380 FRANCIS B. SUMNER 



report in the present paper upon unpublished data derived 

 from two of these crosses. The series included here comprise, in 

 one case 97 Fi and 87 F2 animals, in the other case, 154 Fi and 

 84 F2 animals. Two further series representing the widest of 

 these crosses still remain to be studied, but the F2 generations 

 are not yet old enough to kill and measure. 



The mice here considered are much freer from abnormalities 

 due to captivity than were the hybrids upon which T have re- 

 ported in previous papers. I feel, moreover, that I am more 

 nearly in a position to make proper allowance for such abnor- 

 malities, and to know when they do and when they do not 

 affect the validity of the results. Reference will be made below 

 to this aspect of the case. 



So far as the hybrid series are concerned, I shall restrict my- 

 self, in this preliminaiy discussion, to a comparison of the 

 variability of the Fi and F2 generations, in respect to certain 

 characters. In table 4 the standard deviations for five of 

 these characters have been given. These aie tail length (rela- 

 tive), foot length, ear length, width of tail stripe, and depth of 

 foot pigmentation. Since two different crosses are under con- 

 sideration, and the two sexes have been treated separately, 

 there are twenty pairs of figures to be compared, in our endeavor 

 to ascertain the relative variability of the two hybrid generations. 

 Owing to the fact that the absolute measurements for foot 

 length and ear length are closely correlated with those for body 

 length, and since the variability of the various series differs con- 

 siderably in respect to body length, T have computed the net 

 variability for foot and ear length (see below). 



To sum up the outcome of these computations, out of twenty 

 pairs of comparable figures, that for the F2 generation is greater 

 in 8 cases and less in 4 cases, while the two are equal in 8 others. 

 (I have considered two figures as equal when the difference be- 

 tween them is less than the probable error of that difference.) 



It is not, however, certain that the parent races in either of 

 these crosses differ significantly in respect to ear length, while 

 in one of the crosses (Carlotta-Calistoga) it is also questionable 

 whether the wild stocks differ significantly in respect to foot 



