Evaluation of the culling variate used by breeders in actual selection.

The theoretical basis and computational procedures for estimating the culling variate utilized by breeders in actual selection are presented. The essential part of this procedure is to derive the unknown culling variate in terms of a linear combination of traits based on realized indirect selection differentials of those traits and phenotypic variances and covariances among traits in the population. Special emphasis is placed on the distribution of the variables involved. The accuracy of the culling variates is estimated by use of the biserial correlation with some modifications. Numerical illustrations are given for different types of selection.