J. F. TOCHKR 
341 
be about the correct value. No British adult data are at present available to 
determine the constants involved. The change, however, in passing from juveniles 
(under 19) to adults can be approximately measured from the Aberdeenshire 
data. With the figures given in the following table, r = '24, when we use 
Pearson's fourfold table method for characters not quantitatively measurable*. 
TABLE XXI. Correlation of Age and Hair Colour. 
lied 
Fair 
Medium 
Dark 
Totals 
Adults 
8 
71 
133 
189 
401 
Children 
28-1 
100-2 
188-5 
84-2 
401 
Totals ... 
36-1 
171-2 
321-5 
273-2 
802 
III. Distribution of Colour among the Sane and the Insane. Since the 
Aberdeen data represent a local group, the colour observations on adults theie 
cannot be contrasted with the colour data of the "general insane" population, 
as local groups may or may not be good samples of the general population. It 
has been shown for all characters that they are more likely not to be good 
samples. The rate of change of hair and eye colour with age, however, is not 
so likely to vary in passing from one district to another. On the assumption 
that the rate of change is fairly uniform thruughout Scotland, an approximate 
estimate can be made as to the probable distribution of hair and eje colours 
among the adult sane population from the Aberdeenshire data and the results of 
the observations from the Pigmentation Survey of School Children in Scotland 
about to be published. Let p^, p.,, ... pn = percentage of either hair or eye colour 
among school children in any district ; q^, q2,...qa= similar percentages among 
the adults in the same district; p\, p'o, ... p'„= similar percentages among the 
children in the entire school population ; i2 = (1 + e/(100 — e)) ; e = a constant whose 
value depends on the nature of the distribution and n = number of categories then 
p\q^ R _^ p'.q.,R _^ p'-.q., R _^ p'nqnR _ ^ 
Pi Pi Ih "' Pn 
and gives the corresponding probable percentages of either hair or eye colour in 
the general adult sane population of the country. Applying this equation to the 
Scottish normal data, we obtain the following values, the corresponding values for 
the "general in.sane " population being given for comparison. 
To judge from this result — a tentative one — there is an excess of light-eyed, 
brown and dark-haired persons in Scottish asylums and a corresponding defect in 
the other categories. The colour distributions of the " general insane " population 
cannot therefore be held to be fair samples of the general population of Scotland. 
* It is assumed that selection by hair-colour does not occur; the children are the distributions 
of 401 individuals on the base of the Pigmentation Survey, for Aberdeenshire ; the division is taken 
between ' fair' and ' medium.' 
Biometrika v 44 
