8 
A Study in Crwiinal A7ithro2)077ietry 
For if he significantly exceeds or falls short of, on the average, the ivhole population, 
he will a fortiori be differentiated from the normal-minded population*. We have 
the following results : 
Difference of Means: 
(Weak-minded — whole population) . 
/ Temperature 
Pulse 
Respiration 
•174 ± 
3-205 ± 
-694 ± 
Height 
.Weight 
- -493 ± 
025. 
■628. 
•184. 
•202. 
- 7^750± 1^305. 
We conclude from these results that while the weak-minded criminal probably 
but not very certainly has a somewhat less staturef , he has quite markedly less 
weight, and this is combined with a higher temperature, a quicker pulse and 
a quicker respiration. 
We now come to precisely the same point as we reached in the problem of 
differential temperatures. We ask: How far are these differences explicable on 
the basis of the recognised difference in age between the weak-minded and the 
total population? From Table C (p. 11) we extract the following correlations 
and we give the corresponding regressions: 
Variates 
Correlations 
Regression of Second Variate on Age 
Per Year 
Per 5-36 Years 
Comparable 
Difference 
Age and Temperature 
Age and Pulse 
Age and Respiration 
Age and Height 
Age and Weight 
- -150 zt -022 
+ -121 ± -022 
+ -077 ± -022 
+ -023 ± -030 
+ -136 d= -030 
- -005,026 
+ -092,400 
+ -017,887 
+ -004,177 
+ -159,532 
- -0269 
+ -4951 
+ -0958 
+ -0224 
+ -8548 
•174 
3-205 
•694 
- ^493 
- 7-750 
* Let the means, total frequencies and standard deviations of the general, the normal-minded and 
the weak-minded population be tj, n and jT;, fg, /„, /„,, a-g, cr^^, and (r„, and let /S be the multiple (say, 
about 3) wliich a quantity must be of its probable error to be considered significant. Then in the 
usual way Ti —7r is significant if 
V ,','/ J II' \ Jo /I \ JuKJrj Jw) 
Jj-w is = or>j3(-67449) 
Now with /3=3 and our numbers, roughly /„, = 190, /y = 930, we have 
= 1-0005, 
(■67449/3)3/,, 
and tliis expression may be therefore talcen as unity. Hence if wo compare 
Tj-w- with -67449 ./^+^, 
and it is significant, it will certainly be significant if compared with 
-67449 
V fo 
fw \ fo 
which is a smaller quantity. See Biometrika, Vol. v. pp. 181-183. 
■\ Worked out from the fuller formula the result is — -493±-143, or, the difference is greater than 
three times its probable error and accordingly may be considered probably significant. 
