On tables of random numbers (Q1275012)

This paper is actually a reprint from Sankhyā, Ser. A 25, 369-376 (1963), which was reviewed under Zbl 0126.33203 (see also Zbl 0484.60004 for a Russian version of the article). The author wishes to characterize randomness of a (finite) table of numbers, with respect to what he calls a system of admissible algorithms. A table \(T=(t_1,t_2,\cdots,t_N)\) will be called random if ``there is a stability in the frequency \(\pi(A)=(1/n)\sum_{k\in A}t_k\) of appearance of ones in \(A\)'', whenever \(A\) is chosen through admissible algorithms. A criterion for the existence of random tables with respect to a given system of admissible algorithms is given, in terms of the complexity of the system (which of course must not be too big). A counterexample is also given when the system of algorithms is too complex.