GENETICS: W. E. CASTLE 
263 
B females, the standard deviation of F2 in skull-length (Table 1) is less 
than that of uncrossed race B females and is not half a millimeter 
greater than that of the Fis. The difference can scarcely be regarded 
as significant. In skull-width the F2 females have a smaller standard 
deviation than the FiS, while in femur-length alone is F2 larger than Fi 
and either parent race. 
Among the male hybrids produced by the Cutleri X race B cross, the 
evidence for increased variability in F2 is rather better. In skull-length 
(Table 1) and femur-length (Table 3) the standard deviation of F2 
exceeds that of Fi or either uncrossed race. In skull- width the differ- 
ence is not significant. 
It should be noted in passing that the very low standard deviations 
of pure Cutleri animals do not necessarily indicate lack of variability, 
but are to be explained in part as due to the small numbers of animals 
available for measurement. For Pearson has shown that with numbers 
less than 25, the empirical standard deviation is as a rule too small. 
It will be observed in Tables 1-3 that the highest standard deviations 
among the pure Cutleri animals are found where the numbers studied 
are largest. As between the Fi and F2 animals studied, the numbers 
are not sufficiently different to make any allowance of this sort necessary. 
In the Arequipa X race B crosses, F2 has a higher standard deviation 
than Fi in five out of six cases, the Fi animals being more variable only 
in the skull-width measurements of males (Table 2). 
We may conclude that on the whole F2 shows consistently a higher 
variability than Fi in the croses studied, which is in agreement with 
the observations made in numerous other hybridization experiments 
with both animals and plants. But it does not follow that the differ- 
ence observed is due wholly to multiple genetic factors affecting size. 
For increased variability in F2 would occur if the physiological increase 
of size observed in Fi persisted to some extent in F2 but persisted un- 
equally (i.e., in different degrees) among the different F2 zygotes. Now 
there is some reason to think that the non-genetic or excess vigor of 
Fi does persist slightly into F2, for in 5 out of 6 cases in the Cutleri X 
race B cross (Tables 1-3), the F2 mean is greater than the intermediate 
point between the means of the uncrossed races. But it is clear that if 
this non-genetic vigor is found to a greater extent in some F2 zygotes than 
in others, it will increase the variability of the F2 zygotes as a group. 
As against the multiple factor hypothesis it may be urged further 
that an increased variability of F2 may be satisfactorily accounted for 
in still other ways without involving multiple factors, as for example 
by quantitative variation in a single factor affecting total growth eaergy 
of the zygote. 
