142 EDWIN CAKLETON MACDOWELL 



Accessory factors vs. one variable factor 



Although the evidence was^ against it, the possibihty was ad- 

 mitted'* of applying to the first eleven generations of selection 

 an alternative hypothesis involving one ever-varying factor, 

 similar to the hypothesis adopted for the selection experiments 

 of Castle.^ The hypothesis of accessory factors has formerly 

 been discussed in full ; its application to all the data will now be 

 made, and then the impossibility of so applying the alternative 

 hypothesis, without numerous subsidiary hypotheses, will be 

 shown. 



The germ plasm of the original pair of flies, besides having a 

 factor that permitted more than the four normal bristles to 

 appear, had accessory factors that influenced the numbers of 

 extra bristles that developed. Some of these accessory factors 

 were in a heterozygous condition so that, due to their segre- 

 gation, the germ cells of these flies were of different kinds. The 

 effect of these gerriiinal differences was strong enough to make the 

 somas of the offspring fairly good relative indices of their ger- 

 minal constitutions, even though the vitiating influences of en- 

 vironment were active. Selection of extreme variates moved 

 the means in the direction of the selecting, raising them or 

 keeping them low. Generation by generation the selecting 

 reduced the heterozygosity of the accessory factors, so that the 

 segregation was reduced and the germ cells that were produced 

 were less varied. Shortly the differences between germ cells 

 in an individual disappeared or at least became so small that 

 their influence was overpowered by the influence of environ- 

 ment. Variations in bristle number then had no relation to the 

 germinal condition. As long as brother by sister matings were 

 made, the high and low grade flies produced similar offspring; 

 no longer was it possible to move the means either up or down 

 by selection. But as soon as these flies were crossed with 

 normals, the distribution of the extra bristles was lowered; the 



4 MacDowell : loc. cit., p. 91. 



^ Castle, W. E.: Experiments in mass selection. Amer. Nat., vol. 49, no. 588. 

 pp. 722-723. 



