432 



THE GRAMMAR OF SCIENCE 



Eye-Colour of Husbands — 774 cases. 



Now this table is not like the correlation tables up till 

 now presented to the reader. While hitherto each column 

 gives an array corresponding to unit increase in a char- 

 acter, here there is no true quantitative scale at all ; our 

 numbers merely refer to certain groupings, arranged it is 

 true in increasing darkness of colour, but in no way cor- 

 responding to equal increases in colour intensity. Hence 

 new methods have to be employed for quantitatively deter- 

 mining the correlation. Now the general principle em- 

 ployed is very simple and can be easily understood 

 without complex mathematics. Suppose we divide all 

 men into two classes, say for example, (a) those with hazel 

 or lighter eyes ; (d) those with light brown or darker eyes. 

 Let the numbers in these two classes among N men be 

 n^ and fi^. Similarly let the numbers among N women, 

 who fall into these two classes, be w^ and w^. Then 

 what would happen if there were no assortative mating, 

 i.e. if men selected their wives at random ? Clearly the 

 n^ men would select wives falling into the (a) or the (d) 

 classes purely in proportion to the numbers in those 



classes, or we should have (a) mates with (a) in n^ x -^ 

 cases, and (a) with (d) in n^ x ^ cases. Similarly with 



