Karl Pearson 
487 
There was thus closer sexual accord from the anatomical method. But when 
the same anatomical sexing was applied to the character of the head of the femur 
in the vertical plane, I found for right bones: 
$ 5-05 ^ 6-37 A = -1-32, 
and for left bones : 
% 4-91 (/• 6-10 A = -1-19, 
differences far greater than occur in the mathematical sexing from the bicondylar 
widths. Accordingly no great stress can be laid on inequalities in the coefficients 
of variation deduced from either process of sexing. 
It would appear to me that we have reached on the whole a reasonable 
biometric method of sexing. To what extent it can replace the sexing by 
anatomical appreciation must be left to the future. But it is clear that when 
anatomists themselves prefer to that appreciation an appeal to a single character, 
e.g. to the measurement of the femoral head, and only settle by anatomical appre- 
ciation the sex of femora with diameters between 45 and 47 mm., then they do 
not show much confidence in their own method of sexing. An interesting experi- 
ment could be made if some 400 to 500 sexed bones were available, and then, 
without knowledge of the real sex, two or three anatomists and a statistician were 
to be asked independently to determine the mean and variability of two or three 
characters of the bones of each sex in this material. 
I have cordially to acknowledge the help of my colleague Mr E. Soper in the 
determination of equations (xvii) — (xix) and in their solution (C) in the numerical 
case for which I had reached the solution (A) ; also the labour of my colleague 
Miss H. Gertrude Jones in the preparation of the diagram which contrasts 
graphically the mathematical and anatomical solutions of the problem. 
