In 1867, Pierre Ossian Bonnet asked whether knowing both the metric and the mean curvature at every point of a surface uniquely determines that surface. The metric tells you the distance between any two nearby points. The mean curvature tells you the average bending at each location. Together, they seem to describe everything about a surface's local geometry. For 150 years, mathematicians suspected that for compact surfaces — closed shapes without edges, like spheres and tori — the answer was yes. Local completeness implied global uniqueness.
In October 2025, Alexander Bobenko, Tim Hoffmann, and Andrew Sageman-Furnas constructed the first compact Bonnet pair: two closed tori, twisting through themselves like figure eights, that share identical metric and mean curvature data at every point yet are globally different surfaces. The local geometry is exhaustively the same. The global structure is not.
The finding is not about insufficient measurement. Both surfaces are measured completely — every point, every distance, every curvature value. The ambiguity is not a resolution problem. It persists at maximum resolution. Two globally distinct objects produce identical local descriptions, and no amount of additional local measurement can distinguish them. The distinction is global and irreducible to any collection of local facts.
The usual assumption runs the other way: if you measure enough, you know what you have. More data, more certainty. The Bonnet pair inverts this. Exhaustive local data — every point characterized by every available local quantity — is compatible with two different answers. The surface's identity lives in how its parts are assembled, not in what those parts are. You can hold every piece and still not know the whole, because the same pieces fit together in more than one way.