360 



DON C. WARREN. 



has made the difference between the two strains, the difference 

 should be most evident in the later generations of the experi- 

 ment. This is not true, for a moment's calculation will show that 

 the last three generations in each strain do not show as extreme 

 an average ratio in the desired direction as those which preceed 

 tehm. Although the difference between the totals of the two 

 strains may be sufficiently large to be considered significant, it 

 seems more probable that it is a chance variation, especially since 

 the first experiment showed a similar variation in the direction 

 opposite selection. 



SELECTION EXPERIMENT III. (Warsaw Stock.) 

 In order to make the work still more conclusive, selection was 

 attempted upon a third stock. This stock was collected at 

 Warsaw, Indiana, and kept as a stock culture in the laboratory 

 for about six months before the experiment was started. 



Here forty-two pairs were mated from among the offspring of 

 a single pair and the extremes of these matings were used as the 

 parents of the high and low strains. 3 There were 11,190 flies 

 examined before the selection began, 5,448 males to 5,742 females, 

 a ratio of i: 1.0539. The counts of these matings are given in 

 Table VIII. 



TABLE VIII. 



Showing Origin of the "High" and "Low" Strains of the Warsaw Stock. 



Grand Total 5,448 cf's to 5,742 9 's 



1 From this mating originated the "high" strain. 



2 From this mating originated the "low" strain. 



3 It will be seen in Table VIII. that the pair used as the parents of the low 

 strain did not produce the most extreme ratio in that direction. This mistake was 

 due to a mathematical error which was not checked up until it was too late to 

 rectify the experiment. 



