Evaluation of Factorial Partial Diallel Crosses 1

Factorial partial diallel crosses with s = 12 and 7 ( s is the number of crosses per parent) and n = 20 ( n the number of parents) were sampled from a diallel cross of 190 F 1 hybrids, reciprocals bulked, in maize ( Zea mays L.), that was evaluated in four environments, represented by two years at two locations. Randomized block designs with four replications were used. Data were recorded on grain yield, plant height, ear height, days to 75% silking, grain moisture at harvest, ear length, ear diameter, and number of kernel rows. In comparison to the full dlallel, partial diallel crosses showed an increased standard error of general combining ability effects, and some inconsistencies in variance estimates which were more frequent at the lower level of s ( s = 7) and for the characters with lower heritability (grain yield and ear length). These fluctuations were, in general, more common for those sources of variation that accounted for relatively smaller magnitudes of variability. The results obtained from the partial diallel analyses were, however, on the whole, reliable, and it is proposed that the study of the characters with low heritability such as grain yield should be supplemented by the traits which have, in general, high heritability, and that the experiments be repeated over environments.