The third door off neutral geometry
The earlier guides in this rung walked you through one half of the story. You watched the parallel postulate survive two thousand years of failed proofs, then you opened the hyperbolic door — a plane with too many parallels, where triangles have an angle defect and you toured the Poincare disk and its cousins. This last guide swings the pendulum the other way. Instead of more parallels than Euclid allows, we ask: what if there are none at all?
Remember the Saccheri quadrilateral from the foundations rung: its summit angles are either right (flat Euclid), acute (hyperbolic), or obtuse. The obtuse case is the one we open today. It is the geometry where every pair of lines meets, where a triangle's angles add to more than 180 degrees, and where the surface bending the rules curves the opposite way to the hyperbolic plane. The friendliest example sits on your desk: a globe. This is spherical geometry, and its tidied-up sibling elliptic geometry.
What counts as a straight line on a sphere
The whole subject turns on one definition: what is a 'line' on a sphere? You cannot draw a Euclidean straight segment — every path on the surface bends in space. So we use the next best thing, the straightest possible path on the surface itself: the shortest route between two points. On a sphere that shortest path always lies on a great circle — a circle cut by a plane through the center of the sphere, like the equator or any line of longitude. Lines of latitude (except the equator) are not lines; they curve the easy way and a plane flying along one is constantly turning.
This is exactly why long-haul flights look bent on a flat map. A jet from Tokyo to New York arcs up over Alaska, not along the latitude line that looks straight on the wall map — because the great-circle route really is shorter. The map is lying, gently; the sphere is telling the truth. These straightest-on-the-surface paths are what later rungs will call geodesics, the honest generalization of 'straight line' to any curved surface. For now, 'line on the sphere' means 'great circle', full stop.
Now the punchline that ends parallels. Any two distinct great circles must intersect — slice a sphere with two different planes through its center and those planes cross in a line through the center, which pokes out through the surface at two opposite points. There is simply nowhere for two great circles to hide from each other. So on a sphere there are no parallel lines at all: give me a line and a point off it, and every line through that point eventually crosses the first. That is the obtuse-angle world, made concrete.
Triangles that bulge: angle excess
Draw a triangle from three great-circle arcs and something wonderful goes wrong. Here is the cleanest example: start at the North Pole, run straight down to the equator, turn 90 degrees, run a quarter of the way around the equator, then turn 90 degrees and head straight back up to the pole. You arrive home having made three right angles. The angles sum to 270 degrees — a full 90 more than a flat triangle. The triangle does not fall short of 180 as in the hyperbolic plane; it overshoots.
That surplus has a name: the angle excess, defined as E = (sum of the three angles) - 180 degrees. It is the exact mirror image of the angle defect you met two guides ago. In the hyperbolic plane the defect was positive and triangles ran short; on the sphere the excess is positive and triangles run long. Flat Euclidean geometry is the knife-edge between them, the unique case where the excess is exactly zero and triangles sum to precisely 180. Three geometries, one dial.
And here is the result that makes the sphere magical, Girard's theorem: the area of a spherical triangle is nothing but its angle excess. On a unit sphere (radius 1), area equals the excess measured in radians. Our pole-to-equator triangle had three right angles, an excess of 90 degrees = pi/2 radians, so its area is pi/2 — and indeed it covers exactly one-eighth of the sphere, whose total area is 4 pi. One-eighth of 4 pi is pi/2. The numbers land perfectly. Angle and area are the same thing here, which simply never happens in the flat plane.
Girard's theorem (sphere of radius r):
area = r^2 * ( A + B + C - pi ) angles in radians
= r^2 * E E = angle excess
Worked check (unit sphere, r = 1, three right angles):
E = (pi/2 + pi/2 + pi/2) - pi = pi/2
area = 1^2 * pi/2 = pi/2
whole sphere = 4*pi , so triangle = (pi/2)/(4*pi) = 1/8 of itCurvature: the flatlander really can tell
Why does the sphere force an excess while the hyperbolic plane forces a defect? Because of curvature, and here the extrinsic/intrinsic distinction earlier rungs were careful about really pays off. A sphere of radius r has constant positive Gaussian curvature K = 1/r^2; the hyperbolic plane has constant negative curvature; the Euclidean plane has K = 0. Angle excess and Gaussian curvature are two views of the same fact — for a triangle, the excess equals the curvature times the area, so positive K forces a positive surplus.
Hold on to one subtle, beautiful point. A creature living on the surface, never able to look at it from outside, can still discover its curvature — purely by measuring triangles and finding their angles do not sum to 180. It needs no view from space, no third dimension to peer into. The curvature is intrinsic: it lives in the surface's own distances, not in how the surface sits in the room around it. A flatlander on a sphere can prove it lives on a sphere. That is precisely the seed of Gauss's Theorema Egregium, which a differential-geometry rung will unfold in full.
From sphere to elliptic plane: gluing the antipodes
Now we settle the unfinished business from the first callout. On the bare sphere, two great circles meet in two opposite points, so 'two points determine a unique line' fails whenever the two points are antipodal — through the North and South Poles run infinitely many lines of longitude. To get a clean geometry that truly mirrors the parallel-free axioms, we make one decisive move: declare each pair of antipodal points to be the same point. Glue every point to the one diametrically opposite it.
After this gluing the troubles dissolve. Two distinct points now lie on exactly one line again; two distinct lines meet in exactly one point — no exceptions, perfect symmetry between points and lines. This glued-up surface is the elliptic plane, and it is the same object as the projective plane you will meet in the projective-geometry rung, now wearing a metric so we can measure angles and distances on it. Spherical geometry was the warm, visual picture; the elliptic plane is its axiomatically clean form.
One honest caveat before we close. Some of Euclid's neutral-geometry theorems used the silent assumption that lines are infinitely long and that a point always lies between two others on a line. On the sphere and elliptic plane that fails — a 'line' is a closed loop of finite length (a great circle has circumference 2 pi r), so betweenness and the usual ordering axioms must be rebuilt. This is exactly the gap Saccheri fell into when he 'refuted' the obtuse case: he assumed infinite lines without noticing. Elliptic geometry is consistent, but it is not simply neutral geometry plus an axiom — it gently revises the order axioms too.
Three geometries, one map
Step back and the whole rung snaps into a single picture. There is one dial, the Gaussian curvature K, and turning it gives you three complete and equally honest geometries. Set K negative and you get the hyperbolic plane: too many parallels, angle defect, triangles pinched. Set K = 0 and you recover flat Euclid: exactly one parallel, angle sum precisely 180. Set K positive and you get the sphere and its glued cousin the elliptic plane: no parallels, angle excess, triangles bulging.
Notice how cleanly the three echo Saccheri's three summit-angle hypotheses — acute, right, obtuse — that opened the foundations rung. What he hoped was a false branch to be eliminated turned out to be two real worlds standing beside Euclid's. And the elliptic plane carries a lovely bonus: it is its own projective plane, so the journey you began chasing the fifth postulate flows straight into the projective and differential geometry waiting on the next rungs.
A final word on honesty, the through-line of this whole rung. None of these three geometries is the 'real' one and the others mere curiosities. Each is true to its own axioms; the sphere was beneath your feet long before anyone wrote down a postulate. Gauss, Bolyai and Lobachevsky did not break Euclid — they discovered that geometry offers a choice, and Riemann's 1854 habilitation lecture then showed how to bend curvature freely from point to point, opening the door to the curved-space geometry that Einstein would one day borrow to describe gravity. You finish this rung not with one geometry but with a map of all of them.