ana JOHNSON
twist into w
anakata

the Johnson solids

A convex solid whose faces are all regular polygons is either one of the uniform ones (the 5 Platonic solids, the 13 Archimedean, the prisms and antiprisms) or one of exactly 92 others. Norman Johnson listed them in 1966 and thought that was all; Victor Zalgaller proved it (1967–69). Nearly all are built from a few parts (pyramids, cupolae, the pentagonal rotunda, prisms and antiprisms) by a few moves: stack two (bi-, ortho- or gyro- as they line up), put a prism or antiprism between (elongated, gyroelongated), stick one onto a face of a uniform solid (augmented), cut one off (diminished), or turn one (gyrate). The last nine are one-offs, like the snub disphenoid, whose coordinates are roots of cubics.

Seven of them have the same faces round every vertex and still aren't uniform, because the faces are arranged differently at different vertices: J27, J34, J37 (the elongated square gyrobicupola, which looks for all the world like the rhombicuboctahedron with a twist) and J72–J75.

four dimensions

In 3D a solid's boundary pieces are its faces. In 4D they are cells, solids themselves, with faces of their own, so "made of regular pieces" splits two ways.

Every cell regular: the Blind polytopes, classified by Gerd and Roswitha Blind (1979–91). Besides the uniform ones, a few pyramids and bipyramids, and one enormous family: take the 600-cell (600 tetrahedra, 120 vertices), choose vertices no two of which are neighbours, and cut each off. Every cut turns the 20 tetrahedra round a vertex into one icosahedron. There are 314,248,344 ways, up to symmetry (Dutour Sikirić and Myrvold, 2007), and one of them, cutting 24 in just the right places, is the uniform snub 24-cell. Try it under 600-cell.

Every 2D face regular (cells may be anything, Johnson solids included): the CRF polychora. Nobody has classified them. A prism over any of the 92 is one (two copies and a prism over each face), and so is a pyramid over any solid whose vertices lie on a sphere of radius under one edge: of the 92 only five (J1, J2, J11, J62, J63). Over J1 and the three diminished icosahedra you can put an apex on both sides; over J2 you can't, because its centre lies outside it. And the "bipyramid" over the square pyramid turns out to be the octahedral pyramid: its square's two cells lie in one hyperplane and fuse into an octahedron. Klitzing's segmentochora (vertices on two parallel hyperplanes) are the 4D cupolae: a stack of two polyhedra joined by whatever cells the gap needs.

how it's drawn, and checked

The 92 are drawn from David McCooey's coordinates (Visual Polyhedra, gathered by Loïs Mignard-Debise). The page doesn't take them on trust: the selftest rebuilds every face from a convex hull of the vertices alone, checks every face is a regular polygon with unit edges (worst error 3·10⁻¹²), and checks volumes against closed forms. The 4D solids are built here: the 600-cell from its 120 vertices, its 600 tetrahedra as the 4-cliques of its edge graph; every cell kept as a half-space, so a 3D slice is the intersection of those half-spaces in your slice (a cube clipped plane by plane), each face coloured by the cell it cuts and tinted teal where that cell leans ana (the slice grows as you go ana) or rose. The shadow is every edge projected from an eye in w, coloured by how far ana or kata it is. Each 4D solid is checked convex, every cell flat, every edge one, and Euler's V − E + F − C = 0.

Controls are CAD⁴'s (kata.mino.mobi/cad4): drag to look round, pinch or scroll to zoom; the pad twists your slice into w about what you see as across and up; the lever moves it ana and kata; ⊞ squares it up. Keys: arrows twist, Q E move, , . step through the solids.