A genetic theory of general varietal ability for diploid crops.

The use of estimates of general varietal ability (g.v.a.) of individual plants, as defined in an earlier paper (Wright, 1973), is examined in the contexts of both the population improvement and synthetic variety building phases of a breeding programme, and its theoretical efficiency compared with those of parental and g.c.a. methods of assessment. It is concluded that selection based on g.v.a. may be useful during the varietal building phase when heritability is low. In the absence of epistasis, the g.v.a. variance is shown to account for all the variance among the synthetics of size s which can be drawn from a large random-bred population, except for [Formula: see text]. The possibilities of the prediction of synthetic variety performance from I1 and polycross progeny data is discussed.