CONTRIBUTIONS TO BOTANY. 
131 
a solitary flower ; but in D. Chilensis the peduncles are as nu- 
merous, are longer, broadly compressed, and always bear on tbeir 
summit six or seven umbellated pedicels, each supporting a 
flower, and these are surrounded at their base by a verticil of 
deciduous bracteoles ; where there are seven pedicels, this invo- 
lucre consists of six bracteoles ; and with six pedicels it has flve 
bracteoles. In D. Granatensis this umhel is formed of only three 
to flve pedicellated flowers, and the bracteoles are far more de- 
ciduous. In D. Winteri the calyx is larger, and the petals shorter 
and broader in proportion, seldom exceeding six in number; 
while in D. Chilensis I have always found ten petals : the former 
has nsually four ovaries, the latter generally five. These differ- 
ences, added to their general aspect, and their distinct geogra- 
phical distribution, appear to me sufficient to establish the 
validity of these several species nearly as they are described hy 
DeCandolle and other botanists. 
The plant from Juan Fernandez I consider to be specifically 
distinct; and I find another, from extratropical Brazil, with 
smaller and very narrow, almost linear leaves, and terminal 
flowers on simple peduncles, which is certainly different from 
D. Brasiliensis ; and likewise a third, of remarkable aspect, col- 
lected by Claussen in the province of Minas Geraes, to which 
Dr. Hooker* alludes, as being a singular state of D. Granatensis ; 
but it ill-accords with that species, on account of its inflores- 
cence, and its smaller and excessively revolute leaves. I have 
referred to Tasmannia, for reasons that will be given, the species 
from the island of Borneo, D. piperita, placed in this genus by 
Dr. Hooker. 
I may here remark, that some of the species (more especially 
T). Granatensis) have a disposition to produce what gardeners 
call double flowers; in the variety Mexicana of that species 
there are usually twelve petals, which are sometimes increased 
to twenty-four in number f. 
The anthers in Drimys are generally described as being dor- 
sally attached to the broad filament, but this is not the case ; the 
two cells are quite separate, and affixed by a median line to the 
margins of the filament, which margins are reflected outwax’d, 
so that the anther-eells are thus approximated, and thrown into 
an exti’orse position. In farther proof of this, I may here men- 
tion that I have sometimes met with a stamen half-transformed 
into a petal, in which case the anther-cells are widely separated, 
and attached to each margin of the petal, a little below the 
middle. 
I propose dividing the species into four sections, distinguished 
by the simple or umhellate, either terminal or axillary flowers, 
* Flor. Antarct. ii. 230. t DeC. Syst. i. 444. 
s 2 
