This paper began as an attempt to understand the Euclidean lattices that were recently constructed (using the theory of elliptic curves) by Elkies [E] and Shioda [Sh I,Sh2], from the point of view of group representations. The main idea appears in a note of Thompson [Th2]: if one makes strong irreducibility hypotheses on a rational representation V of a finite group G, then the G-stable Euclidean lattices in V are severely restricted. Unfortunately, these hypotheses are rarely satisfied when V is absolutely irreducible over Q. But there are many examples where the ring End G ( V) is an imaginary quadratic field or a definite quaternion algebra. These representations allow us to construct some of the Mordell-Weillattices considered by Elkies and Shioda, as well as some interesting even unimodular lattices that do not seem to come from the theory of elliptic curves.In § I we discuss lattices and Hermitian forms on T/, and in § §2-4 the strong irreducibility hypotheses we wish to make. In § 5 we show how our hypotheses imply the existence of a finite number (up to isomorphism) of Euclidean
I. LATTICES AND HERMITIAN FORMSIn this section, we establish the notation that will be used throughout the paper. Let G be a finite group of order g. Elements of G will be denoted s, t , .... Let V be a finite-dimensional rational vector space that affords a linear representation of Gover Q. We view elements of G as linear operators acting on the right of V, and so have the formula: v S / = (v s / for v E V and s, t E G. Let x(s) = Tracev(s) be the character of V.