146 F. B. SUMNER, M. E. MCDANIEL AND R. R. HUESTIS. 



Newcomb (1904), from a study of 7,896 families, concluded 

 that, after making allowance for a slight preponderance of males, 

 the sexes in each family were distributed according to chance. 

 There was no tendency toward an excessive production of fami- 

 lies consisting wholly of one sex. A consideration of multiple 

 births, however, based upon French and German statistics, showed 

 the existence of a pronounced tendency toward agreement in 

 sex on the part of twins and triplets. Biologists would have 

 confidently predicted such a situation, owing- to the familiar phe- 

 nomenon of duplicate ("identical") twinning, but Newcomb 

 strangely overlooked this explanation of his results and found 

 in them support for the idea that sex is determined by unknown 

 causes operating during development. 



RACE. 



In the following table, the mice of "pure" (i.e., non-hybrid) 

 stock 1 are grouped according to the geographic race (subspecies) 

 to which they belong. The La Jolla representatives of the sub- 

 species gambcli are here kept distinct from those coming from 

 the more northern localities (Berkeley and Calistoga). This is 

 partly due to the appreciable morphological differences between 

 these local races of gambeli, partly to the fact that the northern 

 representatives were nearly all born during the months of May to 

 August, a circumstance which may account in part for the ex- 

 tremely low sex ratio found in this group. 



In making a comparison between these races, we may reasonably 

 leave out of consideration the small group of Berkeley and Calis- 

 toga gambeli. The two main groups (numerically speaking), 

 sonoricnsis and La Jolla gambeli, differ from one another by an 



i Except P. maniculatus dubius and P. eremicus, for which the numbers 

 are too small. 



