The Galton board: why do randomly falling balls draw a "bell" shape every time?

Toss a coin: you cannot know in advance whether heads or tails will come up. But if you toss a thousand coins, you can say with confidence that the result will be roughly half and half. In the 19th century an English scientist demonstrated this idea, "a single event is random, but many events obey a law", with a simple wooden board and pins.

How is the board built?

According to the description of the Galton Collection at UCL (University College London), the device consists of an upright board with pins driven into it in rows, staggered so that each row falls between the pins of the row above. A ball dropped from the top bounces either to the right or to the left each time it hits a pin and finally falls into one of the narrow compartments at the bottom. As the number of balls grows, the columns in the compartments take the shape of a bell-like curve.

The device is known by various names: "quincunx" (Latin, the diamond arrangement of the pins), "Galton box", "bean machine". Its author was the English scientist Francis Galton (1822–1911), who, according to the Encyclopedia of Mathematics, is regarded as one of the scientists who laid the foundations of the concepts of regression and correlation in statistics.

When did it appear?

Sources do not agree on the exact year the board was made and first shown. According to the catalogue record of the surviving example in the UCL collection, it was made in London by Tisley and Spiller to Galton's design and is dated 1873. A study published on arXiv in 2026, however, links the construction of the board to 1875. What is known for certain is that Galton described the device in detail in his book "Natural Inheritance", published in 1889, and called it a "quincunx" (the text of the book is available on the galton.org website).

Why exactly a "bell"?

The reason can be understood with simple logic. For a ball to land in the outermost compartment, it has to bounce in only one direction at every pin, for example to the left 10 times in a row. Such "luck" is rare. The middle compartment, on the other hand, can be reached by a great many paths in which the left and right bounces are equal. That is why many balls collect in the middle and few at the edges.

In mathematics this is called the binomial distribution, and as the number of rows grows it approaches the normal distribution. The Galton board shows this theory before your eyes, without formulas. The UCL collection notes that the device also helped Galton explain the idea of "reversion to the average" (he called it "regression to mediocrity"). Today the term "regression" is the name of one of the main methods of statistics.

An important note: the UCL Science Collections page points out that this achievement of Galton's in statistics is rarely mentioned because his name is associated with eugenics (the idea of sorting people by "breed"). The ideas of eugenics were later strongly condemned; here we are speaking only about the device's place in the science of statistics.

Then and now

Aspect	Galton's time (late 19th century)	Today
How randomness was shown	A wooden board, pins, shot	Computer simulation, online interactive models
Scale of the experiment	Lead shot dropped by hand	Millions of "balls" within seconds
Who uses it	A scientist in a lecture	A school pupil, a student, a risk analyst
Main question	Why do measurements cluster around the average?	How spread out is the result, what is the probability of error?

Example: Dilnoza opa's class experiment

The experiment and calculation below were compiled by the author.

Dilnoza opa, a school teacher in Samarkand, made a small board with 10 rows of pins together with her pupils and planned to drop 1 024 beads through it. Why exactly 1 024? Because with 10 rows each bead follows one of 2¹⁰ = 1 024 different paths, and the number of paths leading to each compartment equals the binomial coefficients. So with 1 024 beads the expected numbers come out exactly equal to these coefficients.

Compartment (number of bounces to the right)	0	1	2	3	4	5	6	7	8	9	10
Expected beads	1	10	45	120	210	252	210	120	45	10	1
Computer simulation (one run)	1	8	37	120	205	259	222	127	35	9	1

Three conclusions from the table: - The single middle compartment receives 252 ÷ 1 024 ≈ 24.6% of the beads. - The three middle compartments (4, 5, 6) receive 210 + 252 + 210 = 672 in total, that is ≈ 65.6%. - The four outermost compartments (0, 1, 9, 10) receive only 1 + 10 + 10 + 1 = 22, that is ≈ 2.1%.

The simulation row differs slightly from the expected values: randomness gives its own "noise" every time, but the overall shape does not change. In class the result will be just the same.

The practical meaning: in a business, measurements (for example, the weight of a packaged product or daily sales) also often cluster around the average. If the histogram of the data differs sharply from a bell shape, that is a signal to check: there may be a measurement error, mixed data or an unusual event.

Link with the previous period: in the 19th century Quetelet had shown that people's height and other measurements cluster around the average; Galton made visible why this regularity arises. The next stage, the sampling method of the 20th century, relied precisely on the properties of this distribution to teach how to draw conclusions without asking the whole population.