THE PROBLEM OF THE FUTURE 21 



whether there is any reliable data or method by which 

 we can compare the nature and nurtnre factors in man. 



And, first, as to the method ; that is summed up in the 

 modern theory of statistics, and we term it the calculus 

 of correlation. But just as it would be impossible to 

 describe the principles of any branch of mathematical 

 analysis in the few minutes available in a public lecture, 

 so it is only possible to indicate to you very briefly the 

 nature and scope of this new calculus. The fundamental 

 difference between the categories of causation and 

 correlation is that we replace absolute association by 

 a weaker link. In causation A is always followed by B ; 

 in correlation A may be followed by B, C, D, or E, but on 

 the average with certain definite frequencies. Making 

 the frequencies of C, D, E, indefinitely small, we again 

 fall back on B absolutely associated with A, or our 

 correlation passes into the causation of the physicist. 



I will try and illustrate the idea as follows : Let us take 

 a pack of cards containing hearts and other suits, and 

 a bag of coloured balls which for simplicity we will say 

 contains the same number of balls as the pack contains 

 cards, and as many yellow balls as the pack contains 

 hearts. Now suppose we take n draws from the pack 

 and in the process x hearts appeared. We now take .v 

 yellow balls out of the bag. It will be clear that the 

 number of yellow balls is absolutely associated with the 

 number of hearts. There is a complete causal relation 

 between hearts and yellow baUs and we can predict one 

 from the other. This is the causation of the physicist, it is 

 the perfect correlation of the biometrician ; he represents it 

 by the ratio of balls to cards, or calls the correlation unity. 



Now let us modify our procedure. We will make p 

 card-drawings and if ii hearts occur we will draw u 



