APPENDIX. 167 
fruit, which is most generally 2-celled, with many seeds fixed to 
thickened placentze adnate to the dissepiment, and having a 
terete embryo, more or less curved, with an inferior radicle, cha- 
racters that are common to the whole of this large group. So 
gradual is the transition from one link to another of this chain, 
that it is difficult to discover any decided break in their conti. 
nuity, but notwithstanding this, they form too large an assem- 
blage to constitute one single family. The Solanaceae, as distin- 
guished from the Scrophulariacee in general, exhibit characters 
sufficiently marked, but the difficulty lies with the large interme- 
diate group above indicated, that equally partake of the features 
of both these extremes. I am quite averse to the practice of 
multiplying unnecessarily the amount of natural orders beyond 
the smallest possible number: it is not therefore any idle no- 
tion of proposing a new family that leads now to this sugges- 
tion, which would defeat its own object unless supported by 
facts, and urged by the necessity of the case; but it is the desire 
of grappling with a formidable obstacle, that would otherwise 
prevent us from establishing any decided limits between these 
two great families. If this difficulty presented itself to me in so 
prominent a degree three years ago (Lond. Journ. Bot. v. 188, 
note), when I first noticed the anomaly in Lycium, and suggested 
its separation from Solanacee on that account, with how much 
more force must this discrepancy present itself, when the ex- 
ceptionable cases now amount to so extensive an accumulation 
in point of number! The zstivation of the corolla has hitherto 
been considered to form an unerring line of demarcation between 
the Solanacee and Scrophulariacee, but if we place in the former 
family a large proportion of genera possessing an imbricate zsti- 
vation, and offering frequently — anisomerous flowers (cha- 
racters peculiar to the last-mentioned order), we lose at once the 
only valid features that can serve to discriminate the boundaries 
of these great families. It is clear that the intermediate group 
here proposed to be collected together can only be disposed of 
in three modes: they must be associated either with the So/a- 
nacee, or be attached to the Scrophulariacee, or else they must 
remain as a distinct family. In the first case, the Solanacee would 
be then divided into two suborders : 1. the Solaninee, having a 
corolla with valvate estivation ; and 2. Atropinee, with imbricate 
wstivation. In the second case we should associate, 1; Atropinee, 
with flowers nearly isomerous ; and 2. Scrophularinee, with ani- 
somerous flowers. In either of these two cases we find that in- 
consistency to a great extent would be unavoidable ; for in the 
former instance we admit a large circle of exceptions to the only 
leading characteristic mark of the order ; and in the second case 
we include a considerable number of genera, nearly = ina 
