HETEROSIS IN POPULATION GENETICS 155 



perimental conditions. It seems, however, that the study of selection in 

 favor of the heterozygote for single loci deserves more careful analysis. The 

 whole problem of heterosis for several genes affecting quantitative characters 

 will be solved, I think, only when the more simple cases of heterosis where 

 single gene differences are involved shall be cleared up. I have been lucky 

 enough to come across some useful experimental material for the purpose. 



For a number of years I have kept about one hundred different wild 

 stocks of Drosophila melanogaster coming from diflferent geographical locali- 

 ties. Such stocks were maintained by the usual Drosophila technique of 

 transferring about once a month some 30-40 flies from one old vial to a new 

 one with fresh food. About twice a year I look at the flies under the micro- 

 scope. Since all such stocks were wild type, no change by contamination was 

 expected, as these stocks were phenotypically alike. Contamination by mu- 

 tants kept in the laboratory could not have produced any appreciable result, 

 owing to the well known fact that both under laboratory and natural condi- 

 tions mutants are generally less viable than the normal type. To my sur- 

 prise, however, I happened to observe at two different times, in two different 

 wild stocks, that a fairly large number of the flies showed an eye color much 

 lighter than the normal. These two mutants proved to be indistinguishable 

 recessive alleles at the same locus in the third chromosome. The presence 

 of the homozygotes has been checked at different times over a number of 

 years. 



In the summer of 1947 while collecting flies in the wild for other purposes, 

 I found in the neighborhood of Suna, a small village on the western shore of 

 the Lake Maggiore, in Northern Italy, several individuals of both sexes show- 

 ing the same eye color. From these a homozygous stock for such mutant was 

 obtained. Crossing tests proved that it was another allele of the same locus as 

 the above mentioned. The occurrence of several individuals mutant for an 

 autosomal recessive within a free living population was remarkable enough. 

 But finding that the same gene was concerned as in the laboratory stocks, I 

 suspected that such a mutant might have a positive selective value, both 

 under laboratory and natural conditions. 



I began an experiment to check this point. Two populations in numerical 

 equilibrium were started, applying the method previously used by Pearl for 

 the study of population dynamics of Drosophila, described in detail else- 

 where (1947a). Sixteen light-eyed individuals, eight males and eight females, 

 were put together in one vial with sixteen wild type flies. The gene frequency 

 at the beginning of the experiment was therefore .5. Under the experimental 

 conditions the population reached an equilibrium in respect to the amount 

 of available food at about 700-900 flies per vial. After about twenty genera- 

 tions, assuming that each generation takes 15 days, the frequency of recessive 

 homozygotes was about 40 per cent. Assuming random mating within the 

 population, taking the square root of .40 one gets a gene frequency for the 



