134 Influence of "Broad Categories " on Correlation 
Very Robust, Robust, Normally Healthy, Rather Delicate, Delicate and Very 
Delicate. Applied to 2037 schoolgirls (Biometrika, Vol. III. p. 166) it gave: 
We have : — = 
V. R. 
E. 
N. H. 
R. D. + D. 
V. D. 
Total 
109 
- 2-0342 
578-5 
- -9033 
803-5 
+ -0936 
478-5 
+ 1-0837 
R7-5 
+ 2-2293 
2037 
leading to 
If we divide into four classes, thus : 
' 1 3f cT 2 ] = ' 897 > 1616 > and r *c, = ' 9472 - 
V. R. 
R. 
N.H. 
R. D. + D. + V. D. 
Total 
109 
578-5 
803-5 
546 
2037 
We have : - 3 = 
<r x 
-2-0342 
- -9033 
+ -0936 
+ 1 -2253 
8 =L } = -858,9714, and r xC =-9268. 
leading to 8 [ ^1 — , 
And lastly if we divide into three classes, thus 
V. R. + R. 
N. H. 
R. D. + D. + V. D. 
Total 
687-5 
803-5 
546 
2037 
We have : — = 
- 1 -0826 
+ -0936 
+ 1-2253 
leading to 
S f^-J = •801,3866, and r xC =-8952. 
Now if we put these results with those on p. 121 together we find: 
Number 
of 
Groups 
Gaussian Frequency 
Equal 
Subfrequencies 
Pearson's Scale 
General 
Intelligence 
Pearson's Scale 
General Health 
8 
•891 
•881 
•895 
4 
•928 
■932 
•927 
5 
•947 
•947 
6 
•959 
•966 
It will be noticed that the frequencies differ considerably, although not 
extremely, from equality. Accordingly we may find it sufficient in many cases, 
