DISCONTINUOUS VARIATION. 63 



for Bowing the following year. The proportion of four 

 leaflet plants steadily increased, and in the fourth gen- 

 eration the most widely diverging plants had the fol- 

 lowing (percentage) proportions of leaves with from 

 3 to 7 leaflets. 



Extreme variation, 12 9 22 17 40 97 



The plant considered as " normal " was obviously of 

 a five leaflet type, the numbers of leaves with 3 and 4 

 leaflets, and those with 6 and 7 leaflets, being dis- 

 tributed symmetrically around it. A comparison of the 

 numbers of leaflets in the " atavistic plant " with those 

 in the " extreme variation " is interesting, as showing 

 the range of variation possible in plants of the same 

 stock. 



As an instance of the formation of a variety, De 

 Vries' experiments with Chrysanthemum segetum 

 grandiflorum may be quoted.* Starting in 1896 with 

 plants which had 21 ray florets occurring most fre- 

 quently in their capitula, and none of which had more 

 than 23 florets, he picked out each year the two or 

 three plants richest in florets for breeding with, and 

 sowed their seed the following year. In 1897 a single 

 flower was obtained having 34 florets, but the 21 floret 

 form was still the commonest. In 1898 one of the 

 flowers had 48 florets, the commonest forms now having 

 26 or 34 florets. In 1899 one had 67 florets, the com- 

 monest forms having 26 or 33 to 35 florets, and in 1900 

 *Loc. cit.,p. 523. 



