Raymond Pearl 
31 
All of the four series used in this paper are in good agreement, both among 
themselves and with the skull capacity values. They are distinctly lower than the 
figures from the admittedly heterogeneous, Reid, Peacock, Sims and Clendinning 
English series, worked over by Pearson. Taking all the evidence together we are 
driven to the conclusion that the series are fairly homogeneous. If any one of our 
series is markedly heterogeneous all the others must be equally so, which would be 
a very improbable result. That the series cannot be very heterogeneous in their 
make-up is also shown by the values of the coefficients of variation for skull 
capacity from two admittedly homogeneous series, the Whitechapel skulls studied 
by Macdonell, and Ranke's Bavarian series. 
A comparison of the variability in brain-weight with that shown by other 
organs and characters of the human body may next be undertaken. I have 
arranged in the following table (VII) the coefficients of variation for a variety of 
characters which have been studied by biometrical workers. The arrangement is 
in general that of descending order of values in the male series. The attempt has 
been made in the table to include representatives of all the different classes of 
organs and characters for which we have biometric data available. 
The most noticeable and remarkable fact brought out by the foregoing table 
is that with the exception of capacity, all skull characters are roughly only about 
half as variable as brain-weight. Some such a relation as this might have been 
predicted, on the general ground that brain-weight measurements and statistics 
are on many accounts rather " loose," and would indicate a higher variability than 
would exact measurements on skulls, even though it had no real existence. Such 
reasoning, however, takes no account of the agreement, which is really remarkably 
close when we recall the numerous sources of error in brain-weight returns, between 
skull capacity and brain-weight in their variability. All will admit that capacity 
is the most difficult skull character to measure accurately, yet no one would 
maintain that the difference in variability between cephalic index, for example, 
and skull capacity was entirely, or even in any considerable part, due to the 
element of error in the measurement of the latter. The agreement in variability 
between skull capacity and brain-weight is, of course, to be expected on theoretical 
grounds. That it should turn out in practice to be so close is a first-rate guarantee 
of the general trustworthiness of brain-weight statistics. 
As to the explanation of the great variability in brain-weiglit and skull capacity 
as compared with the other skull characters we may tentatively reason about the 
matter in the following way. If the list of organs and characters given in the 
foregoing table be examined a natural division into three groups almost im- 
mediately suggests itself First, we have at the bottom of the list the " bone " 
measurements, in general including all those characters which depend primarily 
for their values on the dimensions of various parts of the skeleton. These 
characters give values for the coefficient of variation up to from 5 to 7, certain of 
the mandibular variabilities exceeding this limit. Next comes the group giving 
values for the coefficients of from 7 to 10, with the limits fairly sharply marked off. 
