DOMINANCE AND OVERDOMINANCE 295 



mum amount of para-amino benzoic acid than either homokaryon. Other 

 cases, less known biochemically, may be similar. 



I think that it is doubtful whether such a system would persist for long 

 evolutionary periods. Alleles of intermediate productivity could arise and 

 replace the originals. Also modiiiers altering the expression of the homozy- 

 gotes would have considerable selection pressure. Or if the alleles were anti- 

 morphic, the situation might be resolved by duplication, as Haldane (1937) 

 has suggested. It is significant that the system reported by Emerson is not 

 one which is ordinarily of importance, but acts only in the presence of the 

 sulfanilamide-requiring mutant. 



A form of gene action that appears more likely to account for instances 

 of overdominance is one in which the two alleles differ qualitatively or each 

 does something that the other fails to do. Instances of mosaic dominance 

 provide excellent examples. This has been demonstrated for the scute series 

 of bristle characters in Drosophila and for color pattern in beetles (Tan, 

 1946). Other examples are provided by the ^4 and R loci in maize. 



Similar examples of physiological mosaic dominance are found where the 

 heterozygote apparently produces something approximating — at least 

 qualitatively — the total effect of the two homozygotes. An example is rust 

 resistance in flax, where each strain is resistant to a certain rust but the hy- 

 brid is resistant to both (Flor, 1947). By the usual tests for allelism, the two 

 resistance factors are alleles. Another series of examples is found in the 

 blood group antigens in man, cattle, and elsewhere. In almost every instance 

 the heterozygote has all the antigenic properties of both homozygotes 

 (Irwin, 1947). The presence of both the normal and abnormal types of 

 hemoglobin in humans heterozygous for the gene for sicklemia provides 

 another example (Pauling, 1950). 



Many instances of overdominance may have a similar explanation. This 

 is the kind of action that East (1936) postulated in his discussion of heterosis 

 due to cumulative action of divergent alleles. It is not necessary that the 

 effects be completely cumulative; only that the net effect on the phenotype 

 be greater in the heterozygote than in the homozygote. Any system in which 

 the alleles act on different substrates to produce the same or different prod- 

 ucts, or convert the same substrate into different products — neomorphs, in 

 Muller's terminology — could result in overdominance. 



Any of the examples listed above may turn out to be closely linked genes 

 (pseudoalleles) rather than alleles. In most cases it is impossible to distin- 

 guish between these alternatives. If the overdominance effect is due to 

 linked genes, eventually a crossover should result in a situation where the 

 desirable effects could be obtained in a homozygous individual. If there are 

 position effects, it may be that no homozygous arrangement is as advanta- 

 geous as one which is heterozygous. Unless there are position effects, it does 

 not seem likely that heterosis due to pseudoallelism would persist for any 



