452 FRED H. HULL 



tion for both traits together each time, with normal distribution, would pro- 

 vide theoretical recovery in two generations at the rate of 1 per 400 or less. 

 Multiplication of selection differentials in recurrent selection was sufhciently 

 understood at the inception of hybrid corn. Nevertheless, hybrid corn has 

 been developed with virtually no use or benefit from recurrence of selection. 

 Hybrid corn is almost wholly an empirical development, but I think we may 

 now consider applications of genetic science to improve the process. 



Recurrent selection (Hull, 1945a) was meant to include reselection genera- 

 tion after generation, with interbreeding of selects to provide for genetic 

 recombination. Thus, selection among isolates, inbred lines, or clones is not 

 recurrent until selects are interbred and a new cycle of selection is initiated. 

 Recurrent selection for specific combinability would seem to require a special 

 breeding plan to provide heritability through successive cycles. 



Shull's original plan for developing superior corn hybrids was designed for 

 maximum immediate employment of specific combinability. Selection was 

 mainly among specific Fi crosses of lines which had been isolated and stabi- 

 lized by inbreeding, thus providing repeatability of crosses. This plan was 

 consistent with a theory of heterozygosis of a degree here termed over- 

 dominance. Shull's plan did not involve recurrent selection to accumulate 

 higher frequencies of favorable genes in successive cycles. 



The apparent heterozygosis which Shull proposed to use was interpreted 

 by Jones about ten years later as the expectation of repulsion phases of ran- 

 dom linkages of dominant favorable factors and recessive less favorable al- 

 leles. This interpretation was particularly attractive because it seemed to 

 eliminate any necessity of accepting overdominance. Overdominance is a 

 contradiction of the time-honored principle that purity of blood is to be 

 sought and maintained. Vigor was no exception to the old principles of like 

 begets like and breed the best to the best. Moreover, the postulated linkage rela- 

 tions would appear to be inevitable where many loci are involved. 



In the decade following appearance of the Jones hypothesis, most corn 

 breeders began more intense selection for vigor within and among lines during 

 the inbreeding process, and selection among lines for general combinability. 

 Most of the very considerable success of hybrid corn came quickly after these 

 modifications of Shull's method were adopted. Selection within and among 

 inbred lines to improve frequencies of dominant favorable factors became the 

 guiding principle for developing superior hybrids of corn, other crops, and of 

 livestock. Selection among specific Fi crosses was retained as a final step, but 

 with very little verbal emphasis. 



Initial successes with hybrid corn (which so far have not been greatly sur- 

 passed) were obtained with inbred lines which were, for the most part, iso- 

 lated directly from the open-pollinated varieties. Corn breeders then had at 

 least two alternatives for further work. Empirically, the choice might well 

 have been to continue isolation and testing of additional new lines from the 



