Karl Pearson and David Heron 
271 
from the contingency -5189*. It is thus easy to see why Mr Yule gets such 
an absurdly low value. If it be said that two out of the nine values of tetra- 
choric r t are lower than the Yulean, the reply is a simple one : They are not 
divisions which would be made with our present knowledge of hair and eye 
pigmentation. There is no such thing as a " black" eye, at best it is only a dark 
shade of brown, and the division between brown and black is largely a matter of 
personal equation. Again the relative amount of pigmentation in a true grey eye 
may be as small as in many so-called blue eyes. The natural physiological division 
is between the blue-grey and the brown-black groups, although, if Livi put his 
hazels and greys with some pigment partly into blues and partly into browns, a 
division between blue and grey might give as good results. Passing now to the 
hair-colour, we believe the distinction between brown and black to be again a 
matter of personal equation, the shades of brown range up to black. Again red 
is a tint which may contain less pigment granules than the blond f, but in some 
varieties it has more granules than many browns. Hence the difficulty of such 
a division as that betwixt brown and red. The real division would come between 
blond and brown with reds omitted or better still microscopically tested and 
placed in their appropriate division. Anyhow the divisions corresponding to the 
four left-hand top quadrants of the last table seem to us the most satisfactory, and 
they all agree in giving the correlation of hair and eye colours = circa "52, a result 
in excellent accord with the mean square contingency. 
Pearson's data for British school children give by corrected contingency "52, 
and the following table shows that the values for all nations must be in excess of 
that determined by Mr Yule's method : 
Correlation of Hair and Eye Colours by corrected 
Contingency. 
Italian Recruits -519 
British School Children ... ... ... ... -524 
Baden Recruits ... ... -484 
German Jewish Children -444 
Prussian School Children -401 
Swedish Recruits -414 
Italians by Yule's Method of Pseudo-Ranks ... -368 
* Assuming normal correlation, the correction for ranks is r = 2sin(30°x •3680) = -39o3. For bail- 
groups the class-index correlation is - 8686 and for eyes -9010, and variate correlation 
= -3953/(-8686x '9010) = -5051. 
Of course these are only rough approximations, for Mr Yule's theory of unit subranges precludes our 
using the corrections actually given. But any corrections will be roughly of this order. 
t Some reds lack all granular pigmentation ; there are blonds which also do this, but tbey are very 
few in number. Both reds and blonds without melanin pigment granules are truly albinotic hairs, 
whether accompanied or not by albinotic eyes. 
