What a probability means
Saying a fair coin lands heads with probability 1/2 is not a promise about the next flip. It describes the long run: over many flips, about half come up heads. A fair die gives each face, 1 to 6, the same chance of 1/6. The average of a roll is 3.5, a number the die can never show, yet the average of many rolls gets closer and closer to it.
Each flip and each roll is independent: what happened before changes nothing. That is why any particular sequence of 5 flips, such as heads five times, has the same chance as any other: 1 in 32.
The law of large numbers
The first version of the law of large numbers was proved by the Swiss mathematician Jacob Bernoulli and published in 1713, in his book The Art of Conjecturing; he called it his “golden theorem”. It says that the average of many independent trials gets closer and closer to the expected value: the share of heads tends towards a half, the average of a die towards 3.5.
The law only speaks of large numbers. There is no rule that a few flips will match the expected share, or that a streak will soon be “balanced” by the opposite result. In fact the gap between the number of heads and the number of tails usually grows as you keep flipping; it is only the gap as a share of all the flips that shrinks. The same law lets a casino, which can lose any single spin of a roulette wheel, count on a predictable share of what is bet, and lets insurers plan for many customers at once.
Why 7 is the commonest total of two dice
Two dice can land in 6 × 6 = 36 different pairs, all equally likely. The totals are not. Only one pair makes 2 (1 and 1) and only one makes 12 (6 and 6), but six pairs make 7: 1 and 6, 2 and 5, 3 and 4, and the same three the other way round. So a total of 7 turns up 6 times in 36, a 2 only once in 36.
The more random pieces are added together, the more their totals crowd towards the middle. With two dice the chart of totals is a triangle; add more dice, or more left-or-right bounces, and the shape rounds into a bell.
The Galton board
The English scientist Francis Galton presented his board at the Royal Institution in London on 27 February 1874. Beads fall through rows of pegs, bounce left or right at each one, and collect in bins at the bottom, where the piles form a bell shape. The number of paths that lead to each bin follows Pascal’s triangle: with 4 rows the bins can be reached in 1, 4, 6, 4 and 1 ways, out of 16 in all.
With a fair bounce at each peg, the chance of each bin is given by the binomial distribution, and with many rows and many beads its shape approaches the smooth normal curve. Abraham de Moivre first worked out that curve in 1738, thinking about the number of heads in 3,600 coin tosses. Tilting the pegs so beads go right more often shifts the whole pile. Galton also used his board to argue for eugenics, ideas that have long been rejected; what the board really shows is how much spread pure chance produces on its own.
Centre and spread
For n rows of pegs, or n coins, with a chance p of going right, or of heads, the average lands at n × p and the typical spread around it, the standard deviation, is the square root of n × p × (1 − p). A board with 10 rows and fair pegs centres on bin 5 with a spread of about 1.58 bins. For de Moivre’s 3,600 coins it is 1,800 heads on average, give or take about 30.
The spread grows only like the square root of the number of tries. Toss 100 times as many coins and the spread in the number of heads is 10 times larger, but as a share of all the tosses it is 10 times smaller: the law of large numbers seen from another angle.
Three common mistakes
“After five heads, tails is due.” This is the gambler’s fallacy, also called the Monte Carlo fallacy after a night in 1913 when a roulette wheel at the Monte Carlo Casino came up black 26 times in a row. The coin has no memory: the next flip is still heads with probability 1/2. Before any flip, five heads in a row and four heads followed by a tail are equally likely, 1 in 32 each.
“Chance evens out quickly.” Only the average settles, and only over many tries. Over a few flips anything can happen, and the difference between heads and tails tends to get bigger, not smaller.
“All totals of two dice are equally likely.” Each pair is, but the totals are not: 7 can be made in six ways, 2 and 12 in one way each.
Questions visitors ask
What is the law of large numbers?
It says that the average result of many independent trials gets closer to the expected value as the number of trials grows. The share of heads in many coin flips approaches 1/2, and the average of many dice rolls approaches 3.5. It says nothing about the next single result.
After five heads in a row, is tails more likely?
No. Each flip is independent, so the chance of heads is still 1/2. Believing otherwise is called the gambler’s fallacy. Five heads in a row is unusual before you start, 1 chance in 32, but so is any other particular sequence of five flips.
Why is 7 the most common total with two dice?
Because more pairs add up to 7 than to any other total. Of the 36 equally likely pairs, six make 7 (1+6, 2+5, 3+4, 4+3, 5+2, 6+1), while only one makes 2 and only one makes 12.
What does a Galton board show?
That many small random choices add up to a predictable pattern. Each bead bounces left or right at random, but the piles at the bottom follow the binomial distribution, which looks more and more like the bell-shaped normal curve as the rows increase.
Can a casino lose money?
On a single spin, yes. Over a very large number of bets, the law of large numbers makes its earnings tend towards a predictable percentage: any winning streak by a player is eventually overcome by the rules of the game.
Sources and further reading
- Law of large numbers — Wikipedia
- Gambler’s fallacy — Wikipedia
- Galton board — Wikipedia
- Binomial distribution — Wikipedia
- De Moivre–Laplace theorem — Wikipedia
Cite this exhibit
Eurekarium. (2026). Chance Lab. https://eurekarium.com/probability/chance/
Last updated: