Parnassia and its bearing on the Affinities of the Genus. 505 
essential organs of a flower of Parnassia palustris , that Lindley’s view is 
undoubtedly correct. The placentation for an extremely short distance 
close to the base is definitely axile, the ovary possessing four loculi (Text- 
fig- 3> F )- A little higher up the tissue connecting the septa breaks down 
(Text-fig. 3, E), but the septa themselves persist, terminating in T-shaped 
placentas (Text-fig. 3, d). The latter are at first almost in contact, but 
higher in the ovary, by the gradual shortening of the partial septa, they are 
brought closer to the ovary wall, so that the placentation, as seen in 
transverse section, appears more decidedly parietal (Text-fig. 3, c). 
The particular kind of parietal placentation found in Parnassia is thus 
easily seen to be a mere variant of the axile type. Further, if we examine 
the ovary of Hypericum Elodes (Text-fig. 4, B-E) for comparison, we find 
that the structure is essentially the same as in Parnassia. At the base, 
the ovary is divided into three loculi (Text-fig. 4, e), while, higher up, the 
septa become disconnected and give rise to parietal placentas (Text-fig. 4, d). 
In various other Hypericums the ovary is multilocular throughout, so that 
in this genus we can follow the transition between axile placentation and 
that form of parietal placentation characteristic of Parnassia. 
The resemblance between the seeds of Hypeidcum and Parnassia is 
generally recognized. Both are anatropous, and certain species of each 
genus have no endosperm in the ripe seed, while others are supplied with 
a single thin layer. 
When we attempt to compare the androecium in these two genera, we 
find one feature in which they differ very noticeably. The androecium of 
Hypericum o. onsists of either three or five ‘ fascicles ’ or ‘ phalanges ’ of fertile 
stamens, which may or may not alternate with an outer whorl of staminodes- 
In the cases where there are five of these fascicles they are placed opposite 
to the petals. In Parnassia , on the other hand, the fertile stamens alternate 
with the petals, and it is the staminodes which occupy a position correspond- 
ing to that of the stamens in Hypericum. When we turn, however, from 
the question of the relative positions of the sterile and fertile whorls to the 
form and structure of the individual members of which they are composed, 
we can scarcely fail to be struck by the analogy between the stamen- 
phalanges of Hypericum and the branched staminodes of Parnassia. This 
point is one of great interest, and it appears to have played a considerable 
part in suggesting the possibility of an affinity between these genera. 
Payer 1 has shown that, in ontogeny, the lateral stalked glands of the 
nectaries of Parnassia palustris are developed basipetally, and that they 
thus correspond to the stamen bundles of Hypericum , where the order of 
development is centrifugal. 
If this correspondence is to be regarded as complete, we must assume 
1 Payer, J. B. : Traits d’organogenie comparee de la fleur, Paris, 1857, p. 184 and PL 9, Figs. 26-9. 
