hyperbolic geometry
For two thousand years geometers believed there was only one possible geometry — Euclid's — and that his fifth postulate (through a point off a line there is exactly one parallel) simply had to be true. Generations tried to prove it from the other axioms and all failed. Hyperbolic geometry is what you get when you keep every Euclidean axiom EXCEPT that one, and replace it with its opposite: through a point not on a given line there pass infinitely many lines that never meet the given line. It is a complete, consistent geometry, just as logically sound as Euclid's — only different.
Here is the picture. Take a line L and a point P not on it. In Euclid's plane exactly one line through P misses L. In the hyperbolic plane, swing a line through P: as you rotate it, there is a whole open fan of directions in which it never hits L, bounded by two special 'limiting' lines that approach L without ever touching it. Every line strictly inside that fan is a parallel (a non-intersecting line), so there are infinitely many. The consequences cascade: the angles of every triangle add to LESS than 180 degrees, the shortfall growing with the triangle's size; rectangles do not exist; and two triangles with equal angles are automatically congruent, so there is no such thing as enlarging a figure while keeping its shape.
Three people discovered this independently around 1830 — Carl Friedrich Gauss (who kept it private, fearing ridicule), the Hungarian Janos Bolyai, and the Russian Nikolai Lobachevsky, who published first. The crucial honest point: this did NOT prove Euclid wrong. The parallel postulate is INDEPENDENT of the others — neither provable nor disprovable from them — so both geometries are consistent. Non-Euclidean geometry did not refute Euclid; it revealed that 'the' geometry was a choice all along, and decades later it would give Einstein the language for curved spacetime.
In a hyperbolic plane, take any triangle and measure its three angles. If they sum to, say, 174 degrees, the missing 6 degrees is the triangle's 'angle defect'. Make the triangle bigger and the defect grows; shrink it toward a point and the angle sum creeps back up toward 180 degrees, so very small hyperbolic triangles look nearly Euclidean.
Angle sum below 180 degrees is the signature of hyperbolic geometry, and the shortfall measures area.
A common error is to say non-Euclidean geometry 'disproved' the parallel postulate. It did not — the postulate is independent, so it can be consistently assumed OR denied; both worlds are logically legitimate.