238 
DR. A. LEE AND PROFESSOR K. PEARSON ON 
We have now all the data necessary for reconstructing the skull capacity, and it 
remains for us to consider how we can apply these to our three fundamental 
problems. 
(8.) First Fundamental Problem. The Peconstruction of the Individual from the 
Icnown Fornmlre for his own Race. 
In order to illustrate the degree of exactness with which we can reconstruct the 
individual from their own racial data, a perfectly random selection of twenty skulls 
was taken out of those of each sex for the Ainos and Germans, and their capacities 
reconstructed from each of the nine regression formulm given in Tables Y. to YTII. 
respectively. The results are tabulated in the following Tables XI. to XIY., and will 
enable the reader to appreciate the degree of exactness with which it would he 
possible to reconstruct the capacity of an individual skull from any one or more 
measurements made u 2 :)on it. 
These results show us at once that the last five formulm are, when available, hv 
far the best to use. (3) and (4), namely, reconstruction from the auricular height 
and the cephalic index, give occasionally very poor results. The latter formula, 
while of much interest from the racial stand]Doint,* need never be used for reconstruc¬ 
tion, for the knov.dedge of the cephalic index means a knowledge of L and B, and 
accordingly we can always use (5) if not (8). 
An examination of the actual mean error made when we use all nine formulae and 
take the mean of their results shows that, as a rule, we shall obtain less error by 
selecting one good formula like (8) or (9) and using that only than if we attempt to 
use them all. In round numbers we see that the mean error made in reconstruction 
by these formulae is aljout 5 ^Der cent., but if we use (8) or (9) tbe mean error will lie 
between 3 and 4 j^er cent. Tlie maximum error reached by a good formula like (8) 
or (9) is ujawards of 10 per cent., but its occurrence is infrequent. On the whole, I 
consider this reconstruction of the individual from data for his own race fairly 
satisfactory. It is jDractically nearly as good as we get in the reconstruction of 
stature from the long bones, f I would also remind the reader that the theory of 
correlation shows tliat we cannot hope to get better results. We have solved the 
2 )roblem as closely as it can be solved, so long as tbe skull is a variable organ. From 
a knowledge of the degrees of variation and correlation of an extended number of 
parts of tlie skidl (unj^ublislied data), I feel fairly confident that no external measure¬ 
ments can be taken iq:)on it which will give substantially better results than those 
ah’eady considered.! When we bear in mind that tv'o difterent ol^servers, using even 
* Tf wc wish to identify criminals, wc select characters to he measured and indexed which exhihit the 
least correlation. In the same way to ditferentiate and specify races, it is best to select a group of 
ehai'acters having the least correlation ; one such is certainly the cephalic index. 
t See Pearson ; “On the Reconstruction of the Stature of Prehistoric Races” (‘Phil. Trans.,’ A, vol. 
192, pp. 188-189). 
I An appendix is devoted to a consideration of the horizontal and vertical girths. 
