Letters to the Editors

I refer to Olguõ n and Fearn (1997) and to Bissell (1989, 1992), in which I o€ered a simple -test for determining whether any mean squares in a saturated n-design indicate signi®cant e€ects …n ˆ 2 or more levels of each factor, k is the number of e€ects and thus ˆ nÿ 1 is the number of degrees of freedom per e€ect and there are k mean squares each on degrees of freedom in the analysis of variance). In Bissell (1989, 1992) I showed that, if the mean and variance of these mean squares, m and s, are calculated, then under the null hypothesis