46 PROCEEDINGS : BOSTON SOCIETY NATURAL HISTORY. 
tion holds, for the coefficients of the head are obviously smaller than 
those of the face. 
Not only may we say that the face is more conspicuous than the 
head, but we may go farther and say that the size and shape of the 
nose affect the appearance more than any other feature of the face. 
These three propositions should then be true: — 
(1) The nose should be more variable than the head. 
(2) The face without the nose should be more variable than the 
head. 
(3) The nose should be more variable than the rest of the face. 
In the first column of Table 1 is given, for both sexes, the mean 
of the coefficients of variability of the four head dimensions. Column 
4 of the same table gives the mean of the coefficients of the four 
dimensions of the face, and column 5 the mean of the coefficients of 
the two nose dimensions. From these figures it appears, that (1) 
is true for both sexes; that (2) is also true for both sexes ; while 
(3), though true for the females, is not true for males. Under col¬ 
umns 3 and 6 are given the scores for these two tests. Beneath the 
sign -f- are given the cases of correlation between conspicuousness 
and variability; under the sign — the cases in which con-elation is 
wanting. The two tests give a score of 7 to I in favor of correla¬ 
tion. 
The data of Table A may be made to furnish yet another test of 
this relation. A moment’s consideration will show that vve nearly 
always visualize other persons in full face view rather than in pro¬ 
file, and that we think of other men and races as they look when 
seen face to face. It follows, then, that transverse diameters of the 
head and face, which determine the full face aspect, are, in general, 
more conspicuous than the dorso-ventral dimensions, which are seen 
more clearly in profile ; and that vertical dimensions, which affect 
both aspects, are more conspicuous than either. Here, again, I do 
not imply that every transverse dimension affects the appearance 
more than any dorso-ventral dimension, but only that this statement 
is true in general. We shall expect to find, then, (1) that the mean 
coefficient of the three dorso-ventral dimensions is less than the 
mean of the four transverse dimensions, and (2) less than the mean 
of the four vertical dimensions, and (3) that the mean of the trans¬ 
verse dimensions is less than the mean of the vertical dimensions. 
Columns 7, 8, 9 give these means, and show that all three state¬ 
ments are true for women, but only the third is true for men. The 
