Abstract
Response from selection for combining ability is proportional to the genetic variance among test crosses in each cycle. Theoretically, this variance can be increased appreciably by a generation of inbreeding in the population under selection prior to initiation of a new cycle. for a single locus with two alleles, crossing S1 rather than S0 plants with a tester is expected to increase variance among test crosses within intercross progenies (of selected S1 lines from the previous cycle) by 100%, and over the entire population by 50%. Sib‐pollination within intercross progenies would be only one‐half as effective as selfing in this respect. Random mating of the population for one generation would have no effect on test cross variance so far as individual loci are concerned.Experimental data for grain yield were in good agreement with the expected results based on genetic theory. In populations produced by intercrossing S1 lines, a generation of sib‐pollination within intercross progenies increased the esimate of test cross variance from 0.046 to 0.141 where an increase of 50% was expected. Where S2 lines had been intercrossed, a generation of inbreeding increased test cross variance from zero to 0.172. It was concluded that a generation of inbreeding prior to initiation of a new cycle would be a worthwhile step in recurrent selection programs.