Classification of isosceles eight-point sets in three-dimensional Euclidean space (Q820079)

A subset \(X\) of the \(k\)-dimensional Euclidean space \(\mathbb R ^k\) is called a \(P(n)\)-set, or more descriptively an isosceles \(n\)-point set, if \(X\) contains \(n\) points and if every three of these points form an isosceles triangle. In his solution to a problem proposed by Paul Erdős, \textit{L. M. Kelly} proved in [E735, Am. Math. Mon. 54, 227--229 (1947)] that there does not exist a \(P(7)\)-set in \(\mathbb R ^2\) and that the set consisting of the vertices and center of a regular pentagon is essentially the only \(P(6)\)-set there. He also gave as an example of a \(P(8)\)-set in \(\mathbb R ^3\) the set consisting of the points \((0,0,1)\), \((0,0,0)\), \((0,0,-1)\) and the vertices a regular pentagon located in the \(xy\)-plane and having center at \((0,0,0)\) and circumradius 1. In 1962, \textit{H. T. Croft} [Proc. Lond. Math. Soc. III. Ser. 12, 400--424 (1962; Zbl 0106.35203)] showed that there does not exist any \(P(9)\)-set in \(\mathbb R ^3\). The paper under review proves that Kelly's example is essentially the only \(P(8)\)-set in \(\mathbb R^3\). A \(P(n)\)-set \(X\) is called an \(s\)-distance \(P(n)\)-set if the set \(\{ \| x-y\| : x,y \in X,~x \neq y \}\) consists of \(s\) distinct elements. It has been shown by \textit{S. J. Einhorn} and \textit{I. J. Schoenberg} in [Indag. Math. 28, 479--488, 489--504 (1966; Zbl 0145.17103)] that there does not exist any 2-distance \(P(7)\)-set in \(\mathbb R^3\) and that there exist 6 essentially different 2-distance \(P(6)\)-sets there. The paper under review proves that if \(s \leq 4\), then there does not exist any \(s\)-distance \(P(8)\)-set in \(\mathbb R^3\). In view of the fact that Kelly's aforementioned example is a 5-distance \(P(8)\)-set, the number 4 in Kido's theorem cannot be improved.