Selection upon Meiidehan Character 



191 



but are deficient. On the other hand when selection is made for 

 U, there is observed no increase in the percentage of univoltins. 

 Such increase we should expect in a series of successive selections 

 for U, but not of necessity as a result of a single selection. Thus 

 in series E following a single selection of U females, we expect a 

 smaller percentage of U females than in the parental series, A'. 



TABLE 3. 

 Percenlas^cs of univoltiti females observed in the several series and their relations to the expected Mendelian 



percentages . 



*i\lissMcCracken's "Table of descent" is not in agreement with her text as regards the derivation of 

 series E. I have assumed the correctness of the text, that series E is derived from the univolts (not 

 the bivolts as shown) of series A'. 



But we do not expect a second selection for U to be attended by a 

 further decrease in the percentage of U females, as is observed in 

 series K, and series L. This, as Miss McCracken observes, is a 

 matter deserving explanation. It is unfortunate that we do not 

 know from what particular broods the males were taken in each 

 series. Without such knowledge a complete Mendelian analysis 

 is impossible. If the parents of K and L happen to have been 

 chosen from broods in which bi vol tin individuals predominated, the 

 high percentages of bivoltins in those series are fully explained. 

 On this point it is sufficient to quote a paragraph from Miss 



