74 



from bar-eye. The analysis of the factors involved has yielded the fol- 

 lowing results: 



1. Germinal diversity was present in the stock at the beginning of 

 selection. 



2. This germinal diversity was due to accessory unit factors or 

 genes and not to differences in the bar-gene. 



.3. New accessory genes producing somatic differences of small de- 

 gree have appeared during the course of selection. Some of these are 

 located in the autosomes. 



4. New accessory genes producing somatic differences of marked 

 degree have also appeared during the course of selection. These also are 

 autosomal. 



5. Reverse mutation in the bar gene causing a return to the original 

 full-eye both somatically and genetically was observed several times. 



Original Germinal Diversity. 

 That germinal diversity was present at the beginning of the experi- 

 ments is indicated by the pronounced effect of the early selections. 

 Crosses between the high selected lines and the low selected lines show 

 that the factors causing the difference are not sex-linked as is the main 

 bar factor. This absence of sex-linkage shows that the difference be- 

 tween high and low lines can not be due to original diversity in the bar 

 gene nor to accessory factors originally present in the sex chromosomes. 

 The factors involved must be in the autosomes. Such differences in auto- 

 somal factors might have been present in the parental full-eyed stock 

 from which the bar was derived. They would then have been trans- 

 ferred to the bar-eyed stock at the time of its formation, which involved 

 not only change in the bar gene in a single male but also the crossing 

 back with a full-eyed female to produce the bar-eyed stock. 



Germinal Changes of Small Degree. 

 That the original diversity is not a sufficient explanation of the 

 effectiveness of selection and that germinal changes continued to occur 

 during the pi'ogress of selection in some of the lines is indicated by the 

 continued effect of selection in these lines for many generations. It is 

 highly improbable that a sustained effectiveness of this kind could have 

 lasted for twelve generatioi^s, as in line V, merely as a result of the con- 

 tinued sorting out of an original diversity without additions to the diver- 



