An Experimental Study of Somatic Modifications etc. 343 
dividnals, but by a comparison between averages computed for each 
group when the mice have been divided into groups according to 
size, and when these groups have been still further subdivided accor- 
ding to sex. By calculations of probability it has been shown that 
the chances for the purely »accidental« occurrence of all these dif- 
ferences are very slight. 
4) These differences among the offspring were manifested with 
fullest certainty in an earlier series of measurements, made at the 
age of six weeks. In a later series, made at the mean age of 
3 1 2 months, the same relations were found to exist, though to a less 
striking degree. 
5) The differences were exhibited with a closer approach to una- 
nimity by the females than by the males. It does not seem justi- 
fiable, however, to lay much stress upon this fact without further data. 
I will freely grant that this reappearance of the parental dif- 
ferences in the two sets of offspring is open to a number of interpre- 
tations. Some of these have little to warrant them and may be dis- 
posed of briefly: 
A. The differences in the offspring may have been due to » co- 
incidence* or » accident*. The odds against such an occurrence have 
been shown to be high. Indeed the cumulative improbability that 
all these differences have been accidental is enormous. 
B. They may have resulted unconsciously from a slight though 
constant biasing of the caliper measurements in favour of that result 
which was calculated to give the greatest personal satisfaction. This 
possibility has been excluded in the case of the second series of 
measurements. 
C. Granting their genuineness, these differences may have been 
due to the immediate effect of temperature as such upon the germ- 
cells. Since, however, we are dealing with a warm-blooded animal, 
it would be necessary to assume either that such an effect was im- 
pressed upon the germ-cells during the first few days after birth, 
before the animal had become homoiothermic, or that slight and 
hitherto unmeasured differences in the internal temperature of the adults 
were sufficient to affect the germ-cells. In either case, the produc- 
tion of parallel modifications in parent and offspring would have to 
be accounted for. 
D. These differences may be related to the circumstance that the 
difference in temperature conditions to which the parents were sub- 
