Five cycles of reciprocal recurrent selection in two synthetic varieties of maize (Zea mays L.) were summarized. Yields of the varietal cross and one parental variety were increased significantly, but the yield of the other variety de, creased slightly. Expected gain in the varietal cross yield, calculated from pooled heritabilities and selection differential, was 7.2% per cycle, but observed gain was only 1.7% per cycle.In the sixth cycle, selection among S1 plants was substituted for selection among S0 plants. Selection within the randomly mated varietal cross was initiated for comparison with selection within the two parental varieties individually. Estimates of genetic variance within the parental populations in the sixth cycle were higher than for the three previous cycles, and genetic variance was higher within the randomly mated varietal cross than within the parental varieties.

No abstract

The effectiveness of half-sib selection with a doublecross tester and Si selection per se was evaluated following four cycles of selection in BSK (the Krug Hi I syn. 3 strain of 'Krug Yellow Dent'). Five populations, BSK CO, BSK(HT)C2, BSK(S)C2, BSK(HT)C4, and BSK(S)C4, were selfed, sibbed, and crossed in a diallel series. They were also test-crossed to four single-cross testers.The average of the four single-cross testcross yields of the BSK (S) populations increased 10.6% during the four cycles, and the BSK(HT) populations increased 5.7%. The selfing series populations performed better than the testcross series populations when selfed, as indicated by a 38.7% increase for the BSK(S) bulk S^'s compared with a 12.0% increase for the BSK(HT) bulk S/s. The populations per se increased 16.3% in the selfing series and 6.3% in the testcross series. The BSK(HT) x BSK(S) crosses exhibited heterosis, which indicated that the two methods developed populations that differed in gene frequency.This study has shown that the mean yield and the general combining ability of BSK were significantly improved by four cycles of selection with both methods of recurrent selection, but Sj selection per se was more effective. Additional index words: Zea mays L., Grain yield, Population improvement, Test-cross evaluation, Sj line per se evaluation.A BASIC concern in maize (Zea mays L.) breeding is the identification of superior genotypes. This often is difficult because of the many factors, genetic and environmental, that affect characters under selection. Various methods of visual selection during inbreeding combined with evaluation by topcrossing have lacked efficiency. Recurrent selection was de-

Four years of reciprocal recurrent selection has produced rapid improvement in one of the parental varieties (‘Ecuador 573’), in the variety cross, and in a commercial topcross maize (Zea mays L.) hybrid. The improved strain of Ecuador 573 is now used as the male parent of the current commercial hybrids, ‘H611C’ and ‘H613C,’ and they yield approximately 25% more than the original versions.Estimates obtained from seven years of the selection trials showed no changes in genetic variances for yield. Hence, this rate of improvement is not expected to change in the near future. Progress from modified ear‐to‐row selection was greater in the introduced Ecuador 573 variety and in KCA, (‘Kitale II’ ✕ Ecuador 573)syn‐4, than in the local Kitale II variety. No improvement was detected in KCA from mass selection. An increased number of ears per plant was associated with yield improvement by reciprocal recurrent selection and modified ear‐to‐row selection. Selection for lodging resistance and blight resistance in the breeding nurseries resulted in slight improvements in the varieties under reciprocal recurrent selection and in the variety cross. Greater attention to ear height and lodging resistance is required to reduce harvesting losses in the commercial maize crop in Kenya.