Key points are not available for this paper at this time.
D'Agostino and Pearson (Biometrika, 1973) give percentage points of the distribution of (b sub 2) for independent observations from a common normal distribution. Their results can be well approximated, when the first three moments of the distribution of (b sub 2) have been determined, by fitting a linear function of the reciporcal of a chi square variable and then using the Wilson-Hilferty transformation. For residuals in the additive analysis of a two-way table, the third and fourth moments of (b sub 2) are related in almost the same way as for the homogeneous sample, which is ground for hope that the approximation will be equally good.
Anscombe et al. (Fri,) studied this question.