724 



THE AMERICAN NATURALIST [Vol.XLIX 



The number of these grandchildren is large enough to 

 leave no doubt as to the conclusion that no further regres- 

 sion attended extraction of the hooded character a second 

 time from the wild cross. The proportion of hooded in- 

 dividuals to non-hooded is also an unmistakable mono- 

 hybrid ratio, viz., 263 hooded to 759 non-hooded, or 25.7 

 per cent, hooded in a total of 1,022 individuals. 



This result indicates clearly the untenable character of 

 our provisional hypothesis to explain the altered grade of 

 hooded rats under selection and crossing, by invoking the 

 action of independent modifying Mendelian factors. No 

 evidence is forthcoming from further and more extensive 

 experiments that such modifying factors are concerned in 

 the result. It seems rather that the hooded character, 

 which is a mosaic or balanced condition of pigmented and 

 unpigmented areas, is slightly unstable. It oscillates reg- 

 ularly about a mean condition or grade, these oscillations 

 being not phenotypic merely but in part genotypic so that 

 selection brought to bear upon them is immediately and 

 continuously effective. 



There may exist cases of continuous variation purely 

 phenotypic, as that of Johannsen's beans seems on his 

 showing to be. In other cases phenotypic variations may 

 so largely exceed genotypic variations that it is difficult to 

 discover and isolate the latter, as has been Pearl's ex- 

 perience. But our experiments with rats show beyond 

 reasonable doubt that genotypic variation, as well as phe- 

 notypic, may assume a continuous form, and if it does no 

 one can question its further modifiability by selection. In 

 denying effectiveness to selection in the case of continuous 

 variation, it has been assumed, tacitly by DeVries and 

 expressly by Johannsen, that continuous variation is 

 wholly phenotypic. This assumption being disproved, 

 the pure-line theory which rests upon it lacks adequate 

 support. 



It seems strange looking backward that the idea should 

 have become so widely accepted that continuous or fluctu- 

 ating variations are wholly phenotypic. For a continu- 



