BRISTLE INHERITANCE IN DROSOPHILA 95 



selections have been made. Another difference that becomes 

 illuminating in the light of these facts is that the flies have been 

 inbred absolutely, while in the rats the inbreeding was of a fairly 

 low degree. Return selection proved equally effective in the 

 piebald rats. If the interpretation of accessory factors be ac- 

 cepted for the results of crossing the rats, it is evident that at 

 least part of the success of selection was due to the sorting out 

 of these accessory factors. Now the fact of a successful return 

 selection can not be sighted in this case to prove the theory that 

 a factor has been modified, since as long as selection was showing 

 steady progress, it should be possible to start a successful return 

 selection. As long as the original selection was progressing 

 there still is evidence of some heterozygosis of the accessory 

 factors. This would afford a basis for the return selection. At 

 this time any positive statement would be premature, but the 

 results given by an attempted return selection of the extra 

 bristles after the upward progress had seemed to stop, appear 

 to indicate that this return selection is ineffective. Finally the 

 more complicated hypothesis of Pearl ('12) supported by his 

 extensive investigations on fecundity in fowls, must be sighted 

 as one based fundamentally on the conception of multiple factors. 

 Taken then on their own merits, the results presented in this 

 paper do not give critical evidence in support of either the 

 hypothesis of modification or of accessory factors. However, the 

 failure of selection after success for six generations and the prob- 

 able genetic equality of the various bristle grades in the later gen- 

 erations, seem to bear the balance strongly towards the hypothesis 

 of accessory factors. Taken in the light of much of the recent 

 genetics investigations, many of which have close theoretical 

 similarities, it is almost impossible to avoid the conclusion that 

 the interpretation of accessory factors is the more probable. 

 Besides, this hypothesis affords a more thinkable mechanism 

 and is more readily understood and tested. For these reasons 

 this interpretation has been adopted at least as a working hy- 

 pothesis upon which to base further investigations. Experiments 

 are already under way to attempt the isolation of accessory 

 factors and, by crossing, to prove unquestionably their existence. 



