EVOLUTION 



483 



cients of the offspring with their ist, 2nd, 3rd . . . nth, 

 etc., mid-parents. From these results I have constructed 

 the following table, exhibiting the influence of selection 

 during one, two, three, etc., successive generations in 



establishing a stock. 



I must ask the reader to examine this table attentively, 

 for it means a great deal for the problem of evolution. 

 We have seen (p. 471) that if a mid-parent with a char- 

 acter differing H^ from the type be selected, the offspring 

 of this mid-parent will on the average regress and have 

 only .6H1 of this character. Will not then the offspring 

 of these offspring have only .6 x (.6H1) of the character 

 on the average ? Certainly not, for they are not merely 

 offspring of mid-parents with .6H^, but their grandparents 

 had Hi of the character. In fact, as we shall see later, 

 these offspring may be expected to have .8049H1 of the 

 character. Thus selection is not checked by regression. 

 Reo-ression is merely the result of mediocre ancestry, the 

 moment we give selected ancestry the regression begins 

 to diminish, and in a few generations is hardly sensible. 



Why in our table above does .5H1 appear instead of 

 .6H1 as the amount of character inherited by the off- 

 spring? Because we have not taken all mid-parents of 

 character Hi, but have taken only those mid-parents of 



