MASS SELECTION. 171 



Generation 16 is the largest in the minus-selection series. It includes 

 1,980 young of mean grade 2.63. The grade of the parents is 2.79. 

 (See table 154.) 



Generation 17 (table 155) includes 868 young of mean grade 2.70. 

 The grade of their parents is 2.86. 



Four generations of selection have thus been added to the minus 

 series as it stood at the last report. The mean grade of the parents has 

 been advanced from 2.49 to 2.86; that of the offspring from 

 2.40 to 2.70, the former is an advance of 0.37, the latter of 0.30. 

 In the plus series the corresponding changes for one less generation of 

 selection (three), were 0.32 and 0.19, respectively. In both series a 

 change in the mean of the offspring attends that in the parents, coin- 

 ciding with it in character but not quite equaling it in amount. 



The lagging behind of the offspring, as compared with their selected 

 parents, gives a good illustration of regression, the phenomenon made 

 familiar by Galton's researches, but explained away by Johannsen as 

 due to a sorting-out action of selection on mixed races. The extent 

 to which in these experiments the offspring lag behind their parents 

 or "regress on their parents" is indicated in each table in the column 

 headed " regression." Tables 146 and 150 illustrate particularly well 

 how the offspring regress toward the general average of the race for the 

 time being. The offspring of parents substantially the same grade as 

 the general average of the race show no regression; the offspring of 

 parents below this average show regression upward (indicated hi the 

 tables by the minus sign); the offspring of parents above the racial 

 average show regression downward, the amount of the regression increas- 

 ing with the aberrant character of the parents. 



If one examines either selection series as a whole (compare Castle 

 and Phillips), he will notice that the point (toward which regression 

 occurs) changes with the progress of the selection. At the beginning 

 of the plus-selection series regression was toward a grade of about 

 4-1.75 (see table 1, Castle and Phillips); after about 15 generations 

 of plus selection it has advanced to +4.00. (See tables 148 to 150.) 

 At the beginning of the minus-selection series, regression occurred 

 toward a grade of to 1 (Castle and Phillips, tables 16 and 17); in 

 generation 17 (table 155) regression is apparently toward grade 2.62. 

 These grades toward which regression occurs represent points of racial 

 equilibrium or stability at which the race would tend to remain in the 

 absence of further selection, but these points of equilibrium are capable 

 of being moved either up or down the scale of grades at the will of the 

 breeder, provided he has patience and persistency and will select 

 repeatedly. 



Regression indicates that there is not complete agreement between 

 the somatic and the genetic character of the parents selected. But the 

 steady movement (in the direction of the selection) of the point of 



