GENE INTERACTION IN HETEROSIS 323 



alleles but which, in the course of evolution or under a changed environment, 

 have been displaced, to a greater or lesser degree, by still better alleles. 



4. The prevailing allele at a given locus has reached its pre-eminent posi- 

 tion through the sifting action of natural selection over many generations. 

 Given a stable environment, further improvement, through mutation, at that 

 locus would long since have materialized if the chances for such improvement 

 were high. It is not strange that random mutation should only rarely be able 

 to produce a superior new allele. Nevertheless, once the possibilities for im- 

 provement through recombination of existing genes have been exhausted, 

 further evolutionary progress will be contingent upon just such an event, 

 however rare its occurrence may be. 



5. Whether dominant or recessive, and whether in a naturally self-ferti- 

 lized or naturally cross-fertilized population, a substantially superior mutant, 

 once established in the population, is destined to increase in frequency and to 

 become the prevailing allele in the population. 



6. A deleterious dominant is doomed to eventual extinction. In a cross- 

 breeding population of sufficient size a deleterious recessive may persist in- 

 definitely, its incidence, except for random drift, being determined by the 

 balance between its elimination by selection and the rate at which it recurs by 

 mutation. 



7. The best allele for one environment may not be best for another envi- 

 ronment. The burden of less favorable alleles which cross-fertilized organisms 

 carry along generation after generation is not an unmitigated liability. It 

 serves as a form of insurance by providing a reservoir of adaptability to 

 changing conditions. 



ROLE OF LESS FAVORABLE ALLELES 



Turning now to the role of these less favorable alleles in the heterosis 

 phenomenon as manifested in naturally cross-fertilized organisms we may 

 formulate a second group of postulates: 



1. At many and perhaps at most loci, .la is as good or nearly as good as 

 A A, and both A A and A a are better than aa. 



2. There may be a few loci where aa is better than .LI or .la. This is par- 

 ticularly likely to be the case for loci affecting traits which are advantageous 

 under domestication, but disadvantageous in the wild under natural selec- 

 tion. 



3. There may, for all we know, be occasional loci where .LI' is better than 

 A A or A' A' (overdominance). 



4. There may be many regions in the chromosomes which behave as though 

 A A' were better than /L4 or A' A'. With deleterious recessive alleles in the 

 heterozygous condition at many loci, it seems almost inevitable that some of 

 these will be closely linked in the repulsion phase, as for example Ab/aB, 

 which in the absence of crossing over would behave as a single locus, the 



