In magnis et voluisse sat est.Abstract. We identify the representations K[X k , X k−1 Y, . . . , Y k ] among abstract Z[SL 2 (K)]-modules. One result is on Q[SL 2 (Z)]modules of short nilpotence length and generalises a classical "quadratic" theorem by Smith and Timmesfeld. Another one is on extending the linear structure on the module from the prime field to K. All proofs are by computation in the group ring using the Steinberg relations.We study here certain representations of the group SL 2 (K) as an abstract group; more precisely, we aim at identifying the various symmetric powers of Nat SL 2 (K), conveniently thought of as the various spaces of homogeneous polynomials in two variables with fixed degree, among Z[SL 2 (K)]-modules. Differently put, we study the inclusion of the class of representations of the algebraic group SL 2 over the field K, in the wider class of Z[SL 2 (K)]-modules. The question may sound not quite irrelevant to admirers of the Borel-Tits Theorem on abstract homomorphisms between groups of points of algebraic groups; we deal with abstract modules instead.We cannot use Lie-theoretic, algebraic geometric, nor character-theoretic methods since SL 2 (K) is to us but an abstract group and K is arbitrary. We cannot even use linear algebra since we do not assume our modules to be vector spaces. Our only method is then brute force computation in images of the group ring. So the problem rephrases into: To which extent is the representation theory of SL 2 (K) determined by the "inner" group-theoretic constraints?The present study is therefore yet another instance of the general problem of investigating representations of algebraic groups from a purely group-theoretic perspective, which we tackled in [3] and [4]. It can however be read independently of the latter two articles and was written in this intention.One should simply recall a result first proved by F. G. Timmesfeld and S. Smith separately. In what follows, Nat stands for the natural representation, here the action of SL 2 (K) = SL(K 2 ) on K 2 . Moreover U stands for a unipotent subgroup of SL 2 (K), and the assumption on the U -length being 2 means that U acts quadratically: for all u 1 , u 2 ∈ U , one has (u 1 − 1)(u 2 − 1) = 0 in End(V ). One word on this assumption -since we are dealing with abstract modules instead of vector spaces, there is no dimension around. Unipotence length is then the natural candidate to measure the complexity of target modules; the length of Nat SL 2 (K) is 2 (and more generally the length of Sym k Nat SL 2 (K) is also its dimension over K, namely k +1). MSC 2010: 20G05; 20C07 Institut de Mathématiques de Jussieu -Paris Rive Gauche adrien.deloro@imj-prg.fr 1 2 ADRIEN DELOROTimmesfeld's Quadratic Theorem ([9, Theorem 3.4 of chapter I], also [8]). Let K be a field, G = SL 2 (K), and V be a simple Z[G]-module of U -length 2. Then there exists a K-vector space structure on V making it isomorphic to Nat SL 2 (K).Our original motivation was to find a similar result identifying the adjoint representation, viz. the...