than your 30 second one. That gap is normal, and it is the more honest figure.*

## Why your score jumps around

Two runs back to back can differ by 10 WPM, and nothing is wrong. Three things move the number a lot:

- **The text itself.** Commas, capital letters, numbers and rare words slow everyone down. A passage of plain lowercase sentences is much easier.
- **Your state.** Tired, rushed or watching the clock tightens your hands. A relaxed run is usually the fastest.
- **Your keyboard.** A laptop you do not use daily, a stiff phone keyboard or an unfamiliar layout will cost you real speed.

For the same reason, comparing your score from one website with someone else's from another is nearly meaningless. Different text, different rules for errors, different timing. Compare with **yourself**, on the same test, once a week.

## How to get faster, in the order that works

1. **Accuracy first.** Aim for 98 percent. Drop the speed until you get there, then raise it slowly.
2. **Stop looking at the keys.** This is the biggest single change, and it feels awful for about two weeks. Cover your hands with a cloth if you have to.
3. **Learn the home row.** Fingers rest on A S D F and J K L ; with thumbs on the space bar. Each finger owns a set of keys, and you always return home.
4. **Practice short and often.** Ten focused minutes a day beats an hour on Sunday. Your fingers learn by repetition, not by effort.
5. **Type in a steady rhythm.** Smooth beats fast. Rushing creates the stumble that wipes out the gain.
6. **Practise your own words.** The test lets you paste in your own text. If you type emails in Hinglish or code at work, train on that, since it is the thing you really type.

Most people who follow that routine are 15 to 20 WPM better within a couple of months, and, which matters more, they stop reaching for the backspace every few words.

## A word about phones

Thumb typing on a phone is a different skill with its own average, usually lower than desktop typing, with autocorrect quietly changing the maths. If you are mostly on a phone, a laptop test will understate how fast you are in practice and overstate how much a keyboard course would help. Take the result for what it is: a measure of how you type on this device, on this text, today.

## Frequently asked questions

### What is a good typing speed?

For most office work, 40 to 60 WPM with high accuracy is plenty. Around 65 to 75 is considered very good for professionals, and above 100 is exceptional. Accuracy matters as much as speed.

### How is WPM calculated?

Characters typed divided by 5 gives the number of words, and that is divided by the time in minutes. Five characters, spaces included, count as one word, so the score reflects keystrokes rather than word length.

### What is the average typing speed?

Around 40 to 50 WPM for adults overall. A large 2018 study by Aalto University researchers reported an average close to 50 WPM.

### How can I type faster?

Raise your accuracy first, learn the home row and stop looking at the keys. Practise ten minutes daily and type at a steady rhythm. Most people improve noticeably within a few weeks.

### Why is my typing speed different each time?

The passage, your focus and your keyboard all change the result. A difficult text with punctuation and numbers will always score lower. Compare tests on the same settings over time, not one run against another.



---

# How long is long enough? A field guide to word and character limits

> A tweet, an Instagram caption, a meta description, an SMS, a college essay: all of them have a ceiling, and most of them count something different. What each limit really measures, and how to write inside it without gutting the message.

Published 2026-10-07 by Toolcloak. 5 minute read. Original: https://toolcloak.pages.dev/articles/how-long-is-long-enough-a-field-guide-to-word-and-character-limits
Related free tool: [Word Counter](https://toolcloak.pages.dev/tools/word-counter)

Your Instagram caption is perfect. You paste it in, tap post, and the app politely chops the last sentence off, which happened to be the joke. Or the college portal says 500 words, you write 560, and the form simply refuses to submit until you delete a paragraph you liked.

Limits are annoying, but they are also predictable. Once you know what each platform counts, and where it quietly cuts, you can plan for it from the first line instead of rescuing the draft at the last minute.

## Words, characters and the space question

Most counters, ours included, define a **word** as a run of characters between spaces. That is simple and mostly right. It also means “well-known” is one word, “well known” is two, and a phone number is one. For an essay limit it rarely matters. For a strict count, check how your institution counts.

**Characters** come in two flavours: with spaces and without. Social platforms count with spaces, and so do SMS and search results. Some university forms count without. A good counter shows both, so you are never guessing.

A **sentence** is a stretch ending in a full stop, question mark or exclamation. A **paragraph** is a block separated by a blank line. These are used for readability checks rather than hard limits.

## The limits worth knowing

**Common character limits**

|  | Characters |
| --- | --- |
| SEO page title (shown) | about 60 |
| Instagram bio | 150 |
| SMS, one message (English) | 160 |
| Meta description (shown) | about 160 |
| Post on X | 280 |
| Instagram caption | 2200 |
| LinkedIn post | 3000 |

*Platforms change their limits from time to time, so treat these as a guide and check the app if it matters. The bars are drawn to the same scale, which is why the short ones look so tiny next to a LinkedIn post.*

A few of those deserve a footnote.

- **SEO title and meta description** are not hard limits. Google measures them in pixels, not characters, and cuts them with an ellipsis when they run over. Around 60 characters for the title and 150 to 160 for the description is the safe zone. Wide letters like W and M use up space faster than I or l.
- **SMS** allows 160 characters in a single message when it is plain English. The moment you add a character outside that basic set, the message switches to a different encoding and the limit drops to **70 characters**. Text in Hindi or other Indian scripts always falls in that second group. A Hindi message of 100 characters gets split into two messages, and some phones show it as two.
- **Instagram** shows only the first line or two of a caption before “more”, so the hook has to live there even though 2,200 characters are allowed.
- **X** counts every link as a fixed length regardless of how long the link really is, and emojis count as more than one character.

Try it: [Word Counter](https://toolcloak.pages.dev/tools/word-counter). Words, characters, sentences, paragraphs and reading time, updating as you type.

## How long does it take to read?

Silent reading for an average adult is somewhere around 200 to 250 words a minute. Reading aloud is slower, and a presenter speaking naturally covers roughly 130 to 150. These are averages. Dense, technical material is slower and a light novel is faster.

| Length | Silent reading | Spoken |
| --- | --- | --- |
| 300 words | about 1.5 minutes | about 2 minutes |
| 500 words | about 2 to 2.5 minutes | about 3.5 minutes |
| 1,000 words | about 4 to 5 minutes | about 7 minutes |
| 1,500 words | about 6 to 7 minutes | about 10 to 11 minutes |
| 2,500 words | about 10 to 12 minutes | about 17 to 18 minutes |

*Silent reading at about 225 words a minute and speaking at about 140. Use your own pace for a speech: read the draft out loud once with a timer.*

This is the maths behind the “5 min read” label on a article, and it is also the practical answer to “how long should my speech be?”. A five minute talk is about 700 words. Anyone who has watched a presenter try to cram 1,500 into the same slot knows what happens.

## Writing inside a limit

Cutting words well is a skill, and it is easier than it sounds. When you are 40 words over, work through these in order.

1. **Delete the warm-up.** The first sentence of most drafts clears the throat. Try starting at sentence two.
2. **Kill the stack of adjectives.** “A very big and quite exciting opportunity” becomes “a big opportunity” and nobody misses the rest.
3. **Swap phrases for words.** “In order to” is “to”. “Due to the fact that” is “because”. “At this point in time” is “now”.
4. **Cut the doubled idea.** If two sentences say the same thing, keep the better one.
5. **Check for “that”.** In many sentences, “that” can go and the meaning stays.

Do this with the counter open and watch the number drop. It is oddly satisfying, and the writing is almost always better at the end.

## Do not write to a number

There is a persistent idea that web articles need to hit a magic word count to rank. They do not. Search engines do not count words as a quality signal, and plenty of 300 word pages beat 3,000 word ones because they answer the question. Write until the question is answered, then stop.

Word counts are a **ceiling** (a limit you must respect) or a **budget** (a way to plan how much space each section gets), not a target to inflate toward. Padding is easy to spot, and readers leave.

## For students

- Aim for the limit, not far under it. A 500 word limit does not reward 320 words.
- A small overshoot is often tolerated, but never assume. When a portal counts for you, it simply will not accept more.
- Keep a separate version with the exact text you submit, and count that copy. Headers, references and captions may or may not be counted.
- Write the first draft long, then cut. Cutting is easier than stretching.

## A last, small point about unique words

The tool also counts **unique words** and shows an average word length. If your unique-word count is much lower than your total, you are repeating yourself, and the draft may feel flat. For essays, a quick look at that ratio is a decent test of whether you are saying many things or one thing many times.

## Frequently asked questions

### How many words is 1,000 words in minutes?

About 4 to 5 minutes of silent reading at 200 to 250 words a minute, and around 7 minutes if spoken aloud at a natural pace.

### What is the character limit for a tweet?

280 characters on X for standard accounts. Links are counted at a fixed length and emojis count as more than one character, so a counter that matches the platform's rules is safer than counting by eye.

### How long should a meta description be?

Around 150 to 160 characters. Google measures in pixels, so it may cut earlier or later, but staying around that length keeps the important part visible in search results.

### Why does my SMS get split into two messages?

A single SMS holds 160 characters in the basic character set but only 70 when any character outside that set is used, which includes Hindi and other Indian scripts and many emojis. Anything longer is divided into several messages.

### Does a word counter count spaces?

Words are counted by the gaps between them, and characters are normally shown both with and without spaces. Use the version that matches the limit you are working to.