28 
Variation ami Correlation in Brain-Weight 
as between 1500 and 1600 will centre somewhat below 1550. Now, of course, if 
the distribution is exactly or approximately normal and the number of individuals 
is sufficiently large, the errors from this cause on one side of the general population 
mean will balance those on the other, and we shall still be able to get a very close 
value for this mean from the frequency distribution. Unfortunately, in Matiegka's 
series, however, the total numbers are not sufficiently large to overcome entirely 
this error. So we have the following differences : by direct calculation from the 
individual observations Matiegka finds for the mean brain-weight of the " young " 
(20 — 59) series in the male 1450-4 gr., in the female 1305-5 gr. Our corresponding 
values are 1460-150 gr. and 1313-910 gr. or there is an excess of 9 -7 gr. and 8-4 gr. 
respectively. For the " total " series (20 — 80) I have calculated from regression 
values which Matiegka gives (p. 7), based on the individual observations, general 
population means which gives values as follows : males 1441'4 gr., females 
1294-8 gr. The corresponding values from the tables are : males 1454-8 gr., 
females 1310-9 gr., or the excess of the table means is 13-4 gr. for the males and 
16-1 for the females. These differences are small and for practical purposes 
negligible. 
The general conclusion may be drawn that, apart from all differences in stature, 
and in tlie mean age of the samples studied, the Bavarian mean brain-weight is 
lower than the Hessian (roughly about 2 °/^), and the Bohemian mean brain-weight 
is higher than the Hessian (roughly about 3-5 °/^). 
These results seem to be of some importance as indicating a method whereby 
brain-weight may have some scientific validity as an anthropological character. 
This can hardly be said to be the case when nothing more is done than to tabulate 
means without probable errors, and no attempt is made to get rid of disturbing 
stature and age effects. Interracial differences in brain-weight may mean 
something or nothing. With modern biometrical methods it is possible to measure 
exactly these differences, and when such methods are used it is the belief of the 
writer that brain-weight can become a really significant anthropological character. 
In the cases discussed here we see ethnic affinities and differences clearly reflected 
in the brain-weight. Of the four racial groups studied the two most closely 
related in origin and general anthropological characters, the Swedes and the 
Hessians, have mean brain-weights sensibly identical. The Bavarians and 
Bohemians, though close neighbours geographically, arise from totally distinct 
ethnic stocks (" Alpine " and Slav respectively) and associated with this we find 
a wide difference in the weight of the brain. 
On the question of the homogeneity of brain-weight statistics, some evidence 
is furnished in the third and fourth columns of Table IV. If the inclusion of aged 
persons (50 years and over) introduced any considerable element of heterogeneity 
so far as brain-weight constants are concerned it would be expected that the 
interracial differences in the means would not be equal in the " total " and in the 
" young " series. As a matter of fact in the material here discussed they are 
sensibly equal, or perhaps better, they are not significantly different. The whole 
