# Galton board > How does a bell curve build from falling balls? Drop balls through rows of pegs and watch the pile grow into the bell shape of the normal distribution. Drop balls through a Galton board and watch a binomial pile turn into a normal bell curve. Change the rows and the bias and compare with the theory. - Page: https://how.box/galton/ - Runs entirely in the browser. - License: HowToolbox's own text and data (titles, explanations, FAQs, tool descriptions and schemas, this file) are CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/); AI training and answers are welcome. Attribute "HowToolbox (how.box)" with a link to the page used. Third-party data keeps its own terms; the name and logo are not licensed. Publisher: HowToolbox, contact@how.box. Terms: https://how.box/about/terms/ ## Inputs Answer link: https://how.box/galton/?r={r}&p={p}&n={n} - `r` (whole number 2 to 20): Rows of pegs. Example: `12`. - `p` (number 5 to 95): Chance in percent of bouncing right at each peg. Example: `50`. - `n` (whole number 10 to 5000): Number of balls in the first run. Example: `300`. Example answer: https://how.box/galton/?r=12&p=50&n=300 A complete link carries the primary answer in the page HTML; a missing or invalid value shows the default page. ## Agent tools - `galton_calculate`: Galton board: `balls` balls fall through `rows` rows of pegs and bounce right with probability `bias` percent at each peg; the bin is the number of rights (0 to rows), Binomial(rows, bias/100). - MCP, one POST to https://how.box/mcp: `{"jsonrpc":"2.0","id":1,"method":"tools/call","params":{"name":"galton_calculate","arguments":{"rows":12,"bias":50,"balls":300}}}` - GET: https://how.box/mcp/run/galton_calculate?rows=12&bias=50&balls=300 ## How it works At every peg a ball bounces left or right, about as likely either way on a fair board. After the last row its place depends only on how many times it went right. Many different routes lead to the middle bins and only one route leads to each edge, so the middle fills up the most. The pile becomes a bell. The count of rights is a binomial number: with n rows and a chance p to go right, the chance of exactly k rights is C(n, k) p^k (1 − p)^(n − k). Its average is n·p and its spread is √(n·p·(1 − p)). With 12 rows and p = 0.5 that is an average of 6 rights and a spread of 1.73 bins. As n grows the bars follow the normal curve more closely (the central limit theorem). A bias such as 80 percent shifts and squeezes the pile. Francis Galton built the board in the 1870s to show how many small random pushes add up to a bell curve. With only a few hundred balls the pile is bumpy: chance alone moves each bar up and down, so it never matches the curve exactly. ## FAQ **Why does a Galton board make a bell curve?** Each ball's final place is the sum of many small left and right bounces. Sums of many small independent random steps pile up in a bell shape. **Why is my pile not perfectly symmetric?** Balls are random. With a few hundred balls each bar varies by a few balls. More balls bring the shape closer to the theory. **What happens when the bias is not 50 percent?** The pile shifts towards the favoured side and becomes narrower and lopsided. The average moves to rows times the chance of going right. **How many rows give a good normal curve?** More rows give a smoother, closer curve. With about ten rows the bars already follow the curve well on a fair board. **Is the first run the same every time?** Yes. The first run is fixed so a shared link shows the same pile. Dropping more balls adds new random ones. More: [HowToolbox index](https://how.box/llms.txt) · [Agent Skill](https://how.box/skill.md) · [agent terms](https://how.box/about/agents/)