“…From (1),(4) and the assumption A(P) > 0 it follows that the powers of the sets {qg(i)e Ps} tend to infinity; denoting by ns the sum of the powers of the sets {p(i) e Pj}, j 1,..., s, for s 1,..., m, we get" A= min (n-n_l)--..l_s_m Reindex the variables r/i in such a way that {p(i)Ps} {ns_ + 1, ..., n,}. Now condition (5) follows from (4), and (1') from (1); if we set I,,..., (n 1,"" nm) B,, .,,(n nm) f'l then conditions (2') and (3') will follow from(2)and(3). Whence, by Theorem 2, we obtain the required convergence of joint distributions.…”