GALTON'S LAW 247 



the ancestral contributions is expressed by the series (0.5) + 

 (0.5) - + (0.5) ^ etc.] which being equal to 1, accounts for 

 the whole heritage." 



If one attempts to make use of this law by basing upon it 

 predictions as to the character of the offspring in particular 

 kinds of matings, it works fairly well when blending charac- 

 ters are under consideration, but fails completely when ordi- 

 nary Mendelizing characters are under consideration. See 

 Castle (1903). As a useful generalization it is now pretty 

 generally discredited. 



Regression was a name given by Galton to the apparent 

 going back of offspring from the condition of their parents 

 toward that of more remote ancestors, or more correctly 

 toward the general average of the race. Thus he observed that 

 very tall parents have children less tall than themselves, 

 while very short parents have children taller than themselves. 

 In either case the children regress toward the general average 

 of the race, and the regression is greater the more pronounced 

 the deviation of the parents from the general average of the 

 race. Also in sweet peas, Galton observed that when very 

 large seeds are planted, the crop harvested averages smaller 

 in size than the seeds planted; and that when small seeds 

 are planted, the crop averages larger in size. Regression 

 occurs in both cases toward the mean of the race. Galton 

 regarded regression as a feature of ancestral heredity; but 

 Johannsen has shown, as regards size of beans, that regres- 

 sion is due to a lack of agreement between somatic and 

 genetic variations, the latter being more conservative, and 

 that when selection is made within a line pure genetically, no 

 regression occurs. Davenport confirms this view in the case 

 of human stature, showing that the children of parents 

 genetically pure for tall stature do not regress toward me- 

 diocrity, as Galton supposed all classes of a population to do. 

 Gallon's law of ancestral heredity and his principle of regres- 

 sion are now chiefly of historical interest, but it is well to 

 keep them in mind when generalizations based on similar 

 reasoning are brought forward. (See Chapter XXVI.) 



