J. r. Tocher 
333 
Taking the series as a whole, we find the coefficients of variation in B^^ differ 
materially from B%. On running through the values for the various asylums (see 
Table XIV.) this material difference is seen to be due to the inmates at Lenzie, 
where the sexual difference is very significant, and in a lesser degree to the 
inmates at Banff and Govan where the males also show greater relative variability, 
and at Ayr, where the females show the excess. In the other cases, the differences 
are not significant. The relative difference for L in the general population is 
perhaps hardly significant, but here again Lenzie stands out with a prominent 
difference. While the general coefficients for H are nearly equal, there are 
significant differences in the local values at Aberdeen and Perth and less significant 
ones at Fife, Paisley, Dumfries, Govan, Montrose and Haddington. Lenzie is 
again the disturbing factor in BjL and in *S^. The coefficients for BjL and 8 
are approximately equal for most of the other asylums. 
Thus, considering the differences between the sets of coefficients for both sexes 
at the individual asylums, we reach the conclusion that the variability among 
the males and females is very much alike, with the exception of the character B 
and a few local cases in the other characters. The conclusion is confirmed and 
amplified by considering the whole population where the males again appear 
more variable in B, the difference being probably significant, but in the other 
characters the variability among the males, although greater than among the 
females, is only slightly greater, and cannot be said to be at all significant. 
(6) Pigmentation. 
I. Distribution of Hair and Eye Colour. As already stated, the colours of the 
hair and eyes of most of the inmates were noted at the same time that the 
measurements were taken. A complete record of the observations is given in 
the Supplement to this memoir side by side with the record of observations on 
measurable characters. 
In order to make a comparison between the pigmentation of the inmates at 
each asylum and the pigmentation of the "general insane" population, the values 
of ■^^ and logP* were calculated for each asylum. In other words the actual 
frequencies for each colour were contrasted with their most probable values — the 
theoretical numbers which would occur on an even distribution of the "general 
insane" population. The approximate values of log P are given in Table XV. 
and XVI. and show that with the exception of a few cases, the local pigmentation 
diverges in character from the general distribution considerably. It is of interest 
to note that the divergence in colour of any locality from the remaining population 
may be measured by determining the mean square contingency coefficient 
where ;i^'^ = the total square contingency!". Thus, to take a particular case, the 
distribution of hair colour in males at Aberdeen may be contrasted with the 
remaining population as in Table XVII. We find Ci = *1347. 
* Elderton : Biometrika, Vol. i. p. 155, x' °^ Goodness of Fit. 
t Pearson ; Drapers' Company Research Memoirs, Biometric Series, i, p. 16. 
Biometrika v 43 
