470 



THE AMERICAN NATURALIST 



[Vol. LIV 



not an arbitrary one is shown by the bimodal nature of 

 the curve (Fig. 1, solid line), which divides itself near 

 50 per cent.; by the closeness with which the actual as- 

 sortment of W and w (assuming the lower limit of W 

 spotting as 51 per cent.) approaches the expected and by 

 breeding tests of spotted mice slightly more and slightly 

 less than 50 per cent, white. Those more than 50 per 

 cent, white proved to be black-eyed white and those less 

 than 50 per cent, white proved to be piebalds. 



The mode of the black-eyed whites in this distribution 

 is at 95-86 per cent, white, although when all other black- 

 eyed whites are added (Table II, Cross 8, Fig. 1, broken 

 line) the main mode is found to be at 90-86. The range 

 is from 100 per cent, to 51 per cent, and the mean of all 

 black-eyed whites is S:].5 per cent, ± .4 with a standard 

 deviation of 11.4 per cent. ± .3. 



This establishes a wider range of variability for the 

 black-eyed white variation than has been current hereto- 

 fore, by the addition to the distribution of the classes less 

 than 80 per cent, white. It will be shown later in the sec- 

 tion on modifying factors that these darker classes are 

 not merely somatic fluctuations in the expression of the 

 gene for black-eyed white (W) but represent genetic 

 variations. They may be regarded as sub-types of black- 

 eyed white. 



The piebalds produced by the cross of black-eyed white 

 with piebald are represented by that part of the solid line 

 (Fig. 1) lying between 50 per cent, and per cent. Un- 

 like black-eyed whites, piebalds are not always distin- 

 guishable on the basis of dorsal white alone, for some pie- 

 balds have no white at all dorsally. In such cases the 

 ventral white must be used as a criterion, and of the pure 

 piebalds produced in this cross none had less than 12 per 

 cent, of ventral white and this was used as the limit of 

 piebald variability. All animals with 12 per cent, or 

 more of ventral white and with from to 5 per cent, of 

 dorsal white were placed in the 0-5 per cent, class. The 

 correctness of this classification can be inferred from the 



Piebald 



