500 Agnes Arber. — On the S 'true here of the Androecium in 
recognizable long before the bundle leaves the axis. This centripetal xylem 
I regard as representing vascular strands originally destined for super- 
numerary anthers, which have now entirely vanished. The numerous 
bundles of the staminode, also, remain free to the base and enter the axis 
separately, agreeing in this point with the stamen bundles of Hypericum. 
It may be asked why it should be supposed that vestigial vascular 
strands have been retained in the Parnassia stamens, instead of being 
entirely obliterated. I think the answer may possibly be that they owe 
their survival to their conversion to a secondary use — an occurrence that is 
not uncommon in the case of vestigial structures. 1 
The stamens of Parnassia execute remarkable movements, each in 
turn elongating, so as to bring the anther into a position above the 
immature stigmas, and then bending outwards and downwards after 
shedding its pollen. 2 The addition of the internal xylem helps to convert 
the wood of the main bundle into a slender hollow cylinder, which may 
have proved of some value in giving the filament certain additional qualities 
of strength and elasticity. 
IV. Discussion of the Affinities of Parnassia. 
The structure and affinities of the genus Parnassia have been 
admirably discussed by Drude 3 in an exhaustive monograph published 
in 1875. The conclusion at which he arrived was that Parnassia was best 
placed in the special family Parnassieae, included in the large and complex 
‘ Nexus Saxifraginae ’, but forming a transition between this nexus and the 
Droseraceae, on the one hand, and the Hypericineae, on the other. He 
regarded the affinity with the Hypericineae as less close than that with 
either the Saxifragaceae or the Droseraceae. Since I find myself in general 
agreement with this conclusion, and with the majority of the arguments on 
which it is based, I do not propose to attempt an exhaustive discussion of 
the whole question of the affinities of Parnassia. 
The relation between Parnassia and the Droseraceae was dealt with 
very fully by Drude from the standpoint of external morphology. More 
recently the subject has been reopened by Eichinger 4 and by Pace. 5 
The latter author has obtained fresh evidence by the use of modern 
cytological methods. From a detailed study of the embryo-sac and 
1 As Darwin points out, the style of the male florets of some Compositae remains well developed 
although the ovary and stigmas are abortive. This is supposed to be due to the fact that it has 
retained its secondary function of brushing the pollen out of the anther-tube, though it has lost 
its primary function of conducting the pollen-tube to the ovule. The Origin of Species, sixth ed., 
London, 1894, p. 373. 
2 Gris, A. : Sur le monvement des etamines dans la Parnassie des marais. Comptes rendus, 
vol. lxvii, 1868, p. 913. 
8 Drude, O. : Ueber die Bliithengestaltung und die Verwandtschaftsverhaltnisse des Genus 
Parnassia. Linnaea, Bd. xxxix, 1875, p. 239. 4 Eichinger, A. : 1 . c. 
5 Pace, L. : Parnassia and some Allied Genera. Bot. Gaz., vol. liv, 1912, p. 306. 
