Every lesson in this course runs on one toy: a board, some pegs, a coin. This page is about the real one — where it came from, the restless mind that built it, the mathematics it dramatizes, and the darker ideas it was also enlisted to serve.
Each bead makes an independent left-or-right choice at every pin. No single path is predictable — yet the crowd always finds the curve.
0 beads settled
The faint dashed line is the theoretical normal (Gaussian) curve the bins are converging toward. Nudge the bias slider and pour again — the shape survives, only its centre moves.
In the winter of 1874, before an audience at London's Royal Institution, the Victorian polymath Francis Galton tipped a stream of lead shot into a glass-fronted box. The pellets rattled down through rows of pins, each one knocked left or right, their paths hopelessly tangled. And yet when they came to rest in the columns below, they had piled themselves into a smooth, symmetrical mound — the bell-shaped curve statisticians call the normal distribution. Galton had built a machine that made an abstract law of mathematics visible to the eye. He called it the quincunx; today it's the Galton board.
Galton (1822–1911) was one of his era's great scientific magpies — a gentleman of independent means who scattered his energy across an astonishing range of fields.
He was a half-cousin of Charles Darwin, and Darwin's Origin of Species lit a fire in him that never went out. Over his long career Galton drew the first modern weather maps and identified the anticyclone, devised the classification system that made fingerprints usable for identification, and — most consequentially for this story — helped invent the mathematical machinery of modern statistics.
His obsession was variation: why do individuals differ, and can difference be measured? He was captivated by the Belgian statistician Adolphe Quetelet, who had shown that human traits like height, tallied across a population, arrange themselves into the same bell curve astronomers used to describe errors of measurement. If nature scatters people around an average the way a careless hand scatters shot around a bullseye, Galton wanted a way to show it.
"I know of scarcely anything so apt to impress the imagination as the wonderful form of cosmic order expressed by the Law of Frequency of Error. … It reigns with serenity and in complete self-effacement amidst the wildest confusion."
— Francis Galton, Natural Inheritance, 1889
Galton also coined the word "eugenics" and was its founding advocate. Several of his statistical inventions, the quincunx included, were built partly in service of a project to rank and "improve" human heredity — ideas later used to justify grave harms. The mathematics he uncovered is sound and universal; the social program he attached to it is not. This page keeps both in view rather than picking one.
The board's likely ancestor is bagatelle, the parlor game where a ball is launched into a slope studded with pins — the same lineage that later gave us the pinball machine. Galton took that familiar toy and turned it into an argument.
He commissioned the first quincunx in 1873 from the London instrument makers Tisley and Spiller: a framed board with a glass face, a fixed funnel at the top, and roughly two dozen staggered rows of pins. On 27 February 1874 Galton presented it to the Royal Institution as part of a lecture on the "law of deviation from an average," using it to explain how the heights of people across Britain distribute themselves around a mean.
The name comes from Latin. A quincunx was originally a Roman coin worth five-twelfths of an as, marked with five dots — the pattern of the "five" on a die: one in each corner, one in the middle. Gardeners still use the word for trees planted in that staggered arrangement, which is exactly how the pins on Galton's board are laid out. Each pin sits in the gap between the two above it, so a bead striking it has a clean choice of falling left or right.
The board works because it physically performs a piece of probability theory. Picture a single bead. At each row of pins it bounces left or right with roughly equal chance. Its final resting column is decided entirely by how many times it happened to go right on the way down.
Going right zero times lands it at the far left; going right every time lands it at the far right — and, crucially, there are vastly more mixed routes that leave it somewhere near the middle. Count those routes and you rediscover Pascal's triangle: the number of distinct paths to each bin is a binomial coefficient. A bead can reach a central bin thousands of different ways, but only one lonely path reaches each extreme edge. So the bins fill in proportion to the binomial distribution — tall in the middle, thin at the sides.
Now add more rows and more beads. A theorem proved by Abraham de Moivre and Pierre-Simon Laplace in the eighteenth century says that as the number of rows grows, the binomial distribution smooths into the continuous normal, or Gaussian, curve. The board is a working demonstration of the central limit theorem: when you add together many small, independent random nudges — and each bounce off a pin is exactly such a nudge — their sum tends toward a bell curve, almost regardless of the details. Galton was showing his audience that order isn't the opposite of randomness; for large numbers, order is what randomness reliably produces.
Galton didn't stop at the demonstration — he wanted the machine to help him think, specifically about heredity. If tall parents tend to have tall children, and the tendency compounds each generation, why doesn't the population fly apart into giants and dwarves?
Around 1877 he conceived a two-stage quincunx. Beads first fall through one set of pins to form a normal distribution. Then the shot from a single narrow slice is released to fall through a second set of pins. Galton found that each slice spreads back out into its own little bell, and that all those little bells, added together, reconstruct exactly the original wide distribution — a mechanical picture of how inherited variation is passed along and yet held in equilibrium.
Galton's earlier sweet-pea experiments (1875) had already hinted at it: heavy mother-seeds gave heavy daughter-seeds, but the offspring drifted back toward the average weight. Plotted, the relationship fell along a straight line of slope less than one — the first regression line.
That word — "regression" — is Galton's, born from this work, and it still labels one of the most-used tools in all of science. His younger colleague Karl Pearson later put the ideas on rigorous mathematical footing, formalizing the correlation coefficient in 1896. The line from a box of rattling shot to the regression models running in every modern lab and spreadsheet is remarkably direct.
The quincunx never left the stage. Because it turns an equation into something you can watch happen, it became a favorite of museums and science communicators.
Charles and Ray Eames built large mechanical versions for their Mathematica: A World of Numbers… and Beyond exhibition beginning in 1961; those "probability machines" appeared at the Boston Museum of Science, the Henry Ford Museum, and the IBM Pavilion at the 1964–65 New York World's Fair. One original Galton quincunx survives in the science collections of University College London, the institution Galton endowed.
Inventors bent the board to other shapes, too. The Dutch astronomer Jacobus Kapteyn built a variant whose triangular pins multiply rather than add the sideways displacements, producing a skewed log-normal curve — a better model for many biological quantities that grow proportionally. And it migrated into finance: the random, unpredictable path of a single bead is a vivid analogue for a "random walk" in prices, the idea that short-term market moves resist prediction even as the aggregate distribution stays orderly. That's why polished desktop Galton boards still turn up in the lobbies of investment firms.
More than 150 years after Galton first poured shot into a glass box, the lesson holds its charm: individual events can be wholly unpredictable while their collective pattern is almost perfectly lawful. The Galton board is, in the end, a small machine for making peace with that paradox — which is exactly why this course borrowed it.
The simulation above is an idealized model — independent 50/50 bounces — and shows the statistical behavior, not the exact physics of a real board.
Everything above is the real device. The course itself keeps things simpler on purpose — one toy, fourteen lessons, no history required.