100 
Ne-w Method of Correlation 
From these results we conclude that : 
(i) The relation between anaemia and age (7 to 13) is not very marked. 
(ii) Anaemia decreases as the boys grow older and increases as the girls grow 
older. 
(iii) The association of anaemia with age in boys is about trebled in intensity 
when we pass from Sirdar Road to Laxon Street, or from a school with upwards 
of 30 to one with more than 50 "/^ of anaemic boys. 
The above distributions are perhaps not quite the type we should select as 
most suitable for the method — the totals for each age are too nearly alike. It 
would probably, however, be difficult to bring out by any more exact method, 
leading to a unique result, the correlation of age and anaemia. It is obvious that 
a fourfold table method would present considerable diversity according to the age 
partition selected and would require far more labour than the 15 minutes requisite 
to determine a correlation-coefficient by the present method. 
Illustration II. The Influence of Age on the Capacity to pass an Examination. 
I take here the returns for 6156 candidates for the London University Matricula- 
tion Examination from the Calendar for 1.908-9 (pp. 953-955). 
Ages of Candidate 
Passed 
Failed 
Totals 
Iti 
583 
563 
1146 
17 
666 
980 
1646 
IS 
525 
868 
1393 
19—21 
383 
814 
1197 
22—30 (centred at 25) 
214 
439 
653 
Over 30 (centred at 33) 
40 
81 
121 
Totals 
2411 
3745 
6156 
The only difficulty here is the centering of the groups "22 — 30" and "over 30," 
the official statistics (as usual !) clubbing all these ages, which are really so signi- 
ficant for the frequency distribution, together. After some consideration, I made 
the mean ages of these groups 25 and 33. 
The mean age of the passing candidates was 18*4280 and of the total candidates 
18-7865. The S.D. of the latter being 3-2850. Thus p = - -3585, pja^ = - -1091. 
Further (1 - a) = -3917, giving |(l-fa) = -6083 and 2 = -3843. Thus 
5/o-, = -9806, and r = -'in3. 
We conclude that there is a small but sensible correlation between youth and 
the ability to pass this examination. There are some evidences that this holds for 
other examinations. It would be interesting to consider this problem at greater 
length. While the complexity of the brain may be greater at 28 than 18, the 
