218 Variation and Correlation in the Earthworm 
The constants deduced from this distribution have the following values : 
Mean = 1 9-171 + 094 cms. 
Standard Deviation = 3-077 + -067 cms. 
Coefficient of Variation = 16-049 ± '356 7„. 
From these values the following points are to be noted : 
(a) The range of variation in respect to number of somites in the body is very 
great. The explanation of this fact is simple. The sample of worms under con- 
sideration is not homogeneous with respect to age. Up to a certain age at least, 
Lumhricus adds somites at the posterior end of the body with growth. It is clear 
from either Table I. or Figure 1, that the great extent of the total range in the 
variation in this character arises from the presence of a comparatively small 
number of individuals with few somites which unduly extend the lower end of the 
range. It seems reasonably certain from what we know of the method of growth 
in the earthworm that if this sample of worms could have been allowed to go on 
growing the range of variation in total number of somites would have steadily 
decreased with lapse of time. 
The fact that there is no apparent tendency towards extreme extension of the 
range at the upper end indicates that the bulk of the worms included in the 
sample have either completed the proce.ss of adding somites or at least that the 
process is going on very slowly if at all. So then the sample serves to define the 
existing typical condition of the worms in the region with respect to this character 
with a reasonable degree of accuracy. It is to be regretted that an entirely homo- 
geneous sample with respect to age cannot be obtained, but in the nature of the 
case this is practically impossible. It would involve following each individual 
worm from the time that its development began. 
(b) The earthworm is apparently more variable in respect to length of body 
than in respect to number of somites. This would indicate that the observed 
variation in length is not due primarily to variation in number of somites, but 
rather to variations in the lengths of the individual somites in different worms. Of 
course, it immediately occurs to one that possibly some of the greater variation in 
length is due to varying degrees of contraction in individual worms. As has been 
stated above, however, this source of error has been largely eliminated by taking 
large length classes. The range of variation in length due to degree of contraction, 
when the worms are fixed in the way described, is very probably considerably less 
than one length class unit of 2-5 cm. The conclusion, that the earthworm is more 
variable in length than in number of somites, may seem at first sight paradoxical, 
but it is really no more so than to conclude, what is very obviously the fact, that 
man is more variable in respect to sitting height than in respect to number of 
vertebrae. It simply indicates that in the earthworm we have two kinds of growth 
occurring together : one the addition of somites, the other increase in size of the 
individual somites. That these two kinds of growth are acting together will be 
clear when the correlation surface between total number of somites and length of 
body is examined. 
