W. E. Castle 295 



races studied. If there were two major factors involved in this group and 

 part of the mothers were, as Punnett suggests, heterozygous, the varia- 

 tion curve should show two distinct peaks, which it does not. Punnett 

 admits the genetic purity of the White Dutch race with which Dark 

 was crossed, although the total number of individuals is smaller and the 

 variation figure is one-sided. The evidence for the genetic purity of 

 the Dark race is even stronger. 



2. The second objection is that my one-factor hypothesis calls for the 

 recurrence in F^ of the same t3rpes as occurred in Pi and F^ but not for 

 darker animals, whereas Punnett's hypothesis calls for the occurrence 

 through factorial recombination of darker animals (" a small proportion 

 of animals tending towards complete pigmentation "). Now the actual 

 result here confirms my view and disproves Punnett's. Not a single 

 animal in F^ was darker than animals of the uncrossed Dark race, and 

 the proportion of low-grade Dark animals was less among the extracted 

 than among the uncrossed Darks. I think Punnett must have had in 

 mind the F^ of a different cross when he raised this point, which is 

 really decisive in favour of my view. 



3. The third objection is that, if Dark and White are allelomorphs, 

 when Fi is back-crossed with White, " the two groups resulting should 

 be sharply separated and their mean grades should be close to those of 

 the parental stocks." Such is actually the case. Punnett apparently is 

 under the misapprehension that the segregated groups, on my hypothesis, 

 should be pure Dark and pure White respectively, whereas they should 

 be (aside from modifications due to minor factors) similar to F^ and pure 

 White respectively. For if the heterozygote, D W, is back-crossed with 

 WW, it is obvious that the two classes of zygotes to be expected are DW 

 (not DD) and W W respectively. As regards the size of the two segregated 

 groups, they number 65 and 64 respectively with one individual in 

 doubt and not assigned to either group. Certainly this is a very good 

 1 : 1 ratio. As regards the point that the " mean grades should be close 

 to those of the parental stocks," the mean grade of one group was 7*04 

 (as compared with 7"28 for F^) and that of the other group was 15"56 

 (as compared with 16"25 for uncrossed White, and 14"40 for White 

 extracted in F^. These are surely very good agreements. In range the 

 two groups agree exactly with the ranges respectively of F^ and of 

 White extracted in Pg- The groups moreover are monomodal as well 

 as equal in size. In short there is no single criterion for monohybrid 

 Mendelian segregation which is not perfectly met by this case. On the 

 other hand if we apply a two-factor or three-factor scheme to the case. 



