102 fV. E. Castle 



McCracken's paper, p. 756. "In 1907, and again in 1908, pre- 

 caution was taken to make a number of matings within each 

 brood. It was found that all the tested females furnished by a 

 few broods, particularly in 1908, were bivoltin-producing. Many 

 of the broods furnished uni vol tin-producing females only and others 

 furnished females part of whom were bivoltin-producing and part 

 of whom were univoltin-producing. In each case these females 

 were mated with males of similar ancestry." This is very clear 

 evidence of MendeHan behavior of the characters univoltinism 

 and bivoltinism. Had the author traced the descent through 

 individual broods throughout her experiments instead of lumping 

 them into series, I am confident she would never have characterized 

 the inheritance as non-Mendelian. Even without this, had selec- 

 tion been made continuously for univoltinism within the mixed 

 race, as was done for bivoltinism in one case, it can scarcely be 

 doubted that the percentage of univoltins would have increased 

 steadily, though probably less rapidly than bivoltinism in the 

 reverse sort of selection. This at any rate is what we 

 should expect if univoltinism is dominant. Compare Tables i 

 and 2. 



On the whole, notwithstanding the incompleteness of the data, 

 we are, I believe, justified in concluding that univoltinism is a 

 Mendelian dominant to bivoltinism. For when from a mixed 

 race produced by crossing, selection is made for either condition, 

 bivoltinism increases faster than univoltinism. The fact that 

 bivoltin mothers mav produce univoltin daughters when mated 

 with males of unknown character is entirely in harmony with a 

 Mendelian interpretation. It is unnecessary to assume a mysteri- 

 ous "pull of ancestry," a delayed "conjugation," or the "masking 

 of an anlage" for a series of years followed by its reappearance, so 

 long as a simpler explanation in line with established principles 

 of inheritance fully accounts for the phenomena observed. 



It may not be out of place to repeat that if one is to test fairly 

 in a particular case the Mendelian or non-Mendelian character 

 of inheritance, the line of descent must be known through individ- 

 uals, not through masses of individuals. The futility of the mass- 

 method of dealing with inheritance phenomena has been sufficiently 

 illustrated in the results of the biometric school in England. 



