26 



THE AM ERIC AX X AT V BALI ST [Vol.XLIV 



able to do in the case of the petals of Ranunculus could 

 only be determined by direct experiment. Possibly the 

 skew curve here found is merely due to the age or vigor 

 of the individuals. 



Considering next the distribution of the number of 

 primary leaflets per leaf in the material we may seriate 

 the number of leaflets for each class of plants separately, 

 as in Table IV. 



TABLE IV 



Number of 











49 1 126 

 34 8 



1 l 4S 



2 ~£ 3 



1005 i 277 



9 " | 1291 



Totals. 



1088 411 



16 1515 



Only sixteen leaves were available for the inflorescences 

 producing four leaves, but for those with two and three 

 leaves the number is ample. In each of the three classes 

 of inflorescences, and in the total, there is a pronounced 

 mode on undivided leaves and on those with three leaflets. 

 Throughout, those with two leaflets are much less fre- 

 quent than those with either one or three. 



Reducing the frequencies for the inflorescences with 

 two and three leaves to a percentage basis for more 

 direct comparison, and laying them side by side in Figs. 

 14 and 15, we note at once that there is a wide difference 

 in the proportion of single leaves in the two series. The 

 source of this difference is at once clear. In the inflores- 

 cences with the third leaf, the additional leaf is nearly 

 always simple. Thus the proportion springs at once 

 from about 4.5 per cent, to 30.5 per cent. 



Unfortunately, our data, though extensive, are not 

 numerous enough to permit of further analysis. If ma- 

 terial were more ample it might be possible to ascertain 

 more precisely some of the factors determining the con- 



