Effects of External Conditions | 139 
number in heavy type is the larger number.?° This decrease in 
bodily differences originally brought about by differences of 
temperature has not been dueto a withdrawalof thelatter. Indeed, 
during the interval between the “6 weeks’’ measurements and the 
‘“‘24 months” measurements the temperature differences in the 
two rooms have increased rather than diminished. (See Fig. 1.) 
Later, it is true, the temperature differences gradually diminished, 
and commencing with April 2 they were abolished altogether. At 
the latter date, however, the mice averaged nearly five months 
old, and their growth was probably not far from complete. 
While the tendency toward a reduction of the original differ- 
ences between the warm and the cold room groups is thus pretty 
clear, the evidence for a reduction of variability within each group 
is not so certain. An inspection of the tables shows us twelve 
cases in which we can compare a later standard deviation with an 
earlier one for the same character. In six of these cases the later 
standard deviation is smaller, i. e., the decrease in variability has 
been absolute as well as relative. In two cases there has been 
a relative decrease, though not an absolute one; while in two others, 
the relative variability has remained practically unchanged. In 
only two of the twelve cases has there been any appreciable in- 
crease in the relative variability. In view of the lack of uniformity 
in these results, however, and the commonly high probable errors, 
too much significance must not be attached to them. It is worth 
pointing out, however, that the variability for tai] length has de- 
creased absolutely as well as relatively in five out of six cases. 
Series of 1905-1909 
At the date of writing, the experiments of the third winter have 
not been carried very far. A first and second series of measure- 
ments upon the living mice have, however, been made, and the 
results seem well worth comparing with those of the preceding 
39 Tt is likewise true that in all of these seven cases there has been a greater absolute gain as well 
as a greater percentage increase. Rate of growth seems more fairly expressed, however, in terms Of 
proportionate increase. 
