No. 565] CHANGES PRODUCED BY SELECTION 



Eepresentative plants of the plus and minus strains of 

 family 19 as obtained by three years of selection at New 

 Haven are shown in Figs. 4 and 5. 



Family No. 5 (Table VI) shows a decrease in mode 

 from 28 to 26 leaves, and a similar decrease in mean from 

 28.1 ± .06 leaves to 26.6 ± .09 leaves as a result of the first 

 minus selection. A second minus selection, however, in- 

 dicates either that the future progress is to be very slow 

 or that the entire effect of selection was manifested in the 

 first selected generation. 



"With the three parts of Table VII we take up the re- 

 sults on Family No. 6 at all three stations. The minus 

 strain was carried on only two generations at Bloomfield, 

 but with this exception there are data upon three genera- 

 tions. At Bloomfield the two generations of selected 

 minus extremes resulted in 0.6 leaf decrease in the mean, 

 but at New Haven the results were negative, the means 

 advancing from 25.8 ± .06 leaves to 27.9 ± .12 leaves in 

 three generations, while at Forest Hill the mean remained 

 practically the same. Surely selection was unprofitable 

 here. 



The first year of selection from the other end of the 

 curve, however, resulted in marked progress. The mean 

 advanced nearly 5 leaves in each case. The original F 5 

 mean is 25.8 ± .06 leaves, but the three F 6 means are 30.7 

 ± .09, 29.6 ± .08 and 30.8 ± .12 leaves. This is a remark- 

 able concurrence of results. The means in the two suc- 

 ceeding generations were about the same in the Bloomfield 

 and New Haven experiments, but there was another defi- 

 nite advance at Forest Hills. Such a result should not 

 be unexpected. If the F 6 generation were almost but not 

 quite a homozygous lot, and if one assumes that selection 

 of extremes from homozygous population has no effect 

 in shifting the mean, it would frequently happen that 

 some individuals selected to continue the line would be 

 homozygous in all factors and some heterozygous in one 

 or more factors. 



The cause of the peculiar distribution of the population 

 (high variability) of the F s generation grown in Bloom- 



