468 FRED H. HULL 



suggested before that breeders of self-fertilized crops might find greater effi- 

 ciency in more frequent recombinations. It was to emphasize these considera- 

 tions that the term recurrent selection was introduced. The sense of recurring 

 back to the same tester was never intended. 



Breeders of open-pollinated corn need to save no more than 1 ear from 

 500 or more to plant the same acreage again. If selection is only 20 per cent 

 effective, the net effect in ten years is {yqY^. The number of corn plants 

 grown in the world in one year is roughly (10)^'. In 100 million times the 

 world acreage of corn there might be one plant as good as the farmer's whole 

 field after he has done 10 to 12 years of recurrent- selection. That this seem- 

 ingly fantastic theoretical concept is essentially correct is supported very 

 well, I think, by results of selection for oil and protein of the corn kernel in 

 the well-known Illinois experiments and in many other less well documented 

 cases with animals, too. East has proposed that selection for oil and protein 

 in corn might be more efficient with inbred lines. However, East proposed 

 that Si lines from the selected ears after chemical analyses be recombined for 

 another cycle of selection. He employed inbreeding only to avoid open-pol- 

 lination of the ears to be analyzed. It is unthinkable that East meant to pro- 

 pose that selection within and among inbred lines for oil or protein without 

 recurrence of selection should be the more effective process. 



Open-pollinated corn varieties of 50 or 30 years ago were actually pretty 

 good, in yield and in many other respects. The selection differentials by which 

 they were isolated were probably enormous. Nevertheless, specific combina- 

 tions of inbred lines are sometimes 20 to 30 per cent above the varieties in 

 yield. That this gain is mainly due to higher frequencies of dominant favor- 

 able genes in the elite inbred lines isolated from only a few hundred without 

 recurrence of selection is really inconceivable. 



A single corn plant in the variety is a product of two gametes. An Fi of 

 two homozygous lines is a product of two gametic types. The plant and the 

 Fi are genetically the same in mean, variance, and expectation of homo- 

 zygosity in advanced generations as well as the first. It should not be diffi- 

 cult, if asexual propagation were possible, to isolate from the single plants 

 clones that are easily superior to the present elite Fi's. That the reservoir of 

 specific combinability in corn is far from exhaustion by present hybrids is 

 evident in comparisons of Fi's with the range of individual plants in varieties. 

 The animal breeder may look upon a family of full sibs (from four grand- 

 parental gametes) as a double cross of unselected but homozygous lines, for 

 a rough estimate of possibilities with hybrids. But, aside from that, the 

 breeder of open-pollinated corn was selecting among specific combinations of 

 two gametes the same as in selection among Fi's. Continued selection within 

 varieties might have degraded gene frequency below (1 + k)/2k at any locus 



2. Cf. Huxley, Genetics in the 20th Century, p. 595. "Recurrent selection," natural or 

 artificial, is designed to multiply improbabilities; requires heritability in the strictest sense. 

 Selection among inbred lines may go on and on without "recurrence." 



