294 JAMES F. CROW 



another explanation is selective superiority of heterozygotes (Haldane, 1950), 

 at least for those factors of importance in heterosis. 



If yield is determined entirely by dominant factors, the correlation be- 

 tween inbreds and their hybrids should be positive. If it is due to over- 

 dominant loci, the correlation should be generally positive, though there 

 would be negative correlations between yield of hybrids and inbreds when 

 the other inbred is constant and high yielding. If both factors are involved 

 and overdominant loci are relatively important in hybrids while dominants 

 are important in inbreds, the correlation would approach zero. The experi- 

 ence of corn breeders has been that selection for yield during inbreeding is 

 relatively ineffective, and that the correlation of hybrid with inbred yield, 

 though positive, is small. 



With overdominant loci the effect of a certain percentage increase in 

 heterozygosity is to cause the vigor to increase by a certain amount. De- 

 creasing the heterozygosity by the same percentage would cause a decrease 

 of approximately the same amount. On the other hand, with dominant loci, 

 making the original equilibrium population more heterozygous would cause 

 a very slight increase, whereas making the population more homozygous 

 would have a decreasing effect of a much greater amount. Therefore it is 

 easier to account for inbreeding depression by dominant loci than to account 

 for increase in vigor on hybridization above the level of a random mating 

 population. 



I should like to suggest the following interpretation of the effects of in- 

 breeding and hybridization: The deleterious effects of inbreeding and the re- 

 covery on hybridization are mainly due to loci where the dominant is favor- 

 able and the recessive allele so rare as to be of negligible importance in a non- 

 inbred population. Variance of a non-inbred population, and hybrid vigor 

 when measured as an increase over an equilibrium population, are deter- 

 mined largely by genes of intermediate frequency, probably mostly over- 

 dominants. 



OVERDOMINANCE AND GENE ACTION 



In order to have overdominance it is not necessary that the immediate 

 gene products of the heterozygote exceed in quantity or variety those of 

 either homozygote. At the level of the immediate gene product, or any inter- 

 mediate state, the effect of the heterozygote may be intermediate between 

 the two homozygotes and still result in a greater final result. Any kind of 

 situation in which something is produced for which an intermediate amount 

 is optimum could be such that the heterozygote is nearer this optimum than 

 either homozygote. 



A model for such cases is found in the sulfanilamide-requiring strain of 

 Neurospora reported by Emerson (1948). When this mutant is present the 

 heterokaryotic state of the suppressor gene results in more nearly the opti- 



