VARIATION AND MENDELIAN INHERITANCE 371 



tions, as we pass southward from Humboldt to Sonoma County. 

 In fact, the collection from Duncan Mills, near the mouth of the 

 Russian River, is about intermediate between 'gambeli' and the 

 more extreme 'rubidus' type of Humboldt County. Neverthe- 

 less, it is a significant and striking fact that the 'gambeli' mice 

 from the vicinity of CaUstoga, in Napa County, much more 

 closely resemble those from La Jolla, 500 miles distant, than they 

 do the 'rubidus' from Duncan Mills, only twenty-seven miles 

 away. 



To a certain extent, and for certain characters, the gradations 

 here considered follow a geographic — and also a climatic — se- 

 quence. The degrees of difference in the characters are, however, 

 by no means proportional to the geographic intervals between 

 the 'races,' and there are other incongruities which greatly com- 

 plicate the situation. Thus, while there is strong evidence for 

 some sort of causal relation between these local differences in the 

 mice and some element or elements in the physical environment, 

 it is not possible to state with confidence at present just what 

 these elements are nor how they operate to modify the organism. 



The order of gradation for certain of the characters agrees 

 with that for certain others. Thus the order for tail length, 

 foot length, and tail stripe is similar, though not identical, while 

 the order for general color and that for skull length each presents 

 some significant points of resemblance to that for the other 

 characters just named. ' 



Certain characters, on the other hand, present a totally dif- 

 ferent arrangement. Thus ear length and the length of the 

 pelvis (innominate bone) offer gradations which bear no relation 

 to one another or to those of the other characters. Again, it 

 must be emphasized that there are well-marked differences even 

 among those characters which approach one another most nearly 

 in their order of arrangement. In respect to tail length, for 

 example, the Berkeley and Victorville races do not present sig- 

 nificant differences, while in regard to the width of the tail 

 stripe they stand far removed. 



Coefficients have been computed, showing the extent of the 

 correlation among certain of these characters within each single 

 race. This, needless to say, is another way of ascertaining 



