92 E. CARLETON MACDOWELL 



more complete removal of these in successive generations by 

 selection will permit the numbers of bristles to gradually increase. 

 However, when all the flies become homozygous for the absence 

 of the accessory restricting factors, no further increase by 

 selection could be expected. Since there were no greater irregu- 

 larities in the environment apparent after, than before, the sixth 

 generation, and since flies from the fifth, sixth and seventh 

 generations were being raised on the same lot of food and at the 

 same time, the other interpretation of the failure of selection 

 becomes very weak. And when the evidence showing that there 

 is apparently no genetic difference in the later generations be- 

 tween flies of different extra bristle grades, is reviewed, including 

 as it does the weighty finding that seemingly norma' parents 

 may produce higher grade extra children than preceding high 

 selected ancestors and the conclusion that after nine selections, 

 high and low bristle grades give the same result in crosses, when 

 all these facts are born in mind, it will be realized that the 

 interpretation of the results by accessory factors seems in closer 

 agreement with the facts than does the alternative hypothesis. 



The hypothesis of accessory factors needs no elaboration or 

 change to explain the phenomena that accompany crosses with 

 a normal race. The main factor for restriction keeps the num- 

 ber of bristles down to four whenever it is present, so the ac- 

 cessory factors can only be detected when the main one is absent. 

 For this reason a simple Mendelian ratio may be found. From 

 the selected flies some if not all the accessory factors have been 

 removed. These would be present in the normal race used in 

 the cross, and, due to the segregation of these independent 

 restrictors, one would find in Fo among the flies lacking the main 

 restrictor, all combinations of the accessory factors, forming 

 groups with various restrictive powers from strong to weak. 

 The strongly restrictive groups would make the bristle numbers 

 lower than in the uncrossed flies under the same conditions, 

 while the weakly restrictive groups would make sUght or no 

 modification. This would result in an increase in the propor- 

 tions of flies with few extra bristles, yet the high bristle grades 

 would still be found. This is shown to be the case in figure 2, 



