20. Let (X"I" ' " X"p), a = 1,... , n, be independently distributed according to p-variate distributions F( X"I - ~"l . . , x"p - t p ) with finite covariance matrix ~, and suppose the es satisfy the linear model assumptions of Section 1. Then under H, S;j(n - s) tends in probability to the (ij)th element aij of [See the corresponding univariate result of Chapter 7, Section 3.]