Decomposition of complete bipartite graphs

Conditions are given for decomposing K-m,K-n, into edge-disjoint copies of a bipartite graph G by translating its vertices in the bipartition of the vertices of K-m,K-n. A construction of the bipartite adjacency matrix of the d-cube Q(d) is given leading to a convenient cu-valuation and a proof that K(d2)d-2,(d2)d-1 can be decomposed into copies of Q(d) for d > 1.