42 First Results from the Oxford Anthropometric Laboratory 
The actual record of how many people succeeded at the first attempt, how 
many at the second attempt and so on is not of much interest, but what is of 
interest are the correlations found to exist between success at this test and 
intellectual eminence as measured by examination. The fourfold table method 
was employed for measuring these correlations, but it is only claimed that the 
coefficients so found give a rough idea of the degree of correlation. 
A.S a specimen table the following may be taken : 
Spot Pattern Test and Class in filial Schools. 
Classes I and II 
Classes III and IV and Pass 
Totals 
5 tries or less 
27 
26 
53 
6 tries or more 
18 
29 
47 
Totals 
45 
55 
100 
r=-2. 
This table includes all those who had taken their final Schools by the end of 1909, 
the following year's results having appeared too late to be included. It will be 
seen that the first and second class men are distinctly better than those included in 
Classes III or IV or who obtained pass degrees, since 60 per cent, of the former 
and only 46 per cent, of the latter mapped the pattern correctly in five attempts 
or less. The correlation coefficient calculated from this table is "2 which indicates 
a low degree of correlation. It should perhaps be explained that when the corre- 
lation is perfect, i.e. when a variation of any particular degree of the one variable 
is always associated with a variation of the same degree of the other variable, 
then the coefficient is equal to unity ; when no correlation exists it is equal to 0. 
Since the numbers are small and the correlation is low, the conclusion drawn 
from this one table could not be regarded as safe if considered independently, but 
as it receives corroboration from the study of the results of scholarship examinations 
it may be accepted with fair confidence. 
The correlation between the possession of a scholarship or exhibition and success 
at the test under consideration is positive for each age group, though as the 
numbers in each such group are small it has a considerable variation. The 
average value for all age groups taken together is "22. Here again the corre- 
lation coefficients are not high, nor, since they are based on small numbers, are 
they particularly constant ; but some degree of positive correlation is -exhibited 
by every age group, so that the probability of the result being partly a chance one 
is so small as to be negligible. 
Further, an investigation was made of the relation between an undergraduate 
subject of study and his success at this test, with the result that those reading 
science or mathematics were found to be superior to those reading other subjects. 
