THE AMERICAN NATURALIST [Vol.LII 



Now it is of interest to note that the F 2 generation, con- 

 sisting of 40 specimens, presented very nearly the same 

 range, in respect to coat color, as did the F r The two 

 dark extremes were of an almost identical shade, as were 

 likewise the pale extremes. 22 No argument for " segre- 

 gation" could be based upon this series which would not 

 apply with equal force to the F 1 series. 



Twenty F 3 mice, resulting from random matings of the 

 F„ animals, presented a range of variation which was 

 actually not as great as that observed in the F 2 genera- 

 tion. The smaller number may perhaps be responsible 

 for this difference. 



It must be added that both of the parent races present 

 a rather wide range of variability as regards coat color, 

 and that series of the two overlap rather broadly in this 

 respect. This circumstance complicates our interpreta- 

 tions much more in the present case than in that of the 

 crosses between rubidus and sonoriensis, which will be 

 considered later. 



The Berkeley and Victorville races have been found to 

 differ in only two of the characters which were subjected 

 to careful measurement. The former race has a broader 

 dorsal tail-stripe and slightly shorter ears. The second 

 of these differences is a trifling one, however, and is not 

 always evident when small series are compared. iMore- 

 over, the absolute length of the ear is largely dependent 

 upon the size of the body. 



Thus the width of the tail-stripe is the only accurately 

 measured subspecific character which is available in con- 

 -idrnnu' hybrids between these two races. Table VII 

 gives the mean value and the variability of this character 

 in the two parent races 23 and in the F x and F 2 gener- 

 ations of hybrids. 



Despite the small numbers of animals here concerned, 

 two facts are of some interest. (1) The mean width of 

 the tail-stripe in both generations of hybrids is very 



