The earliest geometry humans practiced was, paradoxically, not Euclidean. People wanted to navigate by the stars, to measure the Earth, to predict eclipses — and these problems live on the surface of a sphere, where triangles don’t behave like the triangles on a flat sheet of paper.
Greek astronomers from the 4th century BCE onwards developed spherical trigonometry for celestial navigation. The mathematicians of medieval Islam refined it. By the time Euclid had codified plane geometry around 300 BCE, sailors and astronomers were already routinely doing spherical geometry — even though no one yet realized it was a different geometry from Euclid’s.
The idea that there are multiple equally-valid geometries — that Euclid’s flat plane is not the only consistent way to do geometry — emerged only in the 19th century with the discovery of non-Euclidean geometry. At that point, mathematicians realized that the sphere had been a non-Euclidean surface all along, just one nobody thought of as such because it was so familiar.
This article is about the geometry of the sphere: how it differs from flat geometry, why those differences matter, and where it shows up in modern technology.
What’s a “line” on a sphere?
In Euclidean geometry, a “straight line” is the shortest path between two points. On a sphere, the same definition applies — but the shortest path between two points on the sphere’s surface is not a straight line in 3D space. It’s an arc of a great circle — a circle on the sphere whose center coincides with the sphere’s center.
The equator is a great circle. So is any meridian (line of longitude). The arc connecting Tokyo to New York along the shortest path passes nearly over the Arctic. Lines of latitude (other than the equator) are not great circles — they’re parallel circles, and traveling along a latitude line is not the shortest path.
In spherical geometry, the analogue of a “line” is a great circle. From this, the rest of spherical geometry follows.
How spherical triangles differ from flat ones
A spherical triangle is bounded by three great-circle arcs. Its properties differ dramatically from those of a flat triangle.
Angle sum: A flat triangle’s angles sum to exactly 180° ( radians). A spherical triangle’s angles always sum to more than 180°, and the excess depends on the triangle’s area:
where is the sphere’s radius. The angle excess is exactly equal to the area of the triangle (in steradians).
This is the spherical analogue of hyperbolic geometry’s “angle deficit” — only with the opposite sign. On a sphere, large triangles are “fatter” than expected; their angles bulge outward.
Concrete example: a triangle with one vertex at the North Pole and two vertices on the equator, separated by 90° of longitude. Its three sides are: two meridian arcs running from pole to equator (each 90° long), and a 90° arc of the equator. All three angles are 90°. Sum: 270°. Excess over 180°: 90°. The triangle covers 1/8 of the sphere’s surface.
No similar triangles: In Euclidean geometry, you can have triangles with the same angles but different sizes. In spherical geometry, you can’t. Once you fix the three angles, the area is determined (by the angle excess formula), so the size is fixed.
No parallel lines: Two great circles always meet — at exactly two antipodal points. There are no parallel “lines” on a sphere. This is the failure of Euclid’s fifth postulate in the spherical case (in the opposite direction from hyperbolic geometry).
The spherical Pythagorean theorem
The Pythagorean theorem — for a flat right triangle — has a spherical analogue, but it looks very different.
For a right spherical triangle with sides and hypotenuse (all measured as arc lengths or angles from the sphere’s center):
When are small (so the triangle is small compared to the sphere), Taylor-expanding the cosines gives back the flat Pythagorean theorem to leading order. So the flat theorem is the small-triangle limit of the spherical one.
Similar reformulations exist for the law of sines and law of cosines. The full spherical trigonometry — developed by Greeks, Persians, and medieval Islamic mathematicians — gave navigators precise tools for celestial calculations long before Newton invented calculus.
Cartography and map projections
A fundamental fact: you cannot flatten a sphere onto a plane without distortion. Cut up an orange peel and try to flatten it; you can do it only by tearing or stretching. This is a consequence of the difference in curvature: the sphere has positive curvature, the plane has zero, and Gauss’s Theorema Egregium says curvature is intrinsic to the surface.
Every map is a projection that distorts something. Different projections preserve different properties:
Mercator projection (1569): preserves angles. Loxodromes — paths of constant compass bearing — appear as straight lines. This made Mercator the standard for navigation. Its cost: areas are dramatically distorted at high latitudes. Greenland appears the size of Africa. This is misleading; Africa is about 14 times larger.
Equal-area projections (e.g. Mollweide, Albers): preserve relative areas. Greenland looks correctly small. They distort shapes — countries near the edges look squashed.
Equidistant projections (e.g. azimuthal equidistant): preserve distances from a single chosen point. Used in radio communications, where you want to know how far a transmitter is from your antenna.
Conformal projections preserve angles locally (Mercator is one). Stereographic projection is a famous conformal projection used in cartography and complex analysis.
The choice of projection always involves trade-offs. There is no perfect map. This is a precise mathematical statement — the impossibility of unfolding a curved surface into a flat plane is a theorem.
GPS and great-circle distance
Modern GPS systems use spherical trigonometry routinely. When your phone calculates the “distance to your destination,” it’s computing along a great-circle arc.
The formula for great-circle distance between two points and on a sphere of radius — given by latitudes and longitudes — is the haversine formula:
This is the standard distance formula in navigation. It’s accurate for the Earth as a sphere, with ~0.3% error from the actual ellipsoidal shape. For higher precision, Vincenty’s formula handles the ellipsoid properly but is more complex.
GPS satellites broadcast time signals; receivers measure the time-of-flight from each satellite, compute the receiver’s distance to each, and triangulate (or rather, “tetra-late” with four satellites) the receiver’s position. The triangulation uses spherical trigonometry to convert distances into latitude, longitude, and altitude.
If the Earth were treated as flat for GPS calculations, errors of 1-2% would accumulate over distances comparable to Earth’s radius. For most navigation purposes, this would be unacceptable. Spherical geometry is essential.
Aircraft routing and great-circle paths
When a plane flies from New York to Tokyo, the shortest route doesn’t head west across the Pacific. It heads north, passes over Alaska or near the Arctic, and arrives in Tokyo from the north.
This is the great-circle path. It’s “straight” on the sphere, even though it looks curved on flat maps. Pilots and air-traffic-control systems plan around great-circle routes, with deviations only for weather, jet streams, or political airspace restrictions.
A common visualization: take a globe and a piece of string. Stretch the string between two cities and pull it taut along the surface. The path the string takes is the great circle. On a flat Mercator map, this same path looks curved.
The geometry of the sky
Astronomy was historically the main consumer of spherical geometry. The night sky, viewed from Earth, looks like the inside of an enormous sphere — the celestial sphere. Stars have positions on this sphere given by analogues of latitude and longitude (declination and right ascension).
Spherical trigonometry lets you:
- Convert between coordinate systems (equatorial, ecliptic, galactic, horizontal).
- Calculate when celestial bodies will rise and set at a given location.
- Predict eclipses and planetary alignments.
- Design celestial navigation routines for ships.
Modern astronomy still uses these techniques for observing schedules and instrument pointing, even though high-precision work involves more complex coordinate transformations (relativistic, accounting for Earth’s motion, atmospheric refraction).
What spherical geometry teaches
The deepest lesson of spherical geometry is that the geometry you grew up with is one of three options. The flat geometry of paper, the positively-curved geometry of the sphere, and the negatively-curved geometry of hyperbolic space are all internally consistent, each with its own theorems and applications. None is “the right one” — they’re complementary.
For two thousand years, mathematicians treated Euclidean geometry as if it were the only possibility. Practical navigators meanwhile did spherical geometry as a matter of course — and by the 19th century, the synthesis was finally explicit: there are three geometries of constant curvature, plus Riemann’s general framework that includes all of them and many more.
The synthesis matters. It’s why general relativity could be formulated. It’s why modern differential geometry exists. It’s why we can do GPS — without realizing that the Earth’s surface is intrinsically curved (not just curved when viewed from outer space), the math doesn’t work.
For everyday purposes: when you see a flat world map, remember it’s a projection with intentional distortions. When your phone tells you a distance, it’s using spherical trigonometry. When you fly internationally, you’re tracing a great circle that looks weird on the in-flight map but is genuinely the shortest path.
The Greeks did spherical geometry without having a name for what kind of geometry it was. Modern mathematics has the names and the unified framework. The mathematics underneath is exactly what those Greek astronomers were doing: counting angles, measuring arcs, navigating by stars. They just had less to call it.
Frequently asked
Is the Earth's surface a sphere mathematically?
Approximately, yes. The Earth is more precisely an oblate spheroid (slightly flattened at the poles), but for most navigation and calculation purposes, treating it as a sphere is accurate to about 0.3%. This is why spherical geometry is the standard framework for cartography, GPS, and aircraft routing — even though high-precision applications use the more accurate ellipsoid model.
Why do all great-circle paths look 'curved' on flat maps?
Because flat maps must distort the sphere somehow — there's no way to flatten a sphere onto a plane without distorting either angles, areas, or distances. The Mercator projection preserves angles (great for navigation) but distorts areas dramatically — Greenland looks bigger than Africa, when it's actually about 14 times smaller. The straight line on a Mercator map is not the shortest path; the great circle is.
How is spherical geometry the 'third' geometry?
Together with Euclidean (flat) and hyperbolic (negatively curved) geometry, spherical geometry (positively curved) forms the trio of geometries with constant curvature. Each is internally consistent, has its own theorems, and applies to specific physical situations. The unification — that all three are special cases of Riemann's general framework — was one of the major mathematical syntheses of the 19th century.