Non-Euclidean Geometry

For two thousand years, mathematicians tried to prove Euclid's fifth postulate. When they finally gave up, they discovered that the shape of space had never been obvious in the first place.

The most influential textbook ever written is a book of geometry. Euclid's Elements, compiled around 300 BCE, framed mathematics for two millennia. It opens with five postulates — simple assumptions from which every theorem is meant to follow. Four are terse and self-evident. The fifth is not.

The fifth postulate, in Euclid's phrasing, says roughly this: if a straight line crossing two others makes the interior angles on one side add to less than two right angles, then those two lines, extended far enough, will meet on that side. It reads more like a theorem than an axiom. Its statement is longer than the other four combined. And so, for the next two thousand years, mathematicians tried to prove it — to derive it from the other four and put it out of its misery.

None of them succeeded. But their failure, when it finally came into focus, changed geometry forever.

The two-thousand-year itch

Ptolemy tried. Proclus tried. The medieval Persian polymath Nasir al-Din al-Tusi tried. Wallis, Legendre, Lambert — each convinced they had almost got it, each producing a “proof” that turned out to have quietly assumed the very thing it was supposed to prove.

The most heroic failure was Giovanni Saccheri, an Italian Jesuit. In 1733 he published an entire book — Euclides ab omni naevo vindicatus, Euclid Freed of Every Flaw — that tried to prove the fifth postulate by contradiction. He assumed its negation and worked out the strange, unfamiliar consequences: triangles whose angles sum to less than 180°, lines that diverge exponentially, a whole shadow geometry. He kept expecting to hit an outright contradiction. He never did. But he was so certain the postulate had to be true that he stopped just short, declared his own results “repugnant to the nature of the straight line,” and called it a proof.

He had, in fact, done exactly the opposite. He had derived a consistent alternative geometry. He just could not bring himself to see it.

Bolyai and Lobachevsky — the plunge

In the 1820s and 30s, two people made the leap independently. János Bolyai, a young Hungarian army officer, wrote to his father: “I have discovered such wonderful things that I was amazed... out of nothing I have created a strange new world.” Around the same time, Nikolai Lobachevsky, a professor at Kazan, published a paper on what he called “imaginary geometry” — a geometry in which through any point outside a given line, not one but infinitely many other lines fail to meet the original.

This is hyperbolic geometry. Its natural model is a saddle-shaped surface. Triangles on a saddle sag inward: their sides bow toward each other, and their angles sum to less than 180°. Parallel lines, released from Euclid's discipline, spread apart. Circles have too much circumference for their radius. Everything you were taught in school feels subtly, uncannily wrong — and yet nothing contradicts itself.

Bolyai's father passed his son's manuscript on to his old college friend Carl Friedrich Gauss. Gauss wrote back that he could not praise it, because to praise it would be to praise himself — he had reached the same conclusions decades earlier and never published, fearing “the outcry of the Boeotians.” The greatest mathematician of the age had held his tongue for thirty years rather than say, in public, that Euclid might be optional.

Riemann — the general theory

The full generalization came from Bernhard Riemann. In 1854, at the age of 27, he delivered a habilitation lecture — a single lecture — that reorganized geometry as a whole. His idea: geometry does not begin with lines and points fixed in some absolute empty space. It begins with a metric — a rule for measuring the distance between nearby points — imposed on a smooth manifold. Different metrics give different geometries.

On a flat sheet, the metric is Pythagorean: the square of the distance between two nearby points is dx² + dy². On the surface of a sphere, the metric bends. Shortest paths — geodesics — become great circles. Triangles bulge outward, and their angles sum to more than 180°. On a saddle-like manifold, the metric bends the other way, and Lobachevsky's imaginary geometry drops out as a natural consequence.

Riemann classified geometries by their curvature. Positive curvature: sphere-like, elliptic. Zero: flat, Euclidean. Negative: saddle-like, hyperbolic. Euclid's plane was a special case — the middle one — and had never been anything more.

The angles of a triangle told you the curvature all along. Euclid's 180° was never a theorem about space. It was a specification of one particular space, among infinitely many.

The consequence — what shape is the world?

For decades, non-Euclidean geometry was a beautiful abstraction. A curiosity for pure mathematicians, sniffed at by philosophers. Then Einstein turned it into physics.

In 1915, general relativity described gravity not as a force pulling on masses through empty space but as a curvature of spacetime caused by them. Massive objects bend the manifold around themselves; light and freely-falling bodies simply follow the geodesics of the curved geometry, and to us that following looks like the pull of gravity. The mathematical language Einstein needed was Riemann's exactly — his metric tensor, his curvature invariants. Without Riemannian geometry, the field equations could not have been written. Einstein said so himself, and thanked his friend Marcel Grossmann for teaching him the machinery.

The predictions were verified almost immediately. In 1919, Arthur Eddington observed starlight bending around the sun during a total solar eclipse. The bend matched Einstein's calculation to within experimental error. Lobachevsky's “imaginary geometry” — ridiculed as gibberish by the St Petersburg Academy in his own lifetime — turned out to be, in the large, the actual geometry of the actual universe.

Today, GPS satellites correct for it every millisecond. The clocks on a satellite tick at a slightly different rate from clocks on the ground, because time itself flows differently through a curved manifold. Without the correction, GPS coordinates would drift by kilometres per day. Non-Euclidean geometry is quietly running inside the phone in your pocket.

The larger lesson runs deeper still. Euclid's fifth postulate was never a proof waiting to be found. It was a choice — a specification of which world you wanted to describe. Once mathematicians stopped trying to prove it and started asking what happens when you refuse it, the very concept of space broke open. It became plural. Physicists learned to ask which geometry the universe actually inhabits, rather than assume they already knew. And the answer — that spacetime is a dynamical, curved manifold whose shape is written by its contents — is arguably the deepest thing physics has ever said about anything.

Two thousand years of failure. And on the other side of that failure, the world.