44 
Data for the Problem of Evolution in Man 
rate of decadence largely varying with the individual. At any rate, whatever be 
its source, it is very important in its bearing on all attempts to estimate the 
intensity of selection, even in the case of other animals than man. We must be 
careful that there is not an age or growth factor at work, counterbalancing a 
reduction in variability due to selection. The female groups are more irregular in 
their results than the male; in particular the last group for women over 60 
contains only 282 entries. Still the general tendency is the same — increased 
variability with old age. 
Judging by standard deviations man would appear more variable than woman 
in stature. If we consider all the groups of equal weight and take the mean we 
have 6G"'885 for the mean stature of men, and 62"'242 for women, the standard 
deviations being respectively 2''"5928 and 2"'4244. Judged by percentage varia- 
tion we have for the coefficient of variation in man 3'88, and in woman 3"89, 
results sensibly equal. 
The following Table gives relative experience, the agreement in variability is 
wonderfully close. 
CODNTEY 
Great Bkitain 
New South Wales 
Class 
Upper Class * 
Middle Class t 
Criminal Classes 
Sex 
Male 
Female 
Male 
Female 
Male 
Female 
Mean 
Standard Deviation ... 
Coefficient of Variation 
68"-93 
2"-51 
3-6 
63"-82 
2"-42 
3-8 
67"-90 
2"-5.5 
3-7 
63"-30 
2"-40 
3-8 
66"-88 
2 "-59 
3-9 
62"-24 
2" -42 
3-9 
* Cambridge Undergraduates: see Pearson, The Chances of Death, Vol. i. p. 313. 
t Visitors to International Exhibition in 1884 measured in Mr Galton's South Kensington 
Laboratory: see Galton, Natural Inheritance, p. 200, and Pearson, loc. cit. p. 311. 
The British measurements practically include nobody over 51 years of age. 
