32 THE FAMILY AND THE NATION 



the mean height by 4 inches. If we measured the sons 

 of a large number of 6-feet fathers, we should find that, 

 while some were taller than their fathers and some 

 shorter, their average height was very nearly 5 feet 10 

 inches, that they exceeded the mean stature by 2 inches, 

 by half as much as their fathers. 



This result is expressed by saying that the coefficient 

 of correlation is about one-half, or 0.5. Had the sons 

 been just as tall as their fathers, the coefficient would 

 have been unity, and had the sons' average height 

 reverted to that of the race in general, there would 

 have been no relation between the variation of parent 

 and offspring, and the coefficient would have been zero. 

 In this way we can express conveniently the statistical 

 intensity of inheritance of any character. 



Closely similar values for the coefficients of correla- 

 tion have been obtained by investigating many other 

 common characters in men and in animals and plants. 

 The intensity of inheritance in all cases seems to be 

 represented by a coefficient lying between 0.4 and 0.6. 



It follows, then, that the ordinary variations found 

 in the usual races of mixed ancestry, whether of man or 

 other beings, are inherited. They are not wholly mere 

 fluctuations or chance differences of the individual, such 

 as Johannsen found with pure lines of bean plants. 

 They have a definite meaning in heredity, a real 

 selection value. 



If, instead of allowing the race to mate at random, 

 we selected both parents for some one quality, we could 

 raise the intensity of inheritance, and establish gradually 

 by continued selection a strain in which the quality 

 reached a value much higher than its average in the 



