Ficaria verna , and other Members of the Ranunculaceae . 5 1 
more than six perianth segments are present the increased number of 
primordia may be regarded as due to a tendency to produce a polymerous 
perianth of which the bifurcated members are an incomplete expression. 
The two antagonistic tendencies, towards trimery and multiplication of 
parts, have resulted on the one hand in the production of the numerous 
instances of supernumerary perianth segments, and on the other in the more 
or less complete fusions exhibited as notched and bilobed structures. That 
these latter are to be so interpreted seems indicated by the occasional 
presence of slightly notched perianth segments in which the two halves do 
Fig. 3. Eranthis hyevialis. Empirical diagrams showing arrangement of honey-leaves 
and stamens. 
not stand in the same tangential plane. As a result there is a longitudinal 
kink where the two halves join that seems best interpreted as due to fusion 
between two primordia not situated precisely side by side. The rarity of the 
latter phenomenon is, however, strong presumptive evidence that the fusing- 
pairs of primordia were themselves the product of the congenital fission of 
a single primordium, and that subsequent fusion has taken place along the 
original plane of separation ; a view supported by the fact that the super- 
numerary perianth segments are generally much narrower than those 
normally present (Fig. 5, I). The position which they occupy in the flower 
is also in agreement with this hypothesis (cf. Fig. 6 , A and C). Evidence 
based on numerical relations will, moreover, be adduced to show that there 
is no ground for the assumption that the supernumerary members are the 
result of transformation either of honey-leaves or of stamens. Further, the 
