JOVANA
Explore Library Glossary Getting Started Three Levels Fields How it works Mission
Join the mission
All guides

Gauss's Theorema Egregium

Gaussian curvature was born from the way a surface sits in space — yet Gauss proved that an ant trapped on the surface, who can only measure distances and angles, could compute it without ever peeking outside. That stunning surprise, the 'remarkable theorem', is why a flat map of the round Earth must always lie.

Two ways to know a surface, and a wall between them

The previous two guides handed you two very different instruments. The first, the first fundamental form ds^2 = E du^2 + 2F du dv + G dv^2, is the surface's own ruler: from E, F, G alone you can measure the length of any curve drawn on the surface, the angle between two such curves, and the area of a patch. It is intrinsic — everything a creature living inside the surface could ever discover by crawling around with a tape measure, knowing nothing of the space outside. The second, the second fundamental form, records how the surface bends away from its tangent plane as you move; from it came the principal curvatures k_1 and k_2, the sharpest and gentlest ways the surface curves at a point.

Here is the crucial split. The second fundamental form is extrinsic: to define it you need the surface's unit normal — an arrow pointing out of the surface into the surrounding space — so it describes how the surface is placed in three dimensions, not just what it is like to live on it. The cleanest evidence is a sheet of paper. Roll it gently into a cylinder: you have changed how it bends in space, so its principal curvatures change (one of them was 0 flat, now one curves around the cylinder). Yet no distance or angle drawn on the paper changed at all — the ant notices nothing. Bending without stretching leaves the intrinsic world untouched while overhauling the extrinsic one.

Gaussian curvature: the suspicious product

Recall how the previous guide combined the two principal curvatures. Their average, (k_1 + k_2) / 2, is the mean curvature H, and their product, k_1 times k_2, is the Gaussian curvature K. Both were defined through the Gauss map and the second fundamental form, so on first sight both look thoroughly extrinsic — each is built from the unit normal that points out of the surface. K simply multiplies the two bending rates instead of averaging them.

The sign of K already tells a vivid story. Where K > 0, both principal curvatures bend the same way — the surface is dome-like, cupping like the outside of a sphere or the top of a hill; nearby the surface sits entirely on one side of its tangent plane. Where K < 0, the two curvatures bend opposite ways — a saddle, or a mountain pass, curving up along one direction and down along the perpendicular one; the tangent plane slices through the surface. And where K = 0, at least one principal curvature is zero: the flat plane, but also the cylinder and the cone, which curve one way and stay straight the other. That last case is the quiet hint — the cylinder, made by bending flat paper, has K = 0 just like the flat paper it came from.

The remarkable theorem itself

Gauss's Theorema Egregium (1827) states it bluntly: the Gaussian curvature K depends only on the first fundamental form. In full: K can be written entirely in terms of E, F, G and their derivatives — the intrinsic ruler alone — with the extrinsic normal nowhere in sight. The product k_1 k_2, built so visibly from how the surface bends in space, turns out to be something the ant could have computed all along from tape-measure data. Two of those bending numbers were extrinsic, but their product crossed the wall.

Be honest about what this guide can and cannot do. The actual proof runs through a formidable computation — you express the surface's second derivatives, impose the 'compatibility' that mixed partials must agree, and after pages of algebra the extrinsic terms cancel, leaving K as a fearsome but purely intrinsic formula in E, F, G (the Brioschi formula). That derivation belongs to a full course in differential geometry, not an introductory rung, and pretending it has a one-line reason would be a lie. What we can do honestly is make the result believable, see why it must be so, and feel its consequences — and those are extraordinary.

Why the flatlander can feel curvature

Here is the honest intuition behind the theorem, and it needs no normal vector at all. Stand at a point and walk a small geodesic circle of radius r around it — every point exactly distance r away, measured along the surface. On a flat plane that circle has circumference exactly 2 pi r and encloses area pi r^2, the old familiar values. But on a positively curved dome, the geodesics that radiate out lean toward each other, so the circle comes up short: its circumference is a little less than 2 pi r. On a negative saddle the geodesics splay apart and the circle is too long. The deviation, to leading order, is governed by exactly one number — K.

small geodesic circle of radius r, intrinsic measurements:

  circumference  C(r)  =  2 pi r  ( 1  -  K r^2 / 6  +  ... )

  area           A(r)  =  pi r^2  ( 1  -  K r^2 / 12 +  ... )

so, solving for K from data the ant can gather:

  K  =  lim as r -> 0  of  ( 3 / pi )  ( 2 pi r - C(r) ) / r^3
Bertrand-Diguet-Puiseux: a creature confined to the surface recovers K purely from how a small circle's circumference falls short of 2 pi r — no outside view required.

Read that last line slowly: K appears as the limit of a measurement built only from C(r) and r — circumference and radius, both gathered with a tape measure, both intrinsic. No normal vector, no embedding, no peeking out of the surface. This is the whole point, and the prompt's slogan made literal: a flatlander really can detect Gaussian curvature. A two-dimensional surveyor who never suspected a third dimension existed could, by carefully measuring circles, announce 'space here is curved, and by this much' — and be exactly right. That is why the cylinder had to keep K = 0: bending paper never disturbs any of those tape-measure circles, so the intrinsic K cannot budge.

The consequence everyone has held in their hands

Now the famous payoff. A sphere has constant positive Gaussian curvature K = 1 / R^2 everywhere; a flat plane has K = 0 everywhere. A perfect map would be a distance-preserving copy of one onto the other — but distance-preserving means same first fundamental form, and by Theorema Egregium that forces same K. Since 1/R^2 is not 0, no such map can exist. There is no way to flatten any piece of a sphere onto a plane without stretching, tearing, or compressing it somewhere. Every flat map of the Earth must distort distances — Mercator bloats Greenland, equal-area maps shear the shapes, and no clever projection will ever escape it. The orange peel that refuses to lie flat without cracking is Theorema Egregium speaking through your breakfast.

Note the honest limit of this argument: it forbids a perfect map, one that preserves all distances at once. It does not forbid maps that preserve angles (conformal, like Mercator) or that preserve area (equal-area) — you are allowed to sacrifice one thing to save another, and cartographers do exactly that. What you can never have is a map that keeps every length right, because that single demand secretly demands matching curvature, and the curvatures simply do not match. One number, intrinsic and unforgeable, stands in the way.

What you carry forward

  1. Sort surface quantities by the wall: lengths, angles, areas, and Gaussian curvature K are intrinsic (the ant can find them); the normal, the second fundamental form, and the individual principal curvatures are extrinsic.
  2. Remember the slogan of the Theorema Egregium: K = k_1 k_2 looks extrinsic but is secretly intrinsic — computable from E, F, G alone.
  3. Use the test of bending: rolling paper into a cylinder or cone changes the principal curvatures but not K, so any two surfaces with different K can never be wrapped one onto the other without stretching.
  4. Read curvature from the inside: a small geodesic circle's circumference falling short of 2 pi r reveals K > 0, overshooting reveals K < 0 — exactly how a flatlander measures the curvature of its world.

Theorema Egregium is the hinge of this whole rung. Before it, curvature seemed to belong to the embedding — a fact about a surface's posture in space. After it, the deepest curvature, K, is revealed as a fact about the surface's internal geometry, a number woven into its very fabric of distances. That reframing is precisely what lets geometry escape the ambient space entirely: the next guides follow geodesics, the straightest possible paths felt from within, and then the Gauss-Bonnet theorem ties the total of all this intrinsic curvature to the surface's Euler characteristic chi = V - E + F — bending and topology, shaking hands. And further up the ladder, when surfaces grow into manifolds of any dimension with no surrounding space to lean on, it is this intrinsic notion of curvature, the one Gauss first set free, that survives and becomes the language of curved space itself.