112 
CONTKIBUTIONS TO BOTANY, 
proper place of that family in the system is near the Aquifoliacece, 
the structure in both cases being nearly similar, differing prin- 
cipally in the aestivation of the corolla, which is valvate in the 
former and imbricate in the latter. In both families the ovarium 
is normally plurilocular ; and when, as in Villaresia, it is one- 
celled, this is always due to the abortion of the other cells — 
a condition that also prevails throughout the Icacinacece •, it 
therefore very rarely happens that more than one cell and a 
single seed are perfected in this genus. I have recorded the 
fact * that in Pennantin, where the fruit is generally unilocular 
and monospermous, it sometimes occurs that two cells, each 
1 -seeded, are developed. I have witnessed the same excep- 
tional occurrence in Villaresia, in a species growing at Kew, 
where, on the same plant, some of the ovaries were 1 -locular, 
while others were distinctly bilocular, each with two ovules col- 
laterally suspended from the dissepiment. Hence the rule is 
general throughout both families, that, although the normal 
condition of the ovary is plurilocular, there exists a general ten- 
dency to the abortion of most of its cells. Thus in Ilex we have 
constantly four, or more rarely five, cells, in Prinos six to eight, 
in Cassine three, in Nemopanthes three or four, while in Byronia 
they number from twelve to sixteen j in Villaresia it is almost 
universally 1 -celled, as also is the case in JExtoxicum, which 
appears to belong to the same family. Although, as I have 
mentioned, the ovary in the Icacinacece is generally unilocular, 
it is constantly 3- or 5-celled in Emmotum. 
Throughout both families there is a general tendency towards 
the suppression of one of the sexes, so that the flowers are always 
more or less polygamous ; and this is carried to such an extreme 
in jExtoxicum, that they are constantly dioecious, when in all 
other respects the structure corresponds with the Aquifoliacece. 
On the other hand, perhaps no genus in the family approaches 
nearer to regular hermaphroditism than Villaresia. 
One great peculiarity attends the development of the ovary in 
this genus ; the suppressed cells united in the normal axis form 
a prominent longitudinal parietal expansion, which extends far 
towards the centre of the single fertile cell ; and from near the 
summit of this expansion the two collateral ovules are suspended. 
In the fruit, only one of these ovules arrives at maturity, and 
that soon fills the entire cavity ; the seed therefore moulds itself 
about the placental expansion, becoming thus bent round it, so 
that its transverse section is hippocrepiform. A similar struc- 
ture occurs in Bursinopetalum, as is well shown by Dr. Wight, 
in his ' leones^ (tab. 956). This latter genus is referred by Dr. 
* Huj. op. vol. i. p. 77, pi- 12. figs. 2-5-28 j Ann. Nat. Hist. ser. 2. 
vol. ix. p. 488. 
