BRISTLE INHERITANCE IN DROSOPHILA 119 



Discussion of the distributions 



The study of the frequencies of the 49 high selected genera- 

 tions, whether by the means, the extremes, or the standard 

 deviations, has failed, except in the first few generations, to 

 indicate any advance that may be attributed to selection. It 

 seems that breeding only high grade parents has not succeeded 

 in producing increasingly higher distributions, that new ex- 

 treme grades have not appeared after the continued selection 

 of the highest flies as parents. Other than genetic factors seem 

 to influence the character of the distributions, yet it remains 

 possible that a part at least of the fluctuations in the means may 

 be due to corresponding variations in the parents selected. 

 This possibility is considered in the following section. 



Comparison of parents and offspring 



The point of primary theoretical interest is not the ability, 

 or inability, to advance the means and raise the whole dis- 

 tribution of a race by selection. The crux of the selection prob- 

 lem is whether abmodal parents have abmodal children. Spe- 

 cifically it is, do parents with higher bristle grades produce 

 children with higher bristle grades? 



When the means of the offspring, which have already been 

 considered by themselves are compared with the means of the 

 parents that produced them, important evidence is forthcom- 

 ing. Referring again to figure 3, this comparison will be con- 

 sidered. The dotted line in this figure shows the means of the 

 parents selected. These have been weighted according to the 

 numbers of offspring produced by the various grades of parents 

 included in each mean. The means of the parents in each gen- 

 eration are plotted on the same vertical line as the means of their 

 sons and daughters. The basic numbers for these curves will 

 be found in table 1. 



One finds on making this comparison that, excepting the first 

 few generations, the means of the parents and their children 

 vary with great independence. The high peaks of the parents 

 curve are not accompanied by high peaks in the curves of their 



