36 HEREDITY [ch. 



much less than the fathers', say at 67 inches, the 

 regression on the general mean would be considerable 

 and the intensity of inheritance low. If then we can 

 find means of determining the ratio between the 

 deviation of sons in general and the deviation of their 

 parents, we shall have a measure of the intensity of 

 inheritance for the character considered. This ratio 

 is called the ' coefficient of correlation ' between 

 father and son for that character. It should be 

 noticed that correlation simply means that two 

 quantities vary in relation to each other ; the corre- 

 lation between parents and children is a convenient 

 method of estimating the intensity of inheritance, 

 but correlation exists between any two related 

 variables, e.g. between the measurements of two 

 limbs in the same individual, such as an arm and a 

 leg, or between the numbers obtained in successive 

 throws of dice, if not all the dice are picked up for 

 the second throw. The correlation between the same 

 measurement in brothers may be used as a measure 

 of inheritance, for two brothers resemble each other 

 more than two chance individuals because they are 

 children of the same parents. 



The principle of obtaining a coefficient of corre- 

 lation between father and sons is as follows. It will 

 be convenient to assume that the variability of the 

 character considered is normal, i.e. that the frequency 

 curve falls evenlv on either side of the mode, so that 



