400 



FRANCIS B. SUMNER 



Reference has also been made in the introduction to the proba- 

 bihty that the various characters which are associated together 

 in a single local race do not tend to cohere in hybridization, but 

 that they appear to vary nearly or quite as independently of one 

 another in hybrids as they do in pure races. The rather meagre 

 data thus far available for a test of this question are somewhat 



F2 HYBRIDS 

 y\m 90.9 



Fj HYBRIDS 

 Mean 32.5 



VICTORVILLE 

 Mean 81.3 



i^- H h 



n n I 1 



S4 



Parents of F, Broods 



Parents of Fi Broods 



J]£hA 



i~i n n 



Parents of F] Broods 



^''»%\\\''A%V^VAV'AVMVXAAh^V\ V^%\A''A\\\V'^%%%V\i\''^i 



Fig. 7 Variability in Eureka-Victorville cross (see legend for figure 6). The 

 actual parents of the Fi animals here belonged to a cage-born series, and are 

 therefore not included in the larger polygons for the parent races, which are 

 based upon wild material exclusively. 



contradictory. But they do not, I think, furnish any support 

 to the view that racial characters tend to segregate together, if 

 indeed any segregation, in a Mendelian sense, occurs here at all. 

 T will first consider the question of the correlation of relative 

 tail length and width of tail stripe, characters which have been 

 shown to lack significant correlation within the single races, 

 though they are plainly associated with one another when cer- 



