410 Variation and Correlation of the Human Skull 
height, and of the two cephalic indices, were published in the memoir on Spurious 
Congelation of" 1896*. Soon after this Dr Alice Lee took up the subject and 
wrote an academic dissertation on the correlation of the skull in 1899, which 
appeared in the Phil. Trans, as A First Study of the Correlation of the Human 
Skullf. Her memoir deals chiefly with the correlation between the capacity of 
the skull and its length, breadth, and height for a variety of races, but in- 
cidentally a good many other correlations are considered. Meanwhile Dr Franz 
Boas had published in 1899 an interesting paper especially drawing attention to 
the correlation of the cephalic index with the capacity |. Roughly we may say 
that the result of these investigations is to show : 
(i) that the correlation of the parts of the skull varies remarkably from one 
local race to a second ; 
(ii) that the correlation of the measurements most commonly made is re- 
markably small, as compared with that of the principal dimensions of the human 
body, e.g. the long bones of the skeleton or the bones of the hand. 
These conclusions, definite as far as they go, rather strengthened than weakened 
the need of the investigation we had been for some years engaged upon. Our 
object was to determine in this " Second Study of the Correlation of the Human 
Skull " whether, confining our attention to a single race and dealing with a greater 
range of characters, we should find any more highly correlated than those referred 
to above. We desired further to place before the reader the full statistical treat- 
ment of a large series of skulls, adopting the modern methods of reduction, with 
a view, if possible, of making these methods more generally known to craniological 
investigators. The publication of raw material is always of value, but we are con- 
vinced that until the statistical constants for variation and correlation have been 
calculated for such material, but little can be safely asserted as to racial relation- 
ship from purely craniometric investigations. 
(2) Material. 
The great difficulty of the statistician in approaching craniological problems is 
the absence of sufficient material. This is an unavoidable difficulty which must be 
faced. It is very hard to obtain a homogeneous group of skulls, even 50 in 
number, and these again must be distributed between the two sexes. The probable 
errors therefore of constants determined from such series are proportionately large. 
For statistical purposes much of the craniometric data published by anatomists 
is hopelessly inadequate, and if we are to trust reliable craniometric judgments 
rather than anatomical appreciations!, we must impress upon craniologists the 
* R. S. Proc. Vol. 60, p. 495. 
t Phil. Trans. Vol. 196, A, pp. 225—264. 
t The American Naturalist, N. S. Vol. i. p. 448. 
§ We do not contest the value of the anatomical appreciation in the hands of the master, but we do 
contest the cloaking of such appreciation by an apparent array of craniometric data, which are 
statistically inadequate. 
