BOTANICAL INFORMATION. 229 
Rostellaria) distinguished in the first place by the relative 
number and position of the stamens, the lobes of the corolla 
and the petaloid scales, characters in some cases artificial, 
but in so natural an order not the less useful, as being gene- 
raly clearly defined and easy to be seen. Even characters 
derived from the seed (for instance the abundance or absence 
of albumen which separates Sersalisia from Sideroxylon, Di- 
pholis from Bumelia) in most cases of great importance, would 
appear here to be artificial ; that is, unattended by any other 
perceptible generic difference, and consequently, where the 
seed is unknown, the genus is uncertain. Possibly when the 
Structure ofthe seed shall have been ascertained in a greater 
number of species, some corresponding differences may be 
detected in other respects; in the mean time, it is pro- 
ductive of no small practical inconvenience to divide a natu- 
ral genus into two by a character which has only been 
observed in a small proportion of the species. The esta- 
blishment of sections might have answered the same scien- 
tific purpose and obviated the evil of having so many species 
. with doubtful names. 
Of the eight genera of Ebenacee three are new, all well 
distinguished, and in these instances, as wellasin that of most 
newly established or defined genera in orders worked up by 
Alphonse de Candolle, besides the detailed character (some- 
times rather long), a few very useful words are added, point- 
ing out the most striking points in which they differ from 
other genera with which they are likely to be confounded. 
i os itself, consisting of seventy-three species, besides 
twenty-three less known, is divided somewhat artificially, but 
that was the only course to be pursued under our present im- 
Perfect acquaintance with many of the species. 
Styracacee, after weeding out many heterogenous plants 
which had at different times been associated with them, have 
become, under Prof, De Candolle, a small but natural and 
well defined order. The affinities with Ebenacee, Humiria- 
cee and Alangiee are well pointed out; the connexion with 
e@ is perhaps not so close, on account of the remark- 
