No. 588] EXPERIMENTS IN MASS SELECTION 717 



the different basis of selection in the two periods. But 

 it is hard to believe that this entirely explains the dif- 

 ference in result. One notices for example that during 

 the period of ostensible decline the highest average fecun- 

 dity (45.23) is recorded when the number of birds under 

 observation is smallest (48) and the lowest average 

 (19.93) is recorded when the flock is largest (780). Fur- 

 ther, in the later seven-year period of "improvement," 

 the number of birds tested declines as their average 

 fecundity rises. Has not the better environment and 

 lessened competition of small numbers possibly some- 

 thing to do with the changes noted! Is it certain that 

 genetic agencies are responsible for the differences ob- 

 served? Pearl himself nowhere states that the selection 

 practised during the earlier period had produced posi- 

 tive deterioration; he merely states that "there was no 

 change of the mean in the directon of the selection" dur- 

 ing this period when selection was based on high produc- 

 tion without progeny tests. But as soon as progeny tests 

 were made an additional feature of the basis for selection 

 Pearl notes immediate results, viz., the immediate isola- 

 tion of a strain which in its first year made a record for 

 high productiveness only once equalled in the six subse- 

 quent years. How many successive selections were made 

 in this period, we are not informed, but since it would 

 require at least two years to make a combined perform- 

 ance and progeny test, it would seem that not more than 

 three successive selections can have been 'carried out on 

 this basis in the seven year period from 1908 to 1915. It 

 may fairly be questioned whether this is an adequate test 

 of the effectiveness of mass selection. The total number 

 of individuals tested during this period is, according to 

 Pearl's table, 1,655. For the entire seventeen years of 

 selection it is 4,842. 



The total number of animals graded in our selection 

 experiments with rats heretofore published is 20,645, and 

 the number of generations involved 13. Since those 

 figures were compiled, four additional generations of 



