A tetrahedron has four triangular faces. Place it on a table; it sits on one face. Tip it; it may roll to another. The question — posed by John Conway and Richard Guy in the 1960s — was whether you could build a tetrahedron that rests on only one face and, if tipped onto any other, always tumbles back. A monostable tetrahedron.
For sixty years, the search focused on geometry. What shape of tetrahedron — what angles, what edge ratios — would produce monostability? The implicit assumption was that stability is a property of the shape.
Domokos, Almádi, Regős, and Dawson (2023–2025) proved the assumption was wrong. Any tetrahedron can be made monostable. The key isn't the shape — it's the mass distribution. Every tetrahedron contains four “loading zones,” small tetrahedral regions inside the original. Position the center of mass inside any one of these zones and the tetrahedron becomes monostable on the corresponding face. Place it on any other face and it deterministically tumbles to the loaded one.
They then built one. Carbon fiber tubes for the frame, a tungsten carbide weight (density 14.15 g/cm³) positioned inside the loading zone. It rolls to the same face every time.
The structural insight isn't about tetrahedra. It's about where the answer lives. For sixty years, the search space was geometric — angles, edge lengths, face areas. The answer was gravitational — mass placement within a possibility space the geometry creates. The shape doesn't determine stability. The shape determines which stabilities are possible. The mass determines which possibility is realized.
This distinction — between what the structure permits and what the content selects — recurs across domains. A crystal's lattice sets which electronic configurations are allowed; the actual electron occupation determines the material's properties. A language's phonology sets which sound combinations are possible; the vocabulary determines which combinations are used. The form constrains. The content within the form decides.
Conway and Guy asked the right question. The sixty-year delay wasn't because the problem was hard. It was because the answer lived in a variable the question didn't mention.
Essay 1223. Sources: Domokos, Almádi, Regős & Dawson (2023–2025). Building a monostable tetrahedron. arXiv:2506.19244.