IV] STATISTICAL STUDY 35 



sons will on the average diverge less ; they will vary 

 about a modal value lying between the general mode 

 and the fathers' measurement. This fact is called 

 * regression.' It sometimes seems paradoxical to those 

 who have not considered it that the mean deviation 

 of children from the general mode is always less than 

 that of their parents. But of course it does not mean 

 that all sons of tall fathers will be shorter than their 

 fathers ; some will be as tall or taller, but the sons of 

 a number of fathers of given stature will vary about 

 a mode lying between the fathers' stature and the 

 mode of the whole population. 



Now it is plain that the amount of regression is a 

 measure of the intensity of inheritance ; if the modes 

 for sons of fathers of every deviation have deviations 

 nearly as great as those of the fathers, the intensity of 

 inheritance would be high ; if the modes for the sons 

 deviate but slightly from the general mode, whatever 

 be the deviation of the fathers, the intensity would 

 be low. A deJQnite case may make this clearer. 

 Suppose the modal stature of the population is 68 

 inches ; it might then be found that for fathers of 64 

 inches (i.e. deviating 4 inches below), the height of 

 the sons ranged from 61 to 72 inches. If, however, 

 their modal value had a deviation only slightly less 

 than the fathers' deviation, say with a mode at 65 

 inches, the regression would be slight and the intensity 

 of inheritance high ; if the sons' mode had a deviation 



3—2 



