398 FRANCIS B. SUMNER 



due to captivity be really a factor in accounting for the relations 

 to be discussed presently, its mode of operation has been to 

 increase rather than to decrease the appearance of segregation. 



It should be said, too, that the F2 generation, in the Carlotta- 

 Calistoga cross, was visibly more normal than in the Eureka- 

 Victorville one. Indeed, in the former, save for the fact that 

 most of the animals were appreciably smaller than wild ones, 

 there was rarely any indication of abnormality in either hybrid 

 generation. It is in :his set, let us note, that we have the least 

 indication of segregation. 



The possible bearing of this disturbing element upon my 

 hybridization data will be discussed more fully in a later paper, 

 in which I shall present the results of a more satisfactory series 

 of sonoriensis-rubidus hybrids. For the present, let me repeat 

 that I believe the possibility of such an influence merely 

 serves to weaken the slight evidence which I have found in favor 

 of gametic segregation in these racial crosses. 



Let us then consider the comparative variability of the two 

 hybrid generations. Figures 6 and 7 show the actual frequency 

 distributions of the two crosses here considered for two charac- 

 ters which admit of very precise measurement and which are 

 largely independent of the size of the individual. While exact 

 quantitative comparisons between such figures are out of ques- 

 tion, they certainly afford little evidence of an increase of varia- 

 bility in the second hybrid generation, as compared with the 

 first hybrid generation or with either parent race. 



A more accurate comparison of variability among these differ- 

 ent lots is afforded by table 4, which presents the standard devia- 

 tions for the parent races and the two generations of hybrids. 

 So far as tail-to-body ratio, foot lengths, and ear lengths are 

 concerned, these standard deviations have been subjected to a 

 'correction,' as were the averages in table 1. This is for the 

 reason that these three characters, particularly the two last, are 

 correlated with the general size of the animals. Part of their 

 gross variability, therefore, may be supposed to have resulted 

 from the variability of their respective series as regards body 

 length. I have accordingly computed their net variability in 



