Galton Board

Press Start to begin.
Ctrl+wheel to zoom
drag to pan
double-click to reset

Using the Galton Board

Click Start to see balls rolling down the board, striking pegs and bouncing left or right in each of several rows. The default setting drops 3000 balls through 12 rows of pegs. Perform the following steps, in order.

  1. Focus on one ball. In each row, the ball has a 50% chance of bouncing left or right when it hits the peg. The direction the ball bounces in a given row simulates a single coin flip. Think of a left bounce as a tail and a right bounce as a head.

    1. Which way the ball bounces in a given row does not depend on which way it bounced in previous rows. That is, the coin flips are independent.
    2. After the ball completes all 12 rows, the chute it lands in at the bottom depends on the number of heads (right bounces). For example, the leftmost, middle, and rightmost chutes correspond to 0, 6, and 12 heads, respectively. Where a ball lands at the bottom simulates the number of heads in 12 independent coin flips.
    3. Now focus on a different ball. It may land in a different chute at the bottom because where it lands is random, just as when you repeat the experiment of flipping a coin 12 times, you may see a different number of heads.
  2. Now focus on the bars above the chutes at the bottom.

    1. Bar heights are proportional to the numbers of balls landing in different chutes.
    2. After many balls drop, bar heights become proportional to probabilities that the number of heads in 12 coin flips are 0, 1, 2, …, 12 — the binomial distribution with n = 12 and p = 1/2.
  3. Change the speed that balls drop using the Speed control in the sidebar. Increasing the speed lets you see the shape of the distribution more quickly. You can also bypass the animation entirely by clicking Finish in the sidebar, which immediately shows the final distribution after all 3000 balls drop.

Interim Summary

  • Each bounce in a given row simulates one coin flip.
  • Where a ball lands at the bottom simulates the number of heads in 12 coin flips.
  • Heights of bars above chutes summarize results after simulating the 12-coin-flip experiment many times. They become proportional to binomial n = 12, p = 0.5 probabilities. Notice the bell shape of the distribution of bar heights.
  1. Now change p in the sidebar from 0.5 to 0.1 and press Start to simulate 12 flips of an unfair coin with probability 0.1 of heads. The distribution of bar heights now looks different — the left side of the bell is mostly missing.

    1. Increase n from 12 to 30 and restart. After many balls drop, bar heights regain their bell shape.
    2. Increase n from 30 to 60 and restart. After many balls drop, the distribution of bar heights looks even more bell-shaped.
    3. Increase n from 60 to 120 and restart. The bell shape is even more pronounced.
  2. Now change p to 0.9 and repeat step 4. The shape of the distribution at the bottom is not bell-shaped with 12 rows, but becomes bell-shaped as the number of rows becomes large.

Interim Summary

  • Changing the probability of heads to close to 0 or 1 makes the shape of the (binomial) distribution of the number of heads in 12 coin flips lose its bell shape.
  • Nonetheless, dramatically increasing the number of rows (coin flips) restores the bell shape.
  • The Galton board illustrates the Central Limit Theorem: The distribution of the sum of a large number of random deflections is approximately normal.
  1. Change the number of balls from 3000 to 10 and restart the Galton board with the default settings of n = 12 rows and p = 0.5.

    1. Notice that the number of balls is too small to get a good idea of the distribution of balls at the bottom.
    2. Re-run the Galton board with 10 balls, n = 12 rows, and p = 0.5. You may see a very different distribution than you saw in part (a) because 10 balls are not enough to accurately estimate the distribution of balls at the bottom.
    3. Now increase the number of balls to 100 and re-run, then increase to 1000 and then to 10 000. Remember that you can click Finish to see the final distribution without waiting for all balls to drop. Notice that the distribution of balls at the bottom remains remarkably stable once the number of balls is 1000 or more.

Final Summary

  • The Galton board is a device for simulating independent coin flips.
    • The number of rows is the number of coin flips; which way a ball goes in a given row simulates a single coin flip.
    • Where a ball lands at the bottom simulates the number of heads in n flips.
    • The distribution of balls in the chutes simulates what happens when the experiment of n coin flips is repeated many times.
    • When the number of balls is large, the distribution of balls at the bottom is proportional to the binomial distribution with n trials and success probability p.
    • When p is close to 0.5, the distribution looks bell-shaped even with a small number of rows.
    • When p is close to 0 or 1, the distribution does not look bell-shaped with a small number of rows, but becomes bell-shaped as the number of rows grows. This illustrates the Central Limit Theorem: the sum (or average) of a large number of independent deflections has an approximate normal distribution.
© 2026 Warren G. Clary. All rights reserved.