40 HEREDITY [ch. 



Prof. Pearson and his collaborators have worked 

 out the correlation between parent and child for a 

 number of measurable characters in Man, Animals, 

 and Plants, and they find that the numbers group 

 themselves about a value not far from 0*48, varying 

 from 0-42 to 0-52. That is to say, on the average the 

 offspring deviate from the mean about half as much 

 as the parent. 



The parental correlation hitherto discussed has 

 taken no account of the second parent, for if in- 

 dividuals mate at random the one parent may be 

 considered alone, and the second will on the average 

 have the mean value for the general population. But 

 it is clear that one may take for the parental value 

 in each class the mean of the two parents (making 

 allowance for any difference in measurement due to 

 sex), and plot the means of the sons (or daughters) 

 against the classes so produced. The value derived 

 from taking the mean of father and mother is called 

 the * mid-parent,' and the correlation so arrived at 

 would give the measure of resemblance between 

 children and their mid-parents. This is naturally 

 higher than the correlation observed when only one 

 parent is considered; for if both parents deviate in 

 the same direction from the mode of the population, 

 the children will average a gi'eater deviation than if 

 only one does so, and still more than if one deviates 

 in one direction, the other in the opposite. We thus 



