1 6 HEREDITY AND SOCIETY 



of the sons of any one of them would tell us very little. 

 Any one family might be affected by quite accidental 

 circumstances. But, if we measure the height of all 

 the sons of all our many six-foot fathers, the accidental 

 circumstances will tell as often in one direction as in 

 the other, and we shall be able to discover whether 

 or no, on the average, the sons of tall fathers exceed 

 the normal height of the race. As a matter of fact, 

 we shall find that the average height of the sons in 

 the case we have taken will be very nearly 5 feet 10 

 inches that is, 2 inches more than the average. Their 

 mean deviation from the normal is just about half that 

 of their fathers. In this case, we have said nothing 

 about the mothers. Had we restricted our choice of 

 six-foot fathers to those who had married women as 

 much taller than the average for women as they them- 

 selves were taller than the average for men, we should, 

 have found that the sons would, on the results of large 

 numbers, have more nearly approached the average 

 abnormal height of their parents, but that they would 

 still have fallen somewhat short of that excessive stature. 

 Let us now return to the relation between one parent 

 and his children of the same sex. We may express 

 the fact that the children deviate from the mean by 

 one-half as much as the parent by saying that the 

 " coefficient of correlation " for that particular character 

 is one-half, or 0*5. 



Turning to mental characteristics, we have greater 

 difficulty in exact measurement. But the marks of 

 candidates in an examination give a favourable instance 

 of variation. In a good examination, when the 



