The Shape That Refused

In 1693, a prince bet a mathematician that a cube could swallow a copy of itself. Not absorb it. Not collapse into it. Literally: cut a tunnel through one cube, slide the other through. Prince Rupert of the Rhine, nephew of a deposed king, spending his late years in a Windsor Castle laboratory melting glass and posing geometric riddles nobody asked for.

He won the bet. John Wallis proved it. A cube can accommodate a passage large enough for an identical cube to pass clean through. A century later, Pieter Nieuwland -- who died at thirty, a year into his professorship at Leiden, his best work published posthumously by a mentor who outlived him -- showed the passage could be even more generous. The cube swallows a copy six percent larger than itself.

This should feel wrong. A thing passing through a tunnel bored through its twin. But the geometry works because a cube's diagonal cross-section is a hexagon, and hexagons have room to spare. The intuition that a shape is its silhouette is the error. Shapes are more spacious than they appear when you approach them from the right angle.

After Rupert, mathematicians started asking the obvious: does everything do this? Every tetrahedron, every dodecahedron, every convex polyhedron -- can each swallow a copy of itself through a properly chosen tunnel?

The answer kept being yes. Scriba proved it for the tetrahedron and octahedron in 1968. Jerrard, Wetzel, and Yuan handled the dodecahedron and icosahedron in 2017 and went further -- they conjectured that every convex polyhedron has this self-swallowing property. That every solid with flat faces and no dents carries, somewhere in its geometry, a hidden corridor wide enough for its own twin.

For eight years, nobody found a counterexample.

Then Jakob Steininger and Sergey Yurkevich built one. Ninety vertices. A hundred and fifty-two faces. Two hundred and forty edges. A shape so deliberately baroque it needed a portmanteau name: the Noperthedron, from "Nope" and "Rupert," coined by an independent researcher named Tom Murphy VII who apparently understood that the best mathematical nomenclature should sound like a toddler's refusal.

The proof is brute and elegant at once. Steininger and Yurkevich divided every possible orientation -- every way two Noperthedrons might face each other, every angle of approach, every tilt -- into eighteen million blocks. Tested the center of each. Found no passage in any of them. Then proved, block by block, that if the center doesn't work, neither does anything nearby. Eighteen million tiny refusals, each one rigorous, adding up to a single geometric no.

What the Noperthedron says isn't complicated. It says: I don't fit through myself. Not at this angle, not at that one, not if you rotate me, not if you try the diagonal, not from above or below or sideways. There is no hidden corridor. The shape is exactly as spacious as it looks and not a micrometer more.

Three centuries of precedent said otherwise. Every Platonic solid, most Archimedean solids, all rectangular boxes -- they all had the trick. The tunnel was always there if you knew where to bore. The conjecture felt inevitable, the kind of mathematical regularity that holds because the universe prefers symmetry.

But the Noperthedron has no interest in what the universe prefers.

It took ninety vertices to say what Rupert's cube couldn't: that generosity isn't geometry. That some shapes carry room they didn't ask for, and others carry exactly what they are, no surplus, no hidden passage, no corridor for a twin to slip through.

The question now is whether the Noperthedron is alone. An anomaly, or the first member of a family. Somewhere, probably, a desktop is churning through the rhombicosidodecahedron -- another suspected rebel -- and the answer is either weeks or years away, depending on how stubborn the shape decides to be.