An Experimental Study of Somatic Modifications etc. 
337 
It will be seen at a glance that the mean length of tail, foot 
and ear is, as was previously the case, greater for the warm-room 
descendants than for the cold-room descendants. But the differences 
between these gross averages signify even less here than in the case 
of the earlier figures, since the two contrasted lots now differ quite 
appreciably in their mean size. The warm-room descendants are 
heavier on the average by more than a gram, while their average 
body length is nearly one millimeter greater. For this reason it is 
even more important that our animals should be divided into groups 
according to size. I have, accordingly, grouped the mice in two dif- 
ferent ways: 1) according to weight, as was done previously, and 2) 
according to body length. The latter method of grouping seems a 
much fairer one than the first, for it seems likely that the length of 
the appendages is correlated primarily with body length and only 
incidentally with weight. The latter, of course, depends in large 
degree upon the nutritive condition of the animal, amount of adipose 
tissue, etc. The single weight-groups, it may be added, contained 
individuals which differed from one another by as much as 4 or 5 mm 
of body length. 
To consider the second of these methods first, the animals were 
divided into groups, within each of which the individuals differed by 
less than one millimeter in length. The appended table (B) permits 
of a comparison of warm-room and cold-room descendants within 
each of these size-groups, for each sex separately, and for the two 
sexes combined. 
Considering the averages for the two sexes combined, we find 
that in the 15 size-groups which admit of such a comparison, the 
»warm« figures for tail, foot and ear length are larger than the »cold« 
figures in 12, 11 and 10 cases respectively. The chances for the 
accidental occurrence of such majorities are approximately 1 in 57, 
1 in 17, and 1 in 7, respectively. But we have the cumulative testi- 
mony of all three of these characters, pointing in the same direction. 
The chances that the »warm« figure shall thus be greater in 33 out 
of the 45 cases are only 1 in 814 1). 
If we deal with the sexes separately, we find 21 groups within 
which comparison is possible between the warm-room and cold-room 
descendants, or 63 pairs of averages, when the three characters are 
considered. In 40 of these the »warm« figure is larger than the 
ij Leaving out of account correlation between these characters (see above). 
Archiv f. Entwicklungsmechanik. XXX. 2. 22 
