Payne: Variations and Selection 29 



number down to normal. In order to account for the fact 

 that "extra bristles" at the beginning of his experiment 

 behaved as a Mendelian recessive, he assumed that this pri- 

 mary inhibitor had been eliminated. By continued selection 

 he increased the number of extra bristles. In order to ac- 

 count for this increase, he further assumed the presence of 

 a series of secondary inhibitors which keep the extra bristle 

 number reduced, and that when in the course of selection 

 these secondary inhibitors are eliminated the number of extra 

 bristles increases. The difficulty in this explanation, it seems 

 to me, is in getting rid of the inhibitors. This MacDowell 

 does not discuss. In fact, I see no way of getting rid of them 

 except by mutation, if the strain is homozygous for them. If 

 the strain is heterozygous for such inhibitors, then we ought 

 to find many more flies in the wild and in our mass cultures 

 with extra bristles. In this connection it is interesting to 

 speculate whether there are also primary and secondary 

 inhibitors which keep the number of bristles from going below 

 normal. Such an assumption would be as logical as the one 

 MacDowell makes. It would give, however, a complicated 

 mechanism for keeping a character constant and seems awk- 

 ward and cumbersome. 



Occasionally flies are found with only two or three bristles. 

 I have made an attempt at a minus selection from a male 

 with three bristles on the thorax. This male was mated to a 

 wild female. In there were 194 normal flies, four with 

 three bristles, and two with two bristles, or three per cent with 

 a reduced number. The flies with the reduced bristle number 

 were mated and gave in Fo 690 flies, 16 or 2.31 per cent of 

 which had three bristles. The others were normal. Again 

 the flies with the reduced bristle number were used as 

 parents. In Fg there were 2,688 normal flies, 93 with three 

 bristles, and three with two. This was 3.44 per cent show- 

 ing the reduced number. This strain was carried on for 

 six generations without any further increase in the reduction. 

 The per cent of reduced flies in F^ was 2.21, in F^ it w^as 2.69, 

 and in F^, 2.10. Unfortunately, at this time the strain died 

 out, all attempts to keep it going proving useless. Six genera- 

 tions may not be long enough to produce results in this case, 

 but I am of the impression that even if the strain could have 

 been continued, no marked results would have been produced. 



