l(i 
PRELIMINARY TREATISE. 
Bulbocodium and Colchicum, distinguished by a tripartite 
and a tripartible style, shew that the actual separation of the 
three styles, which are usually either conjoined or consoli- 
dated, does not furnish a substantial feature to distinguish a 
natural order; and in Dr. Brown’s own character of Melan- 
thacese we find style trifid or tripartite. Posterior anthers 
would have afforded a safe character for that order or rather 
suborder, but Dr. Brown has included plants with anterior 
anthers, which makes the given character nugatory, there 
being no other feature really decisive. With respect to 
difference in the integuments of the seeds, the striking 
anomaly in that feature between the most closely united 
genera of Amaryllideae (as for instance Pancratium and 
Hymenocallis, Lencojum and Galanthus) shews that it is of 
no certain weight in distinguishing natural orders of plants. 
The beaked umbilicus, by which Dr. Brown characterises 
Hypoxideae, is very dissimilar in Hypoxis and Curculigo, and 
I understand that it is entirely wanting in that portion of 
Curculigo which must be detached under the name Moli- 
neria. 
I cannot too strongly disclaim any intention to depreciate 
the valuable labours of a gentleman whose life has been 
devoted to the cultivation of a science, which to l^een 
only the amusement of some leisure hours, and . . whose 
botanical knowledge my own is comparatively light as air. 
He is one of the most conspicuous ornaments of the study 
he has pursued; but in the arrangement of these orders he 
appears to have supposed the features, which he observed in 
some genera, to be more prevalent than they prove to be; 
and I have no doubt that he is fully aware that there are 
imperfections in the characters framed many years ago, 
which more intimate acquaintance with the subject matter 
makes it necessary to remove, and I am far from flattering 
myself that I can perfectly accomplish their reformation. 
We should constantly bear in mind that there can be but 
one real division, that is identity of kind, or, in other words, 
generic distinction; the secret bar by which the Almighty 
has made His works incapable of being blended with each 
other, and confounded in their propagation. All divisions, 
except that of generic identity are artificial, and rest on the 
supposed agreement of different individuals in one or more 
essential points of structure; and when any such point is 
assumed for classification, the question must arise whether 
