34 



THE AMERICAN NATUBAL1ST [Vol. XL VIII 



a changed gametic constitution, — a loss of gametic fac- 

 tors, — was fertilized by an unchanged cell. The un- 

 changed cell may have had any of the gametic possibil- 

 ities open to the germ cells of the 28-leaved plant of the 

 F 5 family in which the mutation arose, and we know that 

 certain factors in this plant were heterozygous, for pro- 

 gressive change followed the selection of a plus extreme 

 in the next generation. The 12-leaved plant was there- 

 fore a hybrid. It resulted from the union of a mutating 

 germ cell of the mother plant that furnished the F 6 gen- 

 eration with an unchanged germ cell. We can even as- 

 sume that the mutating germ cell, if fertilized by another 

 of the same kind, would have produced a plant with less 

 than 12 leaves. The reasons for believing this are simple. 

 There is experimental evidence (Hayes, 1912) that the 

 F 1 generation of a cross between varieties differing in 

 their number of leaves is intermediate in character. Our 

 12-leaved plant is the lone representative of such an F x 

 generation. The F 2 generation therefore should give 

 plants with less than 12 leaves, and in fact such plants 

 did occur. The distribution marked Fa in the table is 

 the F 2 generation, and this accounts for its extreme vari- 

 ability. The distribution marked Fb is the F 3 generation, 

 and its variability is less than half that of the preceding 

 generation. 



Family No. 56 was the second family to be grown at all 

 three of the experimental stations (Table XI). It arose 

 from a 26-leaved plant of the F 5 generation which pro- 

 duced an F 6 progeny with a mean of 24.2 ± .06 leaves and 

 a mode at 24 leaves. The three generations of the minus 

 strain grown at Bloomfield remained practically the same. 

 The last generation did indeed show a mean 1.0 leaf 

 higher than the original population, but no dependence 

 can be placed in data from only 25 plants. The data on 

 the minus selections grown at New Haven are for this 

 reason a little more dependable. They show a fluctuat- 

 ing mean, but no progress due to selection, the F n genera- 

 tion having a little higher mean than the F 6 generations. 

 The three minus selections grown at Forest Hills also 



