26 
CONTRIBUTIONS TO BOTANY. 
indusia, as constantly witnessed in the two former families, and 
frequently in the latter, and in the deeply concave hilum, formed 
by the increment of the seed around the placenta, which is drawn 
into its cavity, and the consequently somewhat arcuate direction 
of the embryo udthin the albumen, seen more especially in the 
tribe Heteroclinea among the latter family. There are other 
considerations to be held in view, that the PrimulacecE, Myrsi- 
nacecR, and Theophrastacece, offer a free central placenta within the 
ovarium, without any appearance of parietal septa, or any con- 
nexion of the placenta with the style : we see also in the Illici- 
hracea, Mesembryanthacece, and Poi-tulacacece, a somewhat ana- 
logous development ; but in these cases we cannot imagine this 
to be the result of the rolling up of the placentaiy margins of one 
or more carpellary leaves, according to the hypothesis generally 
entertained ; but we may rather conceive, that tlie margins of the 
carpellary leaves constituting the ovarium have not the power of 
developing ovuliferous placentae, a power seemingly there con- 
fined to the rudimentary petiolar support or gynophorus, which 
throws out its placentary threads, that are free in Portulacacea, 
&c., but co'Q.^\xent 'va.Primulace(je,Myrsinace(B,Theophrastacea, &c. 
This view is confirmed by the appearance of the lengthened 
thread that grows up from the torus with the elongation of its 
seed, and its placentary attachment, in the instance of JEgiceras. 
We may therefore look upon this mode of development as the 
opposite extreme of the case of the multilocular ovarium, where 
its intrafolded placentations unite in a central axis ; and we may 
look upon the Olacacece, Styracece, &c., as forming an interme- 
diate state of development. Under such an hypothesis, keeping 
in \iew the considerations before mentioned, it would tend to 
a more natural division of the system, to remove all the several 
orders, from the Lentibularice to the Styraceee, from the position 
assigned to them in the arrangement of the ‘ Prodromus.’ Yet 
because the development of the ovaria in these instances may be 
traced to somewhat similar causes, it does not necessarily follow 
that they must all be allied together, for other considerations of 
equal moment may tend to keep them far apart. Thus from 
circumstances before enumerated, the Styracece and Myrsimcece 
might be associated with the Olacacece and Santalacece, between 
Berberidacece and Rhceades, in a grou]) that might be called Cio- 
nospcrmce, as I suggested on a former occasion [hnj. op. vol. vii. 
p. 207), and in this group the anomalous genus Aptandra will 
naturally find its place. On the other hand, the Sopotacece with 
their truly axile placentation, the complete cells of their ovarium, 
and their corolla more pleiopetalous than monopetalous, appear 
more naturally allied to the Aquifoliacece^ in which family the 
])etals are also generally combined at the base into a tube. The 
