Parnassia and its bearing on the Affinities of the Germs. 509 
differs from that of most flowering plants in such a way as to suggest that 
the stamen has been derived by reduction from a * stamen-fascicle 5 , the only 
traces of ancestral complexity being retained vestigially in the vascular 
system. The stamen of Parnassia would, on this view, hold the same 
relation to any ordinary stamen that a ‘ unifoliolate compound leaf ’ 1 bears 
to an ordinary simple leaf. 
In conclusion, we may put in a plea for the reinstatement of Parnassia 
in the separate Order Parnassieae. That it should be included in the tribe 
Saxifragoideae of the Order Saxifragaceae seems most unsatisfactory, since 
this arrangement tends to obscure its relationships in other directions. 
I entirely agree with Drude’s conclusion that Parnassia shows un- 
doubted affinities with Saxifragaceae, Droseraceae, and Hypericineae, but, 
bearing in mind the new evidence adduced in the present paper, I am 
inclined to think that Drude somewhat under-estimated the closeness of the 
relationship between Parnassia and the Hypericineae. 
V. Summary. 
The chief points regarding the androecium of Parnassia brought 
forward in the present paper are as follows : 
1. In the course of a description of the general vascular symmetry 
of the flower of Parnassia palustr is , L., it is shown that the strands destined 
for the stamens arise as independent bundles at a lower level in the 
receptacle than those destined for the staminodes. 
We are thus able, on anatomical grounds, to confirm Drude’s view, 
based on developmental evidence, that the nectaries or staminodes of 
Parnassia form the inner whorl of the androecium. 
3. In Parnassia palustris , L., the bundle which traverses the filament 
is found to be accompanied by centripetal xylem, and there are indications 
of numerous phloem groups arranged round the xylem. A similar structure 
has also been found in the filaments of P . fimbriata> Banks. 
It is suggested that these peculiarities of the stamen anatomy are due 
to the presence of vestigial vascular strands which indicate that each 
individual stamen of Parnassia is reduced from an ancestral stamen-fascicle, 
comparable with that occurring in Hypericum. 
On the general question of the affinities of Parnassia , Drude’s view 
is accepted, namely, that this genus should be placed in the Order Par- 
nassieae, related to the Saxifragaceae, Droseraceae, and Hypericineae. 
1 i. e. a leaf such as that of Berberis § Euberberis , which is regarded as derived by reduction from 
a compound leaf such as that of the related genus Berberis § Mahonia. 
