502 Agnes Arber . — On the Structure of the A ndroecium in 
which are not infrequently ovate, often with a cordate or reniform base. 
It is true that this type of ‘ radical ’ leaf can be closely paralleled in certain 
Saxifrages, e. g. S', diversifolia , Wall., a Himalayan species, but it is no less 
true that this particular leaf-form is one which is markedly characteristic of 
geophytes in general. If we classify the terrestrial Dicotyledons of the 
British Flora, according to their leaf-form, we find that the great majority 
of the leaves, which approach those of Parnassia in shape, are borne by 
herbaceous plants perenniating underground, and producing a large pro- 
portion of their leaves ‘ radically ’. It seems a reasonable deduction that 
there is some correlation between a perennial geophytic habit and this 
particular type of leaf. Hence, it may fairly be argued that the close 
resemblance in habit and general appearance between certain Parnassias 
and Saxifrages growing in mountainous regions is more probably a sign of 
similarity of response to the special conditions of the environment than an 
indication of near relationship. It is a well-known fact that perennial 
geophytic types are markedly prevalent among the plants of the High Alps. 
To regard the case for the affinity of Parnassia with Saxifraga as 
being greatly strengthened by the resemblance in habit and leaf-form 
between certain members of these genera, seems to be a reversion to the 
ideas on taxonomy current in the latter part of the sixteenth century. 
Mathias de l’Obel, in his various works (e. g. ‘ Kruydtboeck ’, 1581), classifies 
plants almost entirely according to their leaf-form and habit. It is thus 
scarcely surprising that he should describe in succession the following 
plants, which we now regard as belonging to a number of different 
families : Soldanella , Convolvulus Soldanella , Parnassia palustris , ‘ Pthora ’ 
(? Ranunculus Thor a, L.), Cyclamen , and Aristolochia. Modern systems of 
classification, on the other hand, are based upon the well-tested hypothesis 
that it is in the reproductive organs, rather than the vegetative parts, that 
indications as to affinities are to be sought. It is true that the vegetative 
structure sometimes gives valuable clues to relationship, but these should 
always be subsidiary to those derived from a study of the reproductive 
organs. 
The leaf-form and habit are not the only vegetative features on which 
stress has been laid in discussions of the affinities of Parnassia ; attention 
has also been drawn to the presence of tanniferous cells in the leaf-epidermis 
of this genus, 1 and the occurrence of similar cells in members of the 
Saxifragaceae, e. g. Saxifraga cymbalaria and Chrysosplenium , as described 
by Engler 2 and Thouvenin. 3 Now, in Saxifraga cymbalaria and its allies, 
1 E. g. Hallier, H. : Uber die V er wandtschaftsverhaltnisse bei Engler’ s Rosalen, Parietalen, 
Myrtifloren und in anderen Ordnungen der Dikotylen. Abhandl. aus dem Gebiete der Naturwissen- 
schaften, Bd. xviii, Hamburg, 1903 (see footnote to p. 58 of reprint) ; and Eichinger, A. : 1 . c., p. 300. 
2 Engler, A. : Monographic der Gattung Saxifraga , L., Breslau, 1872, p. 12. 
3 Thouvenin, M. : Recherches sur la structure des Saxifragacees. Annales des Sci. Nat., ser. 7, 
t. xii, 1890, p. 33. 
