CONTRIBUTIONS TO BOTANY. 
51 
in mind that the normal carpellary leaves would be ovuligerous 
either on their margins or on their midribs : if on their margins, 
supposing them to be only two in number, how are we to account 
foT the production of a thick dissepiment, which could have 
formed no part of such original carpels ? if on their midribs, 
the same objection supervenes, unless we imagine that the mid- 
rib of each carpel grew out in the form of a septum, both of 
which, uniting by their edges, might constitute the dissepiment j 
but such kind of growth is contrary to the usual law of develop- 
ment, and, upon examination of the ovary in its earliest stages, 
there is no appearance of its having taken place. Such a con- 
jecture may therefore be dismissed, because it is quite unsup- 
ported by rule or evidence. 
We may, however, satisfactorily explain the morphological 
changes that have taken place in Eubignoniea upon the hypo- 
thesis that the ovary is normally composed of four carpels with 
sterile margins, and which are ovuligerous on their midribs, all 
severally plicated and arranged in opposite pairs, as in fig. 1 : 
Fig. 1. Fig. 2. Fig. 3. Fig. 4. 
these, by the confluence of their sterile margins and adjacent 
faces, would constitute a bilocular ovary, such as we find it 
in Bignonia and its congeners, with the ovules fixed, not in 
the axis, but in the marginal angles of the two cells (fig. 2). 
We may in this manner account for the separation of the two 
valves of the capsule from the dissepiment close to the four 
lines of placentation, and also for the production of the replum 
before described, which originates in the union of the midribs 
of the normal carpels (fig. 3), that sometimes remain aggluti- 
nated together (as in Pithecoctenium), sometimes become divided 
in two (as in many other genera). This view is further confirmed 
by the fact observed in Peltospermum, where the dissepi- 
ment is fenestrated at its apex for one-third of its length, and 
nearly its whole breadth (as in fig. 4), by a very large aper- 
ture, which no doubt is caused by the incomplete junction of the 
edges of the normal carpels where they meet together to form 
the septum, in the manner shown in fig. 1. This fact presents 
the strongest argument in favour of the view here taken. In 
Tanaecium, Distictis, and probably in Dolichandra, where, as in 
Bignonia, the dehiscence is marginicidal, the valves are partly 
split down their middle, thus partaking of both kinds of dehis- 
cence. I do not see how we can explain the development that 
takes place in the Eubignoniets under any other hypothesis. W'e 
