Variation and Correlatioti in Br ai7i- Weight 
of getting at this question with the data available, but none of them led to any 
result sufficiently well grounded to warrant taking space for its publication. 
Putting all the results so obtained together they seemed to indicate the following 
conclusion, which is of course subject to modification when further material is 
available : viz., that changes in stature affect all parts of the brain equally (i.e., in 
the proportion of their absolute masses) while changes in age have a relatively 
greater effect on the cerebral hemispheres than on the remainder of the brain 
(cerebellum, pons, and medulla). 
The final conclusions regarding variation and correlation in the weight of the 
cerebruin may be summed up as follows: the cerebrum is somewhat more variable 
in weight than is the entire encephalon, and this character is slightly less closely 
correlated with age and stature. The values of the variation and correlation 
constants are, however, as is to be expected, of the same general order of magni- 
tude as those for the total brain-weight. 
13. Concluding Remarks. 
In concluding this paper I wish to call attention to what seem to me to be 
some of the broader aspects of the work. There are at least two general results 
of the work which alone justify its being done, I think. The first and most 
important is that the paper contributes reduced material to the existing collection 
of biometric data on man. In the nature of the case such a collection grows 
slowly, but every increase in it means a definite, although it may be small, step in 
advance in our knowledge regarding the fundamental problems of anthropology. 
I regard as the second most important result the fact that it may now fairly be 
said that the essential trustworthiness of the most important of the existing 
collections of brain-weight statistics has been demonstrated, and that consequently 
reasoning by statistical methods on the problems involved is not of necessity 
lacking in validity. The agreement in the statistical constants from four series 
of data so divergent in their origin as those treated in this paper cannot reasonably 
be held to be fortuitous. It can mean, I think, but one thing, namely, that the 
same kind of general lawfulness underlies the variation and correlation of brain- 
weigiit and the variation and correlation in other characters of the organism 
With this conclusion presumably no student of brain-weight would disagree, but 
some of the most eminent students have disagreed with the converse proposition 
that it is possible to gain a knowledge of the nature of this lawfulness by statistical 
methods applied to large masses of material when brain-weight is the thing 
concerned. This sceptical attitude owes its origin, I think, to the nature of the 
statistical methods which have hitherto been applied to the problem. Brain- 
weight statisticians have erred in two directions in the handling of their material. 
On the one hand, following the much-to-be-condemned practice in other fields of 
anthropology, entirely heterogeneous material has been grouped together to attain 
large total numbers. The culminating example here is, perhaps, Topinard's series 
of the brain-weights of 13,000 Europeans. Secondly, certain brain-weight workers 
