Karl Pearson 
131 
measured on my scale (see Biometrika, Vol. VIII., p. 93) and clothing was divided 
into five categories, IV and V being ultimately classed together. The better 
intelligence has the higher letter, the worse clothing the higher number. The 
table runs : 
Intelligence. 
a 
IS 
o 
O 
B 
C 
D 
E 
F 
G 
Totals 
I 
33 
48 
113 
209 
194 
39 
636 
II 
41 
100 
202 
255 
138 
15 
751 
III 
39 
58 
70 
61 
33 
4 
265 
IV and V 
17 
13 
22 
10 
10 
1 
73 
Totals 
130 
219 
407 
535 
375 
59 
1725 
I shall use x for Intelligence and y for Clothing. Mr Gilby found raw 
= •1014, corrected for 4x6 cells <£ 2 = "0927 and C=-291. I find from left to 
right 
+ 1-1179, + 2-2107, 
s 
X 
Na-J 
= •812,475, 
= •9014. 
a, = -1-8859, - 11009, - -4758, + -2428, 
^ = - 1-0066, + -2173, +1-2130, +2-1317. 
Hence 
fan 'Y 1 - \ 
= -933,393, 
= •9661, r yCy -. 
If we now correct Mr Gilby 's contingency for the class-index correlations of 
x and y we have 
This is very near the values, '343 and - 340, found by converting the table into a 
three-rowed table, classes III, IV and V of clothing being grouped together and 
a bi-serial 77 method used (Biometrika, Vol. VIII. p. 98). 
I then proceeded to work out on the full table given above the value of r 
as determined by the approximate formula (xvii). This can be conveniently 
arranged thus : 
- 1-8859 
- 1 -1009 . 
- -4758 
+ -2428 
+ 1-1179 
+ 2-2167 
-1-0066 
33 
48 
113 
209 
194 
39 
+ -2173 
41 
100 
202 
255 
138 
15 
+ 1-2130 
39 
58 
70 
61 
33 
4 
+ 2-1317 
17 
13 
22 
10 
10 
1 
+ 33-2178 
+ 48-3168 
+ 113-7458 
-210-3794 
-195-2804 
-39-2574 
- 8-9093 
-21-7300 
- 43-8946 
+ 55-4115 
+ 29-9874 
+ 3-2595 
- 47 -3070 
-70-3540 
- 84-9100 
+ 73-9930 
+ 40-0290 
+ 4-8520 
-36-2389 
- 27-7121 
- 46-8974 
+ 21-3170 
+ 21-3170 
+ 2-1317 
-59-2374 
-71-4793 
- 61-9562 
- 59-6579 
- 103-9470 
-29-0142 
