298 F. B. SUMNER AND H. H. COLLINS 



records show that the offspring of two 'atypical' animals gener- 

 ally resemble their parents in this respect. ^^ 



It is not impossible that we are concerned merely with differ- 

 ences of the type which are commonly called 'phenotypic' or 

 'somatic' (i.e., non-hereditary). But such an explanation is 

 hardly consistent with the facts just cited. It seems more likely 

 that we have to do with the presence or absence of 'modifying 

 factors,' or possibly even with the existence of unstable factors, 

 or departures from the simple 'factorial' scheme of heredity. 



Table 3 gives the average and extreme values for the color 

 determinations of twenty-four adult pelages of the 'b' yellows. 

 About two-thirds of these were listed as 'typical' or 'nearly 

 typical,' the others as 'atypical.' 



From tables 1 and 3 it appears that the 'a' strain shows a 

 slightly higher percentage of white, while the 'b' strain shows a 

 slightly higher percentage of black, the values for 'color' being 

 nearly the same for the two. The significance of these differences 

 is doubtful. Of far more importance is the fact that the ratio 

 of red to green^^ jg distinctly higher for the '6' strain than 

 for the 'a,' the mean figure being 3.59 for the former and 3.17 

 for the latter. A study of the frequency distributions shows 

 that this difference is probably a real one, and indeed a casual 

 comparison of the skins reveals it to the eye. Further evidence 

 of such a difference was derived from the examination of living 

 specimens, many of which were not skinned. 



Hybrids between the 'a' and 'b' yellows 



Matings of these two strains resulted in all cases in offspring 

 which were listed as 'yellows.' Three 'a' females were mated 

 to two different males of the 'b' strain. The resulting twelve 

 F] hybrids appeared, on the whole, as intermediate between 



" The interruption of these studies above referred to is largely responsible 

 for these uncertainties. 



12 The values for red and green employed for this purpose are the excess of 

 each that remains after deduction of the amount which combines with the other 

 colors to constitute the 'white.' The lowest color-screen reading (in this case 

 the blue-violet) also serves to indicate the amount of white, there being no 'free' 

 blue-violet. 



