32 
CONTRIBUTIONS TO BOTANY. 
the axis of the upper portion of the albumen, the radicle being 
always superior, and the cotyledons veiy small and compressed, 
directed towards the centre of the nucleus. To such characters 
1 have found the following genera correspond, viz. Ximenia, 
Heisteria, Olax, Schopfia, Strombosia, Cathedra, lodina, Liriosma, 
Opilia, Arjoonu, Quinchamaliurn, and two new genera, Agonaridra 
and Endusa. The order thus restricted is marked by more 
distinct and coextensive characters than those proposed by Mr. 
Bentham, and will be seen to comprise only his tribes Olucacece 
and Opiliece. The latter tribe however cannot be maintained, 
as I find that Cansjera does not belong to the family*, and that 
Opilia, although often with only a single suspended ovule, some- 
* Tlie genus Cansjera, first placed in the Thymelece by Jussieu, was re- 
tained there by all subsequent botanists, till removed to the Olacaceee by 
Mr. Bentham, who concluded it was alhed to Opilia, because he considered 
it to have a small distinct adnate calyx, and an unilocular ovarium, with a 
single ovule susj)ended from the summit of a free central jfiacenta. All 
the specimens I have examined of both known species, fi’om various localities, 
and in (hfferent herbaria, present characters constantly at variance with 
these conclusions and more in accordance with the description given by 
Lamarck (Diet. iii. 433). Here I can observe no trace of any chstinct calyx, 
but the floral envelope, which is a simjfie tubular perianthium, is supported 
at base upon a small and pointed navicular bract : the four stamens are 
adnate in the upper portion of the tube, equal to the number of the lobes 
of the border, and opposite to them ; four tridentated, free, hypogynous 
scales alternate with the stamens ; the long conical ovarium is seated upon 
a narrow glandular support, from which the scales originate, and the style 
is surmounted by a large 4-lobed capitate stigma. The ovarium I find to 
be constantly 4-locular at base, and one or more (generally two or three) of 
these minute cells extend iiTegularly hke narrow and inteiTupted channels, 
to the u])per portion, and the fecundating threads may be traced from all 
of them, most distinctly, to the style : a single ovule is seen, sometimes 
higher, sometimes lower, from a prominent line of placentation on one side 
of each ovuliferous channel which at the jjoint of the development of the 
ovule becomes widened, and here the ])lacenta is somewhat curved, by 
the ascending direction of the ovule. The seed is a ilrupe, apiculated by 
the base of the style, and supported below bj" the remains of the shrivelled 
])erianthium ; it contains an oval coriaceous putamen, which encloses a 
single erect seed ; a short receptacle is seen at the base of the cell, w hich 
enters into a corresponding hollow in the seed, and fi'om it extend, in a 
cruciform dnection, four prominent keels or ridges, which penetrate as 
many furrows obseiwable in the albumen : the testa and integument are 
membranaceous, the albumen solid and fleshy, and an embryo of half its 
length is placed in the axis of the upper moiety : this embryo is slender, 
cylindrical, and terete, its sujrerior radicle is oval, clavate, six times shorter 
than the linear cotyledons, of which there were three, equal in size, in the 
sirecimen I examined : fi-om the extremity of the cotyledons a thread ex- 
tended to the umbilicus in the axis of the albumen, which was probably the 
remains of the embryonary sac. These characters cannot in any single 
respect be made to correspond with the Olacacea, and Cansjera must again 
be assigned to its former place, as an anomalous genus of the Thymeleacece, 
until a more fitting position can be given to it. 
