118 



EDWIN CARLETON MACDOWELL 



nation of this exception is the claim that the higher averages 

 in this period are due to incomplete counts and not to a modi- 

 fication due to selection. 



eCALE OFJvlEAISIS 



EXTRA BRISTLES 



HIGH RACE 



SCALE OF STANDARD DE?VlATIONS 



/ GENERATIONS 



1—1 \ \ \ \ I I I I I I I I I 



2 4 6 8 10 12 14 16 18 20 22 24 28 30 32 34 36 38 40 42 44 46 48 



SCALE OF MEANS 

 6 _ 



EXTRA BRISTLES 



HIGH RACE 



GENERATIONS 



SCALE OF STANDARD DEVIATIONS 

 _ 2.0 



J \ \ I L 



J \ \ L 



J \ L 



_ L2 



2 4 6 8 10 12 14 16 18 20 22 28 30 32 34 3b 38 40 42 44 46 48 



B 



Fig. 5 Standard deviations in the high selected race compared with the 

 corresponding means. It must be noted that the standard deviations are plotted 

 on a scale five times as large as that of the means. Generations 24, 25 and 26 

 have been omitted on account of their small numbers. 'A' is based on the sons, 

 'B,' on the daughters. A close parallelism is demonstrated between these con- 

 stants. All flies from each bottle were not counted in generations 33-49 in- 

 clusive; this probably accounts for the break in this relationship that appears 

 in those generations. 



