Extensive data on estimates of mean squares and components of variance in a quantitative genetic study were found to be in general agreement with less extensive data reported by Comstock and Robinson [1951]. The data examined were found to be in reasonably good agreement with expectation based upon normal and random distribution of the primary variables. The distributions of the mean squares and components of variance calculated from them were of the kinds expected, and standard deviations calculated from the numerous observed estimates were in good agreement with standard deviations calculated from the pooled estimates. Approximate fiducial limits as suggested by Bross failed to include the mean with the expected frequency, but the discrepancy was not great enough to be of serious concern. The general agreement with expectation gives one greater confidence in the methods used in experiments of this kind.

Text in English

Reprints Collection