No. 565] CHANGES PRODUCED BY SELECTION 



39 



the tobacco grower loses an average of from one to two 

 of his most valuable leaves. 



The plus strain of Family No. 56, which we were dis- 

 cussing when we digressed to speak of the critical periods 

 of development, did show a considerable shifting of the 

 mean following the selection of high-leaved mother plants. 

 In the Bloomfield selections the mean went from 24.2 

 =fc .06 to 26.7 ± .08 leaves, then to 26.8 ± .07 leaves ; in the 

 New Haven experiment the mean shifted to 27.4 ± .08 

 leaves, — a gain of 3.2 leaves, — and then dropped to 26.4 

 ± .11 leaves, recovering again in the F 9 generation to 

 27.5 ± .11 leaves ; in the Forest Hills experiment the suc- 

 cessive means were 27.2 ± .08, 28.9 ± .08 and 26.7 ± .06 

 leaves. Summing up the data from this experiment, it 

 may be assumed to be reasonably certain that no progress 

 resulted from the selection of minus extremes, but that 

 there was a slight effect gradually diminishing in quan- 

 tity when plus extremes were selected. 



Representative plants of Family 56 obtained by three 

 years of selection in the effort to produce strains of high 

 and low leaf number, respectively, are shown in Figs. 

 8 and 9. 



Family No. K (Table XII) was grown on a farm near 

 the Bloomfield experiments, in 1910. The records of the 

 F 5 generation consisted of the number of leaves of only 

 31 plants. From among these individuals two plants 

 were selfed to become the mothers of the F 6 generation. 

 Since no dependence can be placed on the F 5 distribution 

 by reason of the few plants and since it is not absolutely 

 certain that the mother plants of F 6 had 20 leaves each, 

 the selection really began in 1911 with the F 7 generation. 

 There is a difference between the minus strain and the 

 plus strain in 1911 and 1912—0.5 leaves the first year and 

 1.3 leaves the second year, — however, so that one may 

 assume the possibility of a slow shifting of the mean in 

 both directions. 



The data on Family No. 73 are shown in Table XIII. 

 This family came from a 28-leaved plant, one of the 

 highest of the F 5 generation. The F 6 progeny of this 



