No. 565] CHANGES PRODUCED BY SELECTION 1 J 



Connecticut. It was a descendant in a collateral line of 

 the plant used by Shamel in 1903 as the female parent in 

 his cross. 



Table II, giving the frequency distribution for the num- 

 ber of leaves of the two parents and the first and the 

 second hybrid generations, is a complete justification of 

 our prediction as to how the hybrid type produced by 

 Shamel originated. The "Sumatra" and the F x genera- 

 tion were grown at New Haven, Connecticut, in 1911, the 

 "Havana" was grown at Bloomfield, Connecticut, in 1911 

 from commercial seed of the same variety as the plant 

 used for the male parent, while the F 2 generation was 

 grown at New Haven, Connecticut, in 1912. The gen- 

 eration, producing an average of 23.3 ± .14 leaves per 

 plant, is intermediate in leaf number, since the "Havana" 

 variety shows an average leaf number per plant of 19.8 

 ± .08 and the "Sumatra" variety 26.5 ± .11. The varia- 

 tion as determined by the coefficient of variability is some- 

 what less for the Fj than for either parent. The value 

 for the "Sumatra" variety is 6.64 per cent. ± .28 per 

 cent., for the "Havana" variety 6.98 per cent. ± .27 per 

 cent, and for the F 1 generation 6.24 per cent. ± .41 per 

 cent. Taking into consideration the probable error in 

 each case, one may say that the variability of the three 

 populations is almost the same. 



The variability of the F 2 generation, however, is greatly 

 increased. This is shown by the high coefficient of vari- 

 ability, 10.29 -± .23 per cent., although a glance at the fre- 

 quency distribution with its range of from 18 to 31 leaves 

 brings home the point without recourse to biometrieal 

 calculation. 



The appearance of the plants in the field corroborated 

 the data of Table II in other characters. The F, genera- 

 tion was intermediate in the various leaf characters, such 

 as shape, size and texture, that distinguish "Sumatra" 

 from ' ' Havana ' ' tobacco, and in these characters it seemed 

 as uniform as either of the parental varieties. On the other 

 hand, the F 2 generation was extremely variable. Some 

 plants could not be distinguished from the pure "Soma- 



