An important source of information on the quality of random number generators is the result obtained when using numbers to simulate a problem, whose solution is known.These algorithms are sensitive to non-randomness, in particular non-uniformity, in successive sequences of a given length. A large number,N, of non-overlapping sequences of the length m, ( ), are generated and in each case the mean is found, The number of cases where the mean is less than a half N, and the greater than a half, N, is tested against expectations using the chi-square statistics, i.e.