76 Salisbury . — Variation in Eranthis hy emalis , 
or two-fifths phyllotaxis, the calyx commonly consists of from three to five 
members. Such correspondence between the orthostichies of the vegetative 
leaves and calyx-segments has no significance unless the latter be derived 
from the former, since the staminal orthostichies are usually much more 
numerous. The case of Clematis alpina might be regarded as an exception, 
since here there are four staminodal petals alternating with the four sepals. 
But the spacial relations here determine the position of the inner whorl just 
as they determine that of the corolla in Ranunculus or of the stamens in 
Rosaceae. Such an explanation cannot, however, be applied to the outer- 
most whorl. 
It is, therefore, obvious that to regard the perianth as an homologous 
structure derived from the stamens involves us in several assumptions for 
which there appears no warrant. On the other hand, these very difficulties 
fall into place if we accept the view that the perianth is in some cases 
(e. g. Anemone) entirely foliar in origin (cf. Prantl, 1887) and in others (e. g. 
Ficaria) derived partly by modification of leaves (the calyx) and partly from 
sporophylls (the corolla) (cf. Grant Allen, 1882). 
Great stress has been laid by some writers on the existence in this 
group of polymerous perianths (Worsdell, 1916, p. 129 ), the argument 
being briefly that the supernumerary perianth segments could not have 
arisen de novo and must therefore be regarded as transformed stamens. 
That fission is a common phenomenon in the calyx of the Ranunculaceae, as 
also in the other floral whorls, has been amply demonstrated in the foregoing 
pages. As pointed out by de Vries (1893), the unilateral character of the 
Galtonian half-curves, as exhibited in the perianth variation of Caltha 
palustris or Ranunculus bulbosus , can be explained if we assume increase to 
be due to dldoublement. The same feature is also encountered in allied 
families; for example, Marchand (1863, p. 127) recorded the occurrence of 
dedoublement in the androecium of Epimedium musschianum , extreme 
examples exhibiting eight stamens in place of the usual four. The structure 
of the normal flower of Podophyllum peltaium illustrates fission both in the 
androecium and perianth. 
Polymerous perianths can therefore be best explained as due to fission 
(cf. also Rendle, 1903), and the fact that increase in their members is usually 
accompanied by an increase rather than a decrease in the androecium 
renders the transformation theory in the highest degree improbable. 
It has been shown that there is a fairly widespread tendency for the 
androecium and gynaeceum in this group to consist of parts which number 
some multiple of three, and examination of the variation * curve , renders it 
very probable that this feature is the result of a trimerous tendency (cf. 
p. 64 ). It seems most likely, then, that an arrangement of parts on six or 
three orthostichies was primitive for the group, but owing to increase in the 
number of members through fission this arrangement has become obscured. 
