No. 583] ROLE OF ENVIRONMENT IN DROSOPIIILA 387 



Two principal obstacles delayed the formation of a 

 pure strain. The new character is a sex linked domi- 

 nant, 1 but both the heterozygous and the homozygous 

 condition overlap the normal typo which makes the selec- 

 tion of pure females difficult. Any male, however, that 

 shows abnormal abdomen at all is pure, for the charac- 

 ter is borne by the X chromosomes of which he has but 



The other obstacle was what at first appeared to be a 

 perpetual reversion of stock, seemingly pure, to the nor- 

 mal. So constantly did this occur, that, for some time, I 

 thought that I had an "ever-sporting" variety — one 

 that reverted to the normal without apparent provoca- 

 tion. I found, however, that the first flies that hatched 

 in the best-fed cultures were entirely abnormal, while 

 those that emerged later were less abnormal, until finally 

 those that emerged when the cultures were nearly at an 

 end were invariably normal flies. It seemed at first pos- 

 sible that such stock might be impure, and that the ab- 

 normal flies hatched sooner than the normal, but this view 

 was negatived bv the fact that normals hatch as soon as 

 do the abnormal flies. 



The one remaining possibility seemed to be that de- 

 velopment of the abnormal abdomen depended on some 

 definite condition of the culture — one that was present 

 when the food was fresh and the bottle wet, but which 

 disappeared as the food was used up and the bottle got 

 dry. I tested this hypothesis in many ways. Stock was 

 used that had been pure for nine generations. As a 

 bottle dried up an ever increasing proportion of normal 

 flies appeared. At intervals lots of flies were taken out 

 and put into new bottles where they were abundantly fed. 

 Their first progeny, as recorded below, shows that under 

 the new conditions the offspring were sometimes ex- 

 tremely abnormal irrespective of the general condition 



