Parnassia and its bearing on the Affinities of the Genus. 499 
the female flower of the Oak. In this authors words : ‘ In some cross- 
sections of a young flower the stamens were found to be represented by 
small outgrowths alternating with the stigmatic lobes. Just below these 
minute outgrowths, small branches from the vascular bundles supplying the 
perianth end in little irregular masses of reticulately thickened cells. In 
another flower of about the same age, the outgrowths are absent, but the 
small branch bundles persist.’ 
From a consideration of the cases just quoted, it may, I think, be 
conceded that there is at least a possibility that the anomalous vascular 
structure of the Parnassia stamen may be explicable on phylogenetic 
grounds. The peculiarities of the bundle appear to suggest that the 
vascular system of the filament may have had a compound origin. The 
numerous strands in the connective, the centripetal xylem in the filament, 
with the accompanying indications of numerous phloem groups, point to 
the existence of vestigial vascular strands, now closely associated with the 
main bundle, but perhaps, ancestrally, totally distinct from it. 
The suggestion I wish to put forward, in order to account for these 
facts, is that the structure of the stamen of Par?iassia is best interpreted as 
a reduction-stage from a phalange of stamens, such as that which is found 
in Hypericum among the Hypericineae . 1 The staminode of Parnassia has 
already been explained on these lines , 2 but, since the fertile stamens show 
no external sign of being compound structures, the theory has not, so far as 
I am aware, ever before been extended so as to include them. In certain 
species of Hypericum the number of stamens in each fascicle is very small ; 
in Hk virginicum , L ., 3 for instance, there are only three. H. Elodes , Huds., 
has the smallest number of stamens of any British St. John’s Wort, and, 
as the filaments are also united for a considerable part of their length, 
it is the most suitable species with which to compare Parnassia. 
Text-fig. 4 (b-e) shows transverse sections of the essential organs of 
a flower of Hypericum Elodes , in which the stamen-phalanges consisted 
of five, five, and three members respectively. It will be noticed that most 
of the vascular bundles in the common filaments preserve their identity to 
the extreme base (Text-fig. 4 , E and F, c.f.). Both in H. Elodes , and in 
two species of Hypericum with numerous stamens in the phalange, I have 
found that the bundles intended for each individual stamen pass, as a rule, 
separately from the receptacle into the, common filament. This strengthens 
the analogy with the Parnassia stamen, since here the centripetal xylem is 
as in the case of the perianth of the male flower of Ephedra or the abortive ovule of that oi 
Wehvitschia .'*) 
1 The question whether the stamen-fascicles of Hypericum are due to chorisis of single stamens, 
or to the fusion of originally free and numerous stamens, does not affect the present argument, and 
will not be discussed here. 
2 Lindley, J. : The Vegetable Kingdom, London, 1846, p. 405. 
3 Gray, Asa : Genera Florae Americae Boreali-Orientalis, vol. i, Boston, 1848, PI. 94. 
M m 
