On the chiral Archimedean solids (Q1610968)

The authors demonstrate that, alone among the Archimedean solids, the snub cube and snub dodecahedron have edge lengths which are not constructible (in the usual straightedge-and-compass sense) from their circumradii. This mildly surprising fact does not seem to have been recognized before. In the case of the snub cube, the authors embed the square faces in those of a cube, and derive a cubic equation satisfied by \((d/e)^2\), where \(d\) is the circumradius and \(e\) the edge length. (Interestingly, the value of \(d/e\) obtained is remarkably close to \(\sqrt 65/3\), differing only in the fifth place after the decimal point. This is probably a coincidence, but\dots.) A similar approach is used for the snub dodecahedron, but taking advantage of the fact that this solid can also be considered as a ``snub icosahedron''. That is, rather than embedding its pentagonal faces in those of a dodecahedron, those triangular faces that do not share an edge with the pentagons are embedded in those of the dual icosahedron. Again, a polynomial is found, the roots of which are not constructible. There are some notational oddities in this paper. In the fourth section, \(\tau\) is first used in its traditional role, representing the golden ratio; it is a little disconcerting, a few lines later, to see it set to 0. It is not clear why the authors choose to use Kepler's Graeco-Latin cubus simus and dodecaedron simum, throughout, of which ``snub cube'' and ``snub dodecahedron'' are exact translations. Moreover, the use of ``silver'' to refer to quantities involving \(\sqrt 2\), by analogy with ``golden'' for those involving \(\sqrt 5\) or \(\tau\), is already somewhat precious; the archaic ``silvern'' used here is overdone.