18 



THE AMEBIC AN NATURALIST [Vol. XLVIH 



isolated seed pans filled with sterilized soil made it im- 

 possible to start more than two sets of plants for each 

 plus and each minus selection. Generally both sets grew 

 perfectly, but occasionally both failed, and in that case it 

 was usually too late in the season to start a third set even 

 if it were available. 



The second part of Table III shows the results obtained 

 on the poor soil of New Haven, Connecticut, with the same 

 family. There was continuous progress in both direc- 

 tions. The minus selections during the three generations 

 show a constant reduction of mode, the figures being 23, 

 22 and 21; the plus selections show an even greater in- 

 crease in mode, the figures being 25, 27 and 28. The same 

 decrease and increase occur in the means until in the F 9 

 generation there is a difference of nearly 9 leaves, the cal- 

 culated means being 20.9 ±.08 leaves and 29.7 ±.14 leaves, 

 respectively. 



Figs. 1 and 2 show typical plants of the plus and minus 

 strains of this family as developed by 3 years of selection. 

 Fig. 3 illustrates an interesting change of phyllotaxy in 

 some plants of (77-2)-l-l as grown at New Haven in 1912. 



Passing to the data on Family No. 76 (Table IV) there 

 is the same evidence of the effectiveness of selection, ex- 

 cluding the minus strain in 1910, of which only 31 plants 

 were healthy. This effect is markedly less than with the 

 other family. The mode of the minus selection remained 

 at 24 leaves and the mean was reduced only from 24.1 

 ± .11 leaves to 23.9 ± .05 leaves,— hardly a significant 

 figure. The mode of the plus selection crept up to 26-27 

 and the mean to 26.9 ± .07 leaves, there being here one 

 more generation than in the case of the minus strain. 



Table V gives the data on plus and minus selections of 

 Family No. 19 at Bloomfield for two generations. The 

 original family stock of the F 5 generation has the mode at 

 27 leaves and the mean at about 26 leaves. A 24-leaved 

 plant of this generation became the parent of the minus 

 strain, giving in the F 6 generation a population with the 

 same mode and a slightly higher mean (26.9 ± .08 leaves). 

 Continuation of the strain through a 24-leaved plant gave 



