J. Blakeman 
131 
We see at once that a general hospital population is about equally variable 
with other English groups, whether of lower grade, like the criminal, or of higher 
grade, like the middle class, chiefly drawn upon for Cambridge graduates or B.A. 
members. School children are more variable — probably a phenomenon of growth. 
But again we see that our skull series is more variable than the head series. For 
the circumference our comparable data are nil ; we can only compare with the 
English skull. This provides : 
TABLE IX. 
Coefficients of Variation for Englisli Horizontal Circumference. 
Group 
6 
? 
General Hospital Population ... 
2-85 
3-16 
English Skull* 
2-87 
2-92 
In the matter of the circumference we should therefore conclude that there is 
equal variability for head and skull measurements. 
Stature provides us with more material. We find : 
TABLE X. 
Variation in Stature in Inches. 
Group 
Standard Deviation 
Coefficient of Variation 
s 
? 
? 
General Hospital Population ... 
3-05 
2-79 
4-55 
4-42 
Galton's South Kensington Records + 
2-55 
2-40 
3-95 
3-79 
English Criminals 1 ... 
2-59 
3-88 
Cambridge Students t 
2-51 
2-42 
3-64 
3-79 
Pearson's Family Measurements, Offspring § 
2-71 
2-61 
3-95 
4-09 
„ „ „ Parents § 
2-70 
2-39 
3-99 
3-83 
It will be seen that both absolutely and relatively the general hospital 
population is, in the case of stature, more variable than any of the other English 
groups. We believe this iticreased variability in stature to be due to the measure- 
ments being taken on the corpse, not on the living subject. As a rule we may say 
that the coefficient of variation of the living stature of the adults of any race is 
under 4. Pearl gives twelve cases of this coefficient for corpse measurements in 
his Table I, p. 23, and in ten of these cases the coefficient of variation is over 4. 
* Macdonell ; Biometrika, Vol. iii. p. 222. 
t Pearson : The Chances of Death, Vol. i. pp. 313 and 311. 
X Macdonell : Biometrika, Vol. i. p. 181. 
§ Pearson and Lee : Biometrika, Vol. ii. p. 370. 
17—2 
