Partitions of Hereditary Variance for Various Genetic Models.

0 UR present knowledge of gene action for qualitative characters indicates these actions may be almost as varied as the number of genes, and there is no reason for believing this not to be true for quantitative characters. While transformations would be useful for particular types of gene action (MATHER 1949), they are of doubtful value in realistic situations where many types are probably involved. For this reason, it is helpful to maintain the linear descriptive model, which was proposed earlier (COCKERHAM 1954) and for which the partitions of variance were derived. Then by considering various genetic models, one can see how the hereditary variation is distributed among the various partitions. In this respect the partitioning method serves as a yardstick for comparing genetic models. More important, however, an opinion can be obtained on the power of the yardstick to discriminate among genetic models from experimental data. More general forms of the models considered by HORNER et al. (1955) will be considered here. The partitions of variance will be given, and the models will be examined in terms of the partitions in a very general way.