THE INTERACTION OF GENES: THE EFFECTS 167 



both disadvantageous (otherwise they could not be mutants but would 

 have become wild-type), are equally or unequally disadvantageous. The 

 effect of the dosage compensation system must, he argues, be to make 

 the effects of the wild- type genes the same in the two sexes. Now a 

 change from one to two doses of a dominant gene appears to have no 

 effect in the female. But, if Muller is correct, there must actually be a 

 difference between the homozygous wild-type and the heterozygote. 

 He suggests that the difference is one of variability. The homozygote is 

 well on the horizontal part of the curve relating phenotypic effect to 

 dosage (see Stern's curve for bobbed), while the heterozygote is near 

 the inclined part of the curve, where the phenotypic effect probably 

 varies with chance environmental influences as well as with dosages 

 changes. This behaviour of the wild-type m.ust have been evolved 

 through the selection of the appropriate modifiers, which (i) give one 

 dose of the gene an effect near the horizontal part of the curve, and 

 (2) give two doses an effect well on the horizontal part. It should be 

 noticed that (2) involves a restriction of the effect of higher doses; it is 

 apparently not usually advantageous simply to allow the homozygote 

 to produce double the minimum effect which can be allowed to the 

 heterozygote, but is better to bring them both to the optimum. 



The argument is therefore as follows : We can see that one dose of a 

 hypomorphic gene in a male is brought to have the same effect as two 

 doses in a female; therefore we must assume that the same thing 

 applies to the wild-type allelomorphs. One dose of a wild- type gene in 

 a female appears to have the same effect as two doses, but this cannot 

 be true; two doses in a female or one compensated dose in a male 

 must be less variable than one dose in the female. 



Exactly the same considerations apply to autosomal wild-type 

 genes, except that here the compensating mechanism need not he in 

 the same chromosome. But again it is presumably advantageous to get 

 the homozygote well on to the horizontal part of the curve so as to 

 limit its susceptibility to environmental variations, and this automati- 

 cally brings the heterozygote nearly or quite to the same apparent 

 effectiveness. The suggested mechanism therefore involves the selec- 

 tion of modifiers so as to make the wild-type gene dominant.^ That 

 selection of this kind may affect the wild-type gene as well as the 

 modifiers is indicated by the fact that two known wild-type allelo- 

 morphs of white, found in different (American and Russian) races of 



^ For an account of dominance modifiers (acting on vestigial in D, melano- 

 gaster), see Goldschmidt 1937. 



