Abstract
In this paper we present the exact null distribution of a rank statistic used in making all treatments multiple comparisons in one-way layout designs. If we let denote the ith observation from the jth population, then we are dealing with independent random samples of sizes from populations respectively. Letting is the combined samples rank of xij among all observations, and N∗ = least common multiple of Tables are given for the upper tail probabilities of the exact null distribution of R∗ for most useful combinations of .