54 SUMiMARY OF CURRENT RESEARCHES RELATING TO 



taken together in that generation. For about eight generations the 

 means rose ; following this were two periods not comparable with each 

 other, within neither of which was there any evidence of further 

 advance. Continued selection did not produce any high extremes that 

 were not obtainable near the beginning of th6 experiment. The range 

 of variation changed only very slightly ; the low limits being most 

 frequently at or 1, the high limit at I) for the females, at 7 for the 

 males. The standard deviations rose and fell together with the means ; 

 as the means of the females are higher than those of the males, 

 so the standard deviations of the females are higher than those of 

 the males. These relationships do not hold true when the complete 

 yields of the bottles are not included (generations 33-49). Changes 

 in the means of the parents are not accompanied by changes in 

 the means of their offspring, except at the beginning of the experi- 

 ment. 



By selecting low-grade parents from the second generation of the 

 extra-bristled race it was found possible to establish a race of flies which 

 had markedly lower means than the high-selected race. 



By selecting low-grade flies from the fifteenth generation of the high 

 race and continuing to select for low grades, it was found impossible in 

 eight generations to establish a race that was distinguishable from the 

 high race. The attempt was repeated, starting from the twenty-sixth 

 generation of the high race, and continued for six generations with 

 similar results. Keturn selection does not reverse the progress made 

 by the advance selection. Flies with high and low bristle grades 

 appear to have very similar offspring. 



By selecting low-grade parents from the Fj of a cross between 

 normals and flies from the sixteenth generation of a high race, a low 

 race was established (extracted low). One selection was sufficient to 

 establish this race as distinct from a high race ; for four generations 

 the curves of the parents and offspring wre parallel, and after that 

 completely independent ; for four generations the low selection continued 

 to lower the means ; except in the first few generations the curves of 

 the progeny rise and fall in harmony with the curves of the high race, 

 when families raised at similar times are compared ; besides being 

 lower than the high race, the variability of the extracted race is less 

 than that of the high race ; in response to the same improvement in 

 conditions it does not advance as far. 



Comparing the different races, it is found that, no matter what the 

 percentage, they all exhibit high points and low points at the same 

 time. This means that environment is accountable for the variations 

 in most of the generations. But the initial rise in the high race was 

 not due to environment, as this rise resulted in a genetic change in 

 the race. 



The supposition of a single varying factor to explain the above 

 results cannot be justified ; it would require numerous other assumptions. 

 All the results are simply explained on the assumption that there were 

 genetic differences present among the original flies with extra'bristles ; 

 that these genetic differences (or genes) are entirely independent of the 

 main factor that occasions the monohybrid ratio in crosses with 

 normal flies. 



