IQO W. E. Castle 



If all the univoltin females employed in these twenty-four matings 

 had been heterozygous univoltins, while their mates were half of 

 them pure bivoltin, half heterozygous univoltin, then we should 

 have expected the offspring to be as four bivoltin: five univoltin, 

 or I :i^, instead of the observed i : 14. It seems probable, there- 

 fore, that the twenty-four univoltin females were not all heterozy- 

 gous, or else that the Y^ males mated with them were univoltin in 

 character to a greater extent than their sisters, the tested Fj fe- 

 males. The data given are insufficient for testing either hypothe- 

 sis adequately. In either case we should expect the subsequent 



generation to contain a mixture of univoltin and of bivoltin fe- 

 es 



males, as actually observed, but in what proportions they occur- 

 red would depend upon a number of contingencies. Concerning 

 these we are largely without information, so that no Mendelian ex- 

 pectations of much value can be calculated. Nevertheless I have 

 calculated one such set of expectations which is contained in 

 Table 3. It is based on the following contingencies: 



(i) That univoltinism is uniformly dominant over bivoltinism, 

 and that, therefore, all bivoltin females transmit that condition 

 only. But since the character of the male mate is in every case 

 uncertain, it is assumed (2) that the males are in every generation 

 of the same sorts as the females, and occur in the same proportions. 

 (3) The actual mates of each group of females are assumed to 

 have been such as one would obtain by random selection, that is 

 they are univoltin and bivoltin in the same proportion as the popu- 

 lation from which they are taken. If, for example, a group of 

 individuals contains thirty univoltins and twenty bivoltins, and 

 from this group five females are selected, it is assumed that 3 of 

 them arc univoltin and two bivoltin. (4) The F, offspring pro- 

 duced by the original cross are assumed to have been half pure 

 bivoltins, half heterozygous univoltins. (5) The "pure" U fe- 

 males, mothers of series A, are assumed to have been heterozy- 

 gous in one out of seven cases. 



It will be seen from Table 3 that selection for B is, in nearly 

 every case, attended by a reduction in the percentage of univol- 

 tins (increase in the percentage of bivoltins) as we should expect, 

 though this reduction is less rapid than we should expect. Con- 

 trary to Miss McCracken's view, the bivoltins are not in excess. 



