122 
CONTRIBUTIONS TO BOTANY, 
seed of the Canellacece, we may with some confidence venture to 
assign the place which this small group should occupy in the 
system. Its structure is so palpably opposed to that of the 
Clusiacea (where Canella has generally been arranged by bota- 
nists), that it is quite unnecessary to enter into any discussion 
upon the value of such an affinity. This incompatibility was 
long ago shown by Prof. Lindley*, as before stated, upon very 
substantial grounds ; he also proved that it could not be asso- 
ciated with Platonia, and that Gaertner was incorrect in his 
description of the fruit of Canella. In this uncertainty, he 
looked to the Pittosporacece as a more probable affinity. There 
are certainly several points of similitude between them, but the 
relationship appears to me very distant : the symmetrical number 
of the parts in the Pittosporacem, their petals united at the base 
into a short tube, their perfectly free stamens with introrse an- 
thers opening by pores in the apex, the deep inflexion and union 
of the carpels in the centre, where they are placentiferous, — all 
combine to prove that these two families are far from being akin. 
On the other hand, there exists, as I have already shown, a 
singular degree of accordance in the general habit of the Canel- 
lacece with Drimys : the same aromatic principle pervades their 
bark, leaves, and flowers ; they have both similar alternate ex- 
stipulate leaves, furnished with transparent dots, and they have 
unsymmetrical biserial petals, with an imbricated aestivation. 
There is also a no less striking analogy between Cinnamodendron 
and Drimys, as well as Illicium, their ovary being unilocular, 
with longitudinal parietal placentation ; and there is a remark- 
able parallel in the form and structure of the seed. There can 
therefore be no doubt that a very close affinity exists between 
these two groups. The Canellacea, however, will be found to 
differ from the Winteracece in their monadelphous stamens, and 
more particularly in their single ovary. In regard to their 
relative position in the system, if we follow the basis of the 
Jussieuan method, adopted by DeCandolle and most botanists, 
and carried out by Endlicher in his ‘ Genera Plantarum,' we 
cannot fail to arrive at this conclusion, that the several distinct 
ovaries, each formed of a single carpel united by its margins 
without any inflexion, which margins being on the ventral side 
and. ovuligerous, form there a longitudinal parietal placentation, 
— characters that belong to the Winteracece, — unquestionably 
place that family in the class Polycarpicce. On the other hand, 
we find in the Canellacece similar carpels ; but instead of being 
distinct, they are united into one compound ovary by the simple 
junction of their placentiferous margins, thus forming a uni- 
locular ovarium with compound parietal placentation : this Order 
* Veg. Kingd. 442. 
