CONTRIBUTIONS TO BOTANY. 
15 
vii. 34) : this has since been so ably discussed by the authors of 
the ‘ Flora Indica’ (p. 170) that it is unnecessary to go over the 
same ground, as I concur in most of their views on the subject. 
The* relationship towards Lardizabalacece, Magnoliacea, and 
AnonacecB, as is there shown, cannot be doubted ; but this is 
not so considered in the new ‘ Genera Plantarum,’ where the 
apocarpous order Lardizabalacece is transferred as a mere tribe 
into the monocarpous family of Berberidacece, and that of the 
Canellacece (intimately delated to this last family) is carried far 
away and placed between Violacece and Bixacece : this appears 
to be a very illogical view of their true relationship. The apo- 
carpous Thalamiflorce, with parietal placentation, constitute so 
natural a group, and are connected together by so many similar 
characters, that it is difficult to conceive why any of them should 
be placed elsewhere ; and, in regard to Canellacece, the fact of 
having two or four lines of parietal placentation, as in some 
Lardizabalacece, the resemblance of the ovary and seed to those 
of Drimys (especially in the shape and position of their small 
embryo), their many-seried imbricated sepals and petals (as in 
Magnoliacece), their extrorse anthers (as in Anonacece), the ex- 
trorse monadelphous stamens (as in many Menispermacece], their 
solitary carpel (as in Berberidacece), and the resemblance to the 
whole of these orders in their mode of placentation, are charac- 
ters extremely manifest *. The weight of this evidence leads to 
the conclusion and confirms the opinion that the Canellacece 
should rest in contiguity with Berberidacece, osculating at the 
same time with the above-mentioned apocarpous group, and not 
with Bixacece or Violacece, with which they have little analogy. 
If these eminent authorities had classed the Canellacece where 
they have placed the Lardizabalacece (before Berberidacece), and 
had retained the Lardizabalacece in their former position among 
the climbing polycarpous families, near Menispermacece, such an 
arrangement would have met with the general accord of botanists ; 
and this it is to be hoped they will be induced to do in a second 
edition of their important work. 
* Having lately defined the characters of the Canellacece (Contrib. Bot. 
1 . 1 12, pi. 23, 24), I cannot be considered presumptuous in venturing to 
diflFer in opinion from the above-mentioned eminent botanists, who acknow- 
ledge the resemblance of the seeds to those of Winteracece, but who object 
that the Canellacece differ widely in the structure of the perianth, stamens, 
and ovary. This can hardly be conceded ; for if we compare the sepals 
and petals of Cinnamodendron (pi. 24) with those of Drimys (pi. 26) or of 
Illicium, they will be found to accord in a remarkable manner ; and if we 
conceive the extrorse stamens of Drimys united by their margins, we have 
precisely the staminal tube of Canella ; in hke manner, by joining the five 
carpels of Drimys by their margins into one, we have exactly the ovary of 
Cinnamodendron with its sessile stigmata and five lines of placentation, as 
shown in pi. 24. 
