An exhibition in twenty-four rooms · History of mathematics
Great Mathematicians Who Changed the World
Twenty-four men, twenty-four ideas, and something in every room that you can move with your own hands.
By Ali Kaya · November 2026 · about 50 minutes, longer if you play

You can name a dozen of the mathematicians in this building without trying. Pythagoras, Euclid, Newton, Gauss, Turing. It is harder to say what each of them actually did. Most of us carry the names around like the labels on jars we have never opened.
This page opens the jars. There are twenty-four rooms, each with one man and the one idea I would choose if he could keep only one. Each idea comes with something to handle. You can cut a rectangle into Euclid’s squares, fire Newton’s cannon until the ball stops coming down, add up Fourier’s waves, step Turing’s machine along its tape and watch Conway’s four rules build a factory. The rooms run in order, from a Greek island around 530 BC to an IBM laboratory in 1980, and the ideas hand on to one another as you go.
It is the companion to Great Women in Science and Math Who Changed the World, which is built the same way. I have kept the two apart on purpose. For most of the twenty-five centuries covered here a woman could not attend a university, and the other wing tells what it took to do mathematics anyway. Several people walk through both.
Abakcus is reader-supported. The book links on this page are Amazon affiliate links: if you buy through one, I earn a small commission at no cost to you.
The idea · c. 530 BC
The square on the long side equals the other two together
Builders in Babylon were using the rule a thousand years before Pythagoras was born. A clay tablet from about 1800 BC lists fifteen sets of whole numbers that make right angles, some of them in the thousands. What the scribes had was a reliable trick. What the Greeks added, and tradition credits to Pythagoras and the secretive community he founded at Croton in southern Italy, was the claim that it is true of every right triangle that could ever exist, together with a reason why.
Nobody knows which proof they had. The one below is among the oldest and needs no algebra. Put four copies of the same triangle inside a square. Arranged one way they leave a single tilted square in the middle, built on the triangle’s long side. Slide them into the corners and they leave two smaller squares, one on each short side. The outer square has not changed and neither have the triangles, so the leftover areas must be equal.
Try itChange the triangle. The two sides of the picture never disagree.
The Pythagoreans believed that everything was made of whole numbers and their ratios. They had found that strings in the ratios 2 to 1 and 3 to 2 sound sweet together, and thought the planets must work the same way. Their own theorem undid them. Apply it to a square with sides of 1 and the diagonal comes out as a number that cannot be written as any ratio at all. Legend says the man who let that secret out was drowned at sea.
The idea · c. 300 BC
All of geometry from five assumptions
Around 300 BC, in the new city of Alexandria, Euclid gathered up three centuries of Greek mathematics and rebuilt it as a single structure. He began with twenty-three definitions and five assumptions so plain that nobody could object, such as that a straight line can be drawn between any two points. From those he proved 465 propositions, each resting only on the ones before it. Nothing is taken on trust and nothing is out of order.
The Elements became the most successful textbook ever written. It was copied by hand for eighteen centuries and has gone through more than a thousand printed editions, and some of the most beautiful of them can still be read for free. It is not all geometry. Book VII contains a method for finding the largest number that divides two others, and it is the oldest algorithm still in daily use: your browser ran a descendant of it to set up the encrypted connection to this page. The idea is to cut the biggest possible square off a rectangle, then do the same to what is left, and keep going until the last piece is a square itself.
Try itSet two lengths. The side of the last, smallest square is the answer.
Two books later he proved that the primes never run out, in an argument short enough to memorise. What really lasted was the method. Spinoza laid out his ethics in Euclid’s format, Newton did the same for physics, and Abraham Lincoln worked through the first six books as a young lawyer in order to learn what it means to demonstrate something. Only the fifth assumption, the one about parallel lines, ever looked less than obvious. It comes back in Lobachevsky’s room.
The idea · c. 250 BC
Trap a number you cannot reach between two you can
The ratio of a circle’s circumference to its diameter is the same for every circle, and nobody can write it down exactly. Earlier estimates came from measuring. Archimedes replaced measuring with a proof. A polygon drawn inside a circle has a shorter outline than the circle, and a polygon drawn around it has a longer one, and the outline of a polygon can be calculated.
He started with hexagons and doubled the number of sides four times, to 96, working out each new perimeter by hand in a number system with no zero and no decimals. The result was a guarantee that π lies between 3 and 10/71 and 3 and 1/7. The number mattered less than the method. He had found a way to trap an unreachable quantity between two reachable ones and squeeze. He used the same squeeze to find the area under a parabola and the volume of a sphere, and nineteen centuries later it grew into calculus.
Try itDouble the sides and watch the two fences close in.
He was also the best engineer of the ancient world. He worked out the law of the lever and why things float, and his war machines held off a Roman fleet at Syracuse for two years. When the city fell in 212 BC a soldier found an old man drawing figures in the sand and killed him, against orders. He had asked for his favourite result on his tomb: a sphere inside a cylinder, with the ratio of their volumes, two to three.
The idea · c. 820
One recipe solves every equation of its kind
In ninth-century Baghdad the caliph al-Ma’mun kept a library and academy called the House of Wisdom, and one of its scholars, Muhammad ibn Musa al-Khwarizmi, was asked for a practical book on calculation. It was meant for dividing inheritances, settling lawsuits, measuring land and digging canals. He delivered something larger than he was asked for.
Earlier mathematicians had solved particular problems one at a time. Al-Khwarizmi sorted equations into types and gave, for each type, a fixed sequence of steps that always works, with a geometric picture to show why. His standard example is a square and ten of its roots that together equal thirty-nine. Draw the square, attach the ten roots as two strips of five, and the figure is a larger square with one corner missing. Fill in the corner, which is five by five, and the whole is sixty-four, so its side is eight and the unknown is three. He did all of this in words. There is not a single symbol in the book.
Try itChange the numbers in the equation. The same four steps always land on the answer.
One word in the book’s title, al-jabr, meaning something like “restoring”, became algebra. A second book of his explained how to write any number with nine digits and a zero. When it was translated into Latin it opened with the words Dixit Algorizmi, “so said al-Khwarizmi”, and his name turned into the word for any step-by-step procedure. An algorithm is, literally, the thing he did.
The idea · 1202
Ten digits, and a rule about rabbits
Leonardo of Pisa grew up in a trading post on the Algerian coast, where his father ran the customs house for Pisan merchants. The local traders did their sums with nine figures and a sign for zero, written with a pen. Back in Italy merchants were still pushing counters around a board and recording the results in Roman numerals, which cannot be multiplied on paper at all.
In 1202 he wrote the Liber Abaci, the Book of Calculation, to change that. It is a thick manual of the new arithmetic, with hundreds of worked problems on currency conversion, profit margins and interest. Within a few generations every Italian trading city had schools teaching from it, and the numerals you are reading now had won. Buried in chapter twelve is a puzzle about rabbits. A pair breeds a new pair every month, and each new pair starts breeding a month after birth. How many pairs are there after a year? Each month’s total is the sum of the two before it: 1, 1, 2, 3, 5, 8, 13.
Try itAdd squares. The ratio of each number to the one before settles toward 1.618, the golden ratio.
He thought so little of the puzzle that he never came back to it. The sequence was named after him six centuries later, and has since turned up in the spirals of sunflower heads, the family trees of bees, computer algorithms and a bookshelf. The book that carried it is the reason Europe could do the arithmetic that everything else on this page needs.
The idea · 1637
Every equation is a shape, and every shape an equation
Until the seventeenth century geometry and algebra were separate subjects. Geometry was about figures, was done with ruler and compass, and was thought the more respectable. Algebra was about unknown quantities and was a tool for accountants. A problem belonged to one or the other.
In 1637 René Descartes published a book on how to reason, the Discourse on Method, and attached three appendices to show the method at work. The third, La Géométrie, joined the two subjects for good. Fix two lines at right angles, and every point on a plane has an address made of two numbers. A curve is then the set of all the points whose two numbers satisfy some equation. Any question about shapes can be turned into algebra, worked out by calculation, and turned back. The story that he thought of it while watching a fly cross his ceiling is probably too good to be true. Pierre de Fermat, in the next room, had the same idea at the same time and did not publish it.
Try itChange the numbers in the equation and the shape follows. Change the kind of equation and you get a different family of shapes.
The same appendix gave us the habit of writing unknowns as x, y and z, known quantities as a, b and c, and powers as small raised numbers. Without coordinates there is no graph, no calculus, no map on a phone and no pixel on a screen. In 1649 Queen Christina of Sweden hired him as her philosophy tutor and set the lessons for five in the morning. He caught pneumonia and died that winter.
The idea · 1637
Two cubes never add up to a cube
Pierre de Fermat was a judge in Toulouse who did mathematics for pleasure and sent his discoveries to friends as challenges, usually without the proofs. He read his copy of the ancient Arithmetica of Diophantus with a pen in his hand. Beside a problem about splitting a square number into two squares, which can be done in endless ways, he wrote that the same thing is impossible for cubes, for fourth powers, and for every power above the second. He added that he had found a truly marvellous proof, which the margin was too narrow to hold.
He never wrote it out anywhere else. His son published the note after his death, and for the next three and a half centuries the best mathematicians alive tried to supply the proof. Euler managed cubes. Sophie Germain, whose room is in the other wing, found the first approach that covered many powers at once. In 1994 Andrew Wiles, who had read about the problem in a library book at the age of ten, finished it after seven years of working alone in his attic. The proof runs to more than a hundred pages and uses mathematics that did not exist before the twentieth century. Whatever Fermat had in mind, it was not that.
Try itAt the second power the grid is dotted with solutions. Move to the third and it goes blank. A few cells miss by one, and no cell ever hits.
Almost everything else he claimed turned out to be right. He shares the invention of coordinate geometry with Descartes and of probability with Pascal, and he found tangents and maxima by a method that was calculus in all but name. A smaller result of his, about remainders after dividing by a prime, now sits inside the encryption that protects card payments.
The idea · 1654
Chance obeys arithmetic
In the summer of 1654 a gambler asked Blaise Pascal a question. Two players are in the middle of a game, the first to win a set number of rounds takes the pot, and the game is interrupted. How should the money be divided? Pascal wrote to Fermat, and over a few weeks of letters the two of them solved it. The stake should be split according to each player’s chance of winning from that point on, and the chance can be found by counting all the ways the rest of the game could go.
It was the first time anyone had treated the future as something to be calculated. Insurance, statistics, weather forecasts, clinical trials and dropping needles to find π all descend from those letters. To do the counting Pascal used a triangle of numbers in which each entry is the sum of the two above it. Chinese and Persian mathematicians had known the triangle for centuries, but he was the one who showed how much it holds. Row by row it gives the number of ways to choose some things out of many, which is the number of ways a run of coin tosses can fall.
Try itWith few rows you can read the numbers. Add rows and shade only the odd ones, and a pattern appears that nobody put there.
He had already built one of the first mechanical calculators at nineteen, to spare his father, a tax commissioner, the endless addition. He proved that the air has weight by sending a barometer up a mountain. Then, on a November night that same year, he had a religious experience so intense that he wrote it down and sewed the paper into the lining of his coat. He gave the rest of his short life to theology and died at thirty-nine.
The idea · 1687
The Moon is falling
Everyone knew that things fall. Everyone knew that the Moon goes round the Earth. Nobody before Isaac Newton thought they were the same fact. The picture he used to explain it is a cannon on a very high mountain. Fire it level and the ball curves down and lands some way off. Fire it harder and it lands further away. Fire it hard enough and the ground, which is round, curves away beneath the ball as fast as the ball drops. It keeps falling and never lands. That is an orbit, and the Moon is doing it.
The Principia of 1687 turned the picture into mathematics. It gave three laws of motion and one law of gravity, and from those four statements it derived the paths of the planets, the tides, the bulge of the Earth and the return of comets. To do it he needed a way of calculating with quantities that change from instant to instant. He had invented one, which he called fluxions and we call calculus, in 1665 and 1666, when the plague shut Cambridge and sent him home to his mother’s farm. He told almost nobody for twenty years.
Try itRaise the speed a notch at a time. There is one setting where the ball comes all the way round.
That silence caused the bitterest quarrel in the history of science, which is the subject of the next room. He was secretive, vindictive and quite possibly the most capable mind there has ever been. When Johann Bernoulli set a problem meant to stump the mathematicians of Europe, Newton solved it in an evening after work and sent the answer in unsigned. Bernoulli said he knew the lion by his claw. The first full French translation of the Principia, still the standard one, was made by Émilie du Châtelet, who has a room of her own in the other exhibition.
The idea · 1684
An area is a sum of infinitely thin strips
Gottfried Wilhelm Leibniz was a diplomat, librarian, philosopher and lawyer who took up serious mathematics at twenty-six and within four years had calculus. He reached it by a different road from Newton. Newton thought about motion. Leibniz thought about sums. The area under a curve, he saw, is what you get by slicing it into strips, adding up the strips, and letting them become infinitely thin and infinitely many.
On a manuscript page dated 29 October 1675 he wrote a tall, stretched letter S, for summa, to stand for that sum. A few days later he added dx for the width of a strip. He published in 1684, twenty years after Newton’s private discovery and three years before Newton printed anything. His notation was so well made that the rules of calculus almost work themselves, and it is the one every student still learns. It is one of the mathematical symbols that changed what people were able to think.
Try itAdd strips. The sum closes in on the true area, and in the limit it is the area.
Newton’s supporters accused him of theft. In 1712 the Royal Society appointed a committee to settle the matter, and it found for Newton. Its report was written, in secret, by the Society’s president, who was Newton. Historians now agree the two men worked independently. British mathematicians stayed loyal to Newton’s clumsier dots for a hundred years and fell behind the continent. Leibniz also built a machine that could multiply, and worked out arithmetic in base two, using only 0 and 1, long before anything existed that could use it.
The idea · 1748
Growth, rotation and waves are one formula
Three numbers kept turning up in eighteenth-century mathematics. One was π, from circles. One was e, about 2.718, which appears wherever something grows in proportion to its own size, like money at compound interest. The third was the square root of minus one, a number that could not exist and that algebra could not do without. They seemed to have nothing to do with one another.
Leonhard Euler showed they were parts of one thing. In a textbook of 1748 he proved that raising e to an imaginary power does not produce growth at all. It produces rotation: the result is a point travelling around a circle, with the cosine of the angle as its horizontal position and the sine as its vertical. Go halfway round and you arrive at minus one. Written out, that is e to the power iπ, plus one, equals zero, which ties together the five most important numbers in mathematics. Every calculation about waves, alternating current or quantum mechanics goes through this formula. The unit circle is where it lives.
Try itTurn the angle. Stop at 180 degrees.
He wrote more mathematics than anyone before or since. There are more than 850 works, and the collected edition, begun in 1911, has passed eighty volumes. He started graph theory by proving that nobody could walk across all seven bridges of Königsberg exactly once, and he gave us the letters e, i and π in their modern use, along with f(x) and Σ. He lost the sight of one eye in his twenties and of the other at fifty-nine, said it left him with fewer distractions, and went on working for seventeen years, doing the calculations in his head.
The idea · 1801
Arithmetic on a clock
Carl Friedrich Gauss was the son of a bricklayer in Brunswick. The story goes that at seven his schoolmaster set the class to add up the numbers from 1 to 100, to keep them quiet, and Gauss wrote the answer down at once. He had seen that the numbers pair off into fifty sums of 101. At nineteen he showed how to construct a regular polygon of seventeen sides with ruler and compass, the first advance in that subject since the Greeks, and decided to become a mathematician.
At twenty-four he published the Disquisitiones Arithmeticae, which turned the study of whole numbers from a collection of curiosities into a science. Its first page introduces a new kind of equality. On a clock, four hours after ten is two, and nobody finds that strange. Gauss took the idea seriously for clocks of any size, wrote it with a three-bar sign, and showed that the arithmetic of remainders has laws as strict as ordinary arithmetic and far richer. Every card number you type is checked with it, and so is every secure connection.
Try itPut the numbers round a clock and join each to its double. Then change the multiplier, and the size of the clock.
There was very little he did not touch. He found the lost asteroid Ceres from a handful of observations, invented the method of least squares to do it, and gave his name to the bell curve and the integral under it. He also worked out, and kept to himself for fear of ridicule, a geometry in which Euclid’s parallel postulate is false. His motto was pauca sed matura, few but ripe, and it left others to make the discoveries he had left in his desk drawer.
The idea · 1822
Any wave is a sum of simple waves
Joseph Fourier wanted to know how heat spreads through a metal bar. In 1807 he presented the Paris Academy with an answer that depended on a claim so bold that the examiners refused to publish it. Any pattern at all, he said, however jagged, can be built by adding together smooth waves of different frequencies, if you take the right amount of each.
The examiners included Lagrange and Laplace, and they could not believe that a curve with sharp corners could be made from sine waves, which have none. Fourier could not quite prove it either. He was right all the same, and the work finally appeared in 1822. The idea reaches far beyond heat. A sound is a sum of pure tones, an image is a sum of ripples of brightness, and any signal can be broken into its frequencies and put back together. I have written about how the Fourier transform works at more length. Every MP3, JPEG, MRI scan and mobile phone call runs on it.
Try itAdd waves one at a time. The dashed square wave has corners and the sum has none, yet it gets as close as you like.
He was a tailor’s son, orphaned at nine and arrested twice during the Revolution. In 1798 he sailed to Egypt with Napoleon as one of the expedition’s scholars, and was left to govern half the country. He came back convinced that heat was good for the health and lived swaddled in overcoats in overheated rooms. In 1824, thinking about why the Earth is warmer than it should be at its distance from the Sun, he described what we now call the greenhouse effect.
The idea · 1832
An equation’s symmetries decide whether it can be solved
Every student learns the formula that solves quadratic equations. Renaissance Italians found formulas for the cubic and the quartic, both far longer. Then, for three hundred years, nobody could find one for equations of the fifth degree. In 1824 the young Norwegian Niels Abel proved that no such formula exists. That left a deeper question open. Some fifth-degree equations can be solved, and others cannot. What decides?
Évariste Galois answered it before he was twenty. He looked at the solutions of an equation as a set and asked in how many ways they can be shuffled among themselves without breaking any of the relations between them. Those shuffles form a structure he called a group. An equation can be solved by a formula exactly when its group can be taken apart into simple pieces in a certain way, and above the fourth degree the group is usually too tangled. The answer had nothing to do with the numbers in the equation and everything to do with its symmetry.
Try itThe eight symmetries of a square are a small group. Turn and then flip, or flip and then turn, and see whether the order matters.
Hardly anyone understood him. He failed the entrance examination for the École Polytechnique twice. The Academy lost one of his papers and rejected another as unreadable. He was a fierce republican, was expelled from his college and jailed. On 30 May 1832 he fought a duel with pistols, for reasons still unclear, and died the next day. He had sat up the night before writing out his results in a letter to a friend. It was published fourteen years later. Group theory is now the language of particle physics, crystallography and chemistry.
The idea · 1829
A geometry where Euclid’s parallels fail, and nothing breaks
Four of Euclid’s five assumptions take a line each. The fifth takes a paragraph. It amounts to saying that through a point not on a line there is exactly one line that never meets it. For two thousand years mathematicians suspected that this was not really an assumption, and tried to prove it from the other four. All of them failed.
Nikolai Lobachevsky, at the remote Russian university of Kazan, tried the opposite. Suppose that through the point there are many lines that never meet the first. He followed the consequences, waiting for a contradiction, and none came. He got a complete and consistent geometry. In it the angles of a triangle add up to less than 180 degrees, and the bigger the triangle the less they add up to. He published in 1829. The Hungarian János Bolyai reached the same result three years later without knowing of him, and Gauss, as his room explains, had been sitting on it for years.
Try itThe disc is a map of the whole infinite plane of his geometry. Small triangles look almost ordinary. Stretch one toward the edge.
It meant that Euclid’s geometry is not a necessary truth. It is one possibility among several, and which one describes real space is a question for measurement. Lobachevsky’s colleagues thought the work absurd and a St Petersburg journal mocked it. He was rector of his university for nineteen years, was dismissed without explanation in 1846, and dictated his last book after he had gone blind. Sixty years after his death Einstein showed that space near a mass is curved.
The idea · 1859
The primes keep time to a hidden music
The primes look as if they were scattered at random. There are four below ten and twenty-five below a hundred, and the gaps between them follow no rule anyone can see. Yet when the young Gauss counted them in bulk he found that their overall thinning-out follows a smooth curve with uncanny accuracy. The question was how the two facts fit together.
In 1859 Bernhard Riemann, newly elected to the Berlin Academy, sent in an eight-page paper, the only one he ever wrote on whole numbers. It takes a function that Euler had studied, the zeta function, and extends it to complex numbers. Riemann showed that the exact positions of the primes are encoded in the points where this function equals zero. Each zero contributes a wave, and the waves added together turn Gauss’s smooth curve into the true jagged staircase. He remarked, in passing, that all the zeros that matter seem to lie on a single straight line, and that he had looked for a proof and put the search aside.
Try itThe staircase goes up one step at every prime. Extend the range and watch how closely the smooth blue curve keeps to it.
That remark is the Riemann hypothesis. It has been checked for the first ten trillion zeros and has never been proved, and a million-dollar prize is waiting for whoever does. There are more strange facts about primes than one page can hold. Five years earlier he had given a lecture on curved spaces of any number of dimensions, with the elderly Gauss in the audience. It became the mathematics of general relativity. He died of tuberculosis beside Lake Maggiore at thirty-nine.
The idea · 1874
Some infinities are larger than others
Before Georg Cantor, infinity was a manner of speaking. Numbers go on for ever, and that was all there was to say. He asked a question nobody had thought could be asked. Are all infinite collections the same size?
Two collections are the same size if their members can be paired off with none left over. By that test the even numbers are exactly as numerous as all the whole numbers, since each number pairs with its double. So are the fractions, which is more surprising. Then, in 1874, Cantor proved that the decimals cannot be paired with the whole numbers, however cleverly you try. His later version of the argument fits in a paragraph. Suppose someone hands you a numbered list that claims to contain every decimal. Go down the diagonal, changing the first digit of the first number, the second digit of the second, and so on. The number you have made differs from every entry on the list in at least one place. The list was not complete, and no list can be.
Try itBuild the number that the list missed. Then try it on another list.
A list that claims to contain every decimal
- 10.91559780…
- 20.87848446…
- 30.36433739…
- 40.77760879…
- 50.30228987…
- 60.61713867…
- 70.84342582…
- 80.48417062…
So there is more than one infinity, and in fact there is an endless ladder of them, each larger than the last. The reaction was savage. His former teacher Leopold Kronecker called him a corrupter of youth and blocked his career, and Poincaré, in the next room, thought the theory a disease from which mathematics would recover. Cantor suffered repeated breakdowns and died in a sanatorium in 1918. By then his theory of sets was becoming the foundation on which all of mathematics is built. “No one shall expel us,” David Hilbert said, “from the paradise that Cantor has created.”
The idea · 1890
A tiny difference at the start can change everything after
Newton’s laws give the motion of two bodies, a sun and a planet, exactly and for ever. Add a third body and nobody could find a formula. In 1887 King Oscar II of Sweden offered a prize for a solution, in a competition organised by Gösta Mittag-Leffler, who also turns up three times in the women’s exhibition. Henri Poincaré won it with a memoir of 160 pages.
While it was at the printer, an editor asked about a passage he could not follow. Poincaré went back to it and found that he had made a serious error. He paid to have the printed copies destroyed, which cost him more than the prize was worth, and rewrote the paper. The corrected version contained something nobody had seen before. The orbits of three bodies can be so intricately tangled that two starting positions differing by a hair lead, in time, to completely different futures. The motion is governed by exact laws and still cannot be predicted in practice.
Try itThe two pendulums start a twentieth of a degree apart. Watch how long they agree, then release them again.
He drew no great attention to it, and the idea slept until the 1960s, when a meteorologist with a computer rediscovered it and it was given the name chaos. Poincaré went on to found topology, the study of shape without measurement, and left a question about three-dimensional spaces that stood for a century until Grigori Perelman settled it in 2003 and declined the million dollars. He is often called the last person to have understood the whole of mathematics.
The idea · 1891
A line can fill a square
A line has length and no width. A square has area. They are different kinds of thing, one-dimensional and two-dimensional, and common sense says that no line, however long or twisted, could cover a square. In 1890 Giuseppe Peano showed by pure algebra that one can. A year later David Hilbert drew it.
His recipe is to divide the square into four and visit the quarters in a U-shaped path. Then divide each quarter into four and do the same inside it, joining up the ends, and repeat without end. At every stage you have an honest curve that never crosses itself. In the limit it passes through every single point of the square. The result forced mathematicians to admit that they did not really know what “dimension” meant, and the effort to say so precisely opened a new field. The curve itself is now used in databases and map software, because points that are near each other on the line stay near each other in the square.
Try itRefine it a step at a time. It is one unbroken line at every stage.
That was a minor work by his standards. In 1900 he stood up in Paris and listed twenty-three unsolved problems that he thought should occupy the coming century, and for the most part they did. He rebuilt Euclid’s geometry on sounder foundations, created the spaces in which quantum mechanics is written, and made Göttingen the centre of the mathematical world. He also fought his own faculty to get Emmy Noether a post, as her room tells. His creed was that every mathematical question has an answer. “We must know. We will know,” he said in 1930, and it is carved on his grave. The next room is about what happened to that hope.
The idea · 1931
Some true statements can never be proved
Hilbert’s programme, in the 1920s, was to put all of mathematics on a finite list of axioms and rules, and then to prove two things about the list. It should be complete, so that every true statement can be derived. And it should be consistent, so that no false one can. In September 1930, at a conference in Königsberg, a twenty-four-year-old from Vienna mentioned quietly in a discussion that he had shown this could not be done. Hilbert gave his “We must know” speech in the same city the next day.
Kurt Gödel’s method was to make arithmetic talk about itself. He gave every symbol a code number, and showed how to combine the codes so that every formula, and every proof, becomes a single enormous whole number. Statements about proofs then turn into statements about numbers, which arithmetic can express. With great care he built a formula that says, in effect, “this statement cannot be proved”. If it could be proved, the system would be proving something false. So it cannot be proved, and therefore what it says is true. Any system strong enough to do arithmetic contains true statements that it can never reach.
Try itPick a statement and watch it turn into one number. The whole proof rests on this trick.
The statement “zero equals zero”
Multiplied together, its Gödel number
243,000,000
Factor the number back into primes and the statement can be read off again, symbol by symbol.
A second theorem followed: no such system can prove its own consistency. He fled Vienna in 1940 by way of Siberia and Japan and settled in Princeton, where Einstein said he went to his office mainly for the privilege of walking home with Gödel. He grew ever more afraid of being poisoned and would eat only what his wife had prepared and tasted. When she was taken to hospital in 1977 he stopped eating, and he died weighing sixty-five pounds.
The idea · 1936
One machine can do what any machine can do
One of Hilbert’s questions was left standing after Gödel. Is there a mechanical procedure that could take any mathematical statement and decide, without fail, whether it can be proved? To answer it, somebody first had to say exactly what a “mechanical procedure” is. In 1936 a twenty-three-year-old Cambridge graduate named Alan Turing did.
He imagined the simplest device that could be said to calculate. It has a paper tape divided into squares, and a head that looks at one square at a time. The head can read the symbol there, write a new one, and move a square left or right. What it does is fixed by a short table of rules and by which of a few states it is in. That is all. Turing argued that anything a human clerk could work out by following rules, this machine could work out too. Then he showed that a single machine of this kind, given the right table, can imitate every other. The rules of the machine to be imitated are simply written on its tape. This is the stored-program computer, described nine years before one was built. And he proved that there are questions no such machine can ever settle, which answered Hilbert: no.
Try itThis machine has two states and five rules, and it adds one to a binary number. Step through it, then let it run.
- state
- seek
- rule used
- Start at the left of the number.
- tape reads
- 11 in binary, after 0 steps
During the war he led the work at Bletchley Park that broke the German naval Enigma, which by most estimates shortened the fighting by years. Afterwards he designed one of the first real computers and proposed a test for whether a machine could be said to think. In 1952 he was convicted of gross indecency for a relationship with another man and accepted hormone treatment as the alternative to prison. He died of cyanide poisoning in 1954, aged forty-one. The British government apologised in 2009.
The idea · 1948
Information can be measured
In the 1940s the engineers who built telephone lines, radio links and telegraph cables each had their own rules of thumb. Nobody had asked what all those systems were actually carrying. Claude Shannon, a mathematician at Bell Labs, published the answer in 1948 in a paper called “A Mathematical Theory of Communication”.
A message, he said, is a choice among possibilities, and its information is the amount of uncertainty it removes. The natural unit is the choice between two equally likely alternatives, and he called it the bit. A fair coin toss carries one bit. A coin that nearly always lands heads carries much less, because you could have guessed. From that definition came a formula for how far any message can be compressed, and a second and more startling result. Every channel, however noisy, has a definite capacity, and below that rate messages can be sent with as few errors as you please. Engineers had assumed that noise always wins in the end.
Try itBias the coin. The less surprising each toss becomes, the less it tells you, and the shorter the record of a thousand tosses can be made.
It took fifty years for hardware to reach the limits he had calculated, and everything digital now runs up against them. He had already shown, in a master’s thesis written at twenty-one, that the logic of true and false can be built from electrical switches, which is how every computer chip works. Then he more or less stopped. He worked out the mathematics of juggling, built a mechanical mouse that could learn a maze, and rode his unicycle down the corridors.
The idea · 1970
Four rules are enough to make a world
John Conway spent the late 1960s in the common room of the Cambridge mathematics department playing games, and insisting that he was working. One of the games used a Go board and a heap of counters. He was looking for the simplest possible set of rules that could produce behaviour too complicated to predict, and after a year and a half of tinkering over coffee he had them.
Each square of a grid is alive or dead. At every tick of the clock, a live square with two or three live neighbours stays alive, and otherwise it dies. A dead square with exactly three live neighbours comes to life. That is the whole of it. He called it Life. Martin Gardner described it in Scientific American in October 1970, and within months it was said to be using up a noticeable share of the world’s computer time. People found patterns that sit still, patterns that blink, and small ones that crawl across the board, which were named gliders. Conway offered fifty dollars for a pattern that grows without limit. A group at MIT found the glider gun you see below, which builds a new glider every thirty ticks, for ever.
Try itWatch the gun fire. Then clear the board and draw on it, or fill it with random soup and see what survives.
With guns and gliders you can build logic gates, and with logic gates you can build a computer, so this toy can calculate anything that Turing’s machine can. Nobody designed that. It was already there in four rules. Conway did deeper work, on symmetry in 196,883 dimensions and on a new kind of number, and he came to resent being known for Life. He died in April 2020, an early casualty of COVID-19.
The idea · 1980
Endless complexity from one line of arithmetic
Euclid’s shapes are smooth. Lines, circles, cones and spheres are what the first rooms of this building are made of. Almost nothing in nature looks like them. A coastline is ragged at every scale, a tree branches into smaller trees, and a cloud has no edge you could draw with a compass. For two thousand years geometry had nothing to say about any of it.
Benoit Mandelbrot spent his career on that roughness. He asked how long the coast of Britain is, and showed that the answer depends on the length of your ruler, because each finer ruler finds more inlets. In 1975 he coined a word, fractal, for shapes that look equally intricate however far you magnify them. Then, on 1 March 1980, on a computer at IBM, he plotted the picture that carries his name. Take a number, square it, add a fixed constant, and repeat. For some constants the result stays small for ever and for others it flies off. Colour the first kind black and you get the shape below. The rule fits on one line, and the boundary it draws never simplifies, at any magnification.
Try itZoom in on the edge. Keep going. Sooner or later a small copy of the whole shape turns up inside it.
Two French mathematicians, Gaston Julia and Pierre Fatou, had studied the same arithmetic in 1918 and proved deep things about it without ever seeing it, because nobody could do the millions of calculations by hand. Mandelbrot had a machine, and an eye. He had learned geometry as a Jewish boy in hiding in wartime France, and said he always thought in pictures. Fractal geometry now describes river networks, lungs, market prices and the antenna inside a mobile phone. “Clouds are not spheres,” he wrote, “mountains are not cones, coastlines are not circles.”
On the way out
What passes from room to room
Walk the rooms in order and you find that very little here was done alone. Euclid leaves one assumption looking shaky, and two thousand years later Lobachevsky pulls on it. Archimedes squeezes a circle between polygons, and Newton and Leibniz turn the squeeze into calculus. Hilbert demands that every question have an answer. Gödel shows that some cannot, Turing shows which machines could ever look for one, and Conway finds such a machine hiding in a board game. Each of these ideas was a reply to an earlier one. You can read the whole story as a history of mathematics written in the margins, or go to the sources themselves, since many of the original papers are free to read.
The other thing the rooms share is that the ideas are small enough to hold. A square cut into squares. A list with a number missing. A tape, a head and five rules. None of the machines on this page needed more than a few dozen lines to build, which says something about where the depth in mathematics comes from. It rarely comes from complication. Twenty-four more rooms, with the same kind of machines, are waiting in the women’s exhibition.
Go deeper · Ali’s picks
Five books to carry on with
Every room above has its own book. These five cover the wider story.
Journey through Genius
William Dunham
Twelve great theorems, each with its proof and the life around it. If this page left you wanting the actual mathematics, start here.
02
Men of Mathematics · E. T. Bell
Written in 1937 and unreliable on the facts, especially about Galois. It also made more people into mathematicians than any other book.
03
The Music of the Primes · Marcus du Sautoy
Gauss, Riemann, Hilbert and Turing, all chasing the same problem across two centuries.
04
Infinite Powers · Steven Strogatz
Calculus from Archimedes to GPS, by the best explainer now writing about mathematics.
05
Gödel, Escher, Bach · Douglas R. Hofstadter
Eight hundred pages on what Gödel’s theorem has to do with minds and machines. Not everybody finishes it, and nobody forgets it.
Questions people ask
- Who is the greatest mathematician of all time?
- There is no agreed answer, but four names come up more than any others: Archimedes, Isaac Newton, Leonhard Euler and Carl Friedrich Gauss. Gauss is the one most often called the “prince of mathematicians”, and Euler was the most productive, with more than 850 published works.
- Who invented calculus?
- Isaac Newton and Gottfried Wilhelm Leibniz invented it independently. Newton worked out his version in 1665 and 1666 but did not publish it for decades. Leibniz developed his in the mid-1670s and published first, in 1684. The notation used today, including the integral sign and dx, is Leibniz’s.
- Who is called the father of mathematics?
- The title is most often given to Archimedes, for the range and rigour of his work in geometry and mechanics. Euclid is called the father of geometry, al-Khwarizmi the father of algebra, and Alan Turing the father of computer science.
- Who proved Fermat’s Last Theorem?
- The British mathematician Andrew Wiles, with a key step completed together with Richard Taylor. Wiles announced a proof in 1993, a gap was found, and the corrected proof was published in 1995, about 358 years after Fermat wrote the claim in the margin of a book.
- What is the most famous unsolved problem in mathematics?
- The Riemann hypothesis, stated by Bernhard Riemann in 1859. It says that the zeros of a function called the zeta function all lie on one line, which would pin down how the prime numbers are distributed. The Clay Mathematics Institute offers one million dollars for a proof.
Portraits: Wikimedia Commons, shown in greyscale. Those of Pythagoras, Euclid, Archimedes, al-Khwarizmi and Fibonacci are later imaginings, since no likeness made from life survives. Public domain: Pythagoras (Roman bust); Euclid by Jusepe de Ribera; Archimedes by Domenico Fetti; Descartes after Frans Hals; Pascal after François II Quesnel; Newton by Godfrey Kneller; Leibniz by Christoph Bernhard Francke; Euler by Jakob Emanuel Handmann; Gauss by Christian Albrecht Jensen; Fourier after Julien-Léopold Boilly; Lobachevsky by Lev Kryukov; Turing by Elliott & Fry; Shannon, Bell Telephone Laboratories; Fermat, Galois, Riemann, Cantor, Poincaré, Hilbert and Gödel by unknown artists and photographers. Al-Khwarizmi: monument in Madrid, photograph by Zarateman, CC0. Fibonacci: statue in Pisa, photograph by Taty2007, CC BY-SA 4.0. Conway: Thane Plambeck, CC BY 2.0. Mandelbrot: Steve Jurvetson, CC BY 2.0.











