Abstract
The problem of deciding which subsets of a given set of variables have the same covariance matrices in each of two multivariate normal populations is considered. A simple technique for determining all such subsets is described. The procedure maintains (at least asymptotically) a specified upper bound on the error rate for the family of hypotheses tested. Similar methods for determining uncorrelated subsets of a given set of variables, and for finding subsets of variables whose covariance matrices are submatrices of a given matrix are also noted