SYNOPTIC ABSTRACT
A class of (multistage) schemes for selecting the most probable event from a set of k events is proposed. Each scheme in this class consists of m experiments, where each experiment is a single-sample procedure on t events as proposed by Bechhofer, Elmaghraby and Morse. The total sample size in the m experiments is n. A scheme is optimal if, under its least-favorable configuration, among all schemes with the same parameters k, m, t and n, it maximizes the probability that the final selection is the most probable event. This paper proves some properties of an optimal scheme.