50 
CONTRIBUTIONS TO BOTANY. 
On the other hand, in some few genera, as in Jacaranda, 
Fridericia, Eccremocarpus, and Calampelis, there is no actual 
dissepiment, but the placentations are parietal along the middle 
of two parallel valves, which lines of placentation nearly touch one 
another, but do not meet in the axis : this linear placentation 
projects but little above the surface of the valves, and is retlected 
right and left into narrow plates parallel with the valve, which 
plates are crenately lobed, each lobe bearing a seed. In Jaca- 
randa the seeds are arranged parallel to the valves. In Eccre- 
mocarpus (I speak rather of Calampelis, which has sufficient 
claim to be generically distinct) the seeds are more imbricately 
pendent. In Fridericia the structure of the capsule and seeds 
completely agrees with that of Jacaranda. 
It has been held as a general rule that, throughout the order, 
the direction of the radicle is centrifugal in regard to the axis 
of the fruit : this is true in all the Eubignoniea, and in such ge- 
nera of the Catalpece as have a simple transverse dissepiment ; 
but many exceptional cases occur. The instances of Spathodea 
and Stereospermum have already been mentioned (p. 160), where 
the radicle is neither centrifugal nor centripetal, but lies in a 
transverse direction parallel to the diameter of the fruit : a 
similar direction exists in Sparattosperma and Heterophragma : 
in Jacaranda it has the same position, while in Calampelis it is 
almost centrifugal. On the other hand, in Platycarpum and 
Henriquezia the radicle is decidedly centripetal, as I have shown 
it to be in certain species of Spathodea. In Oocycladus its direc- 
tion is equally abnormal : here the ovary is bilocular, and the 
ovules, all heterotropal, are in a divaricating position, placed in 
collateral pairs down the middle of the dissepiment, with the 
micropyle directed to the axis. One of the cells becomes abor- 
tive, and in the other one only a single superior ovule becomes 
perfected, which occupies the entire space of the cell, and by the 
pressure of growth thus becomes pendent, so that the radicle is 
neither centrifugal nor centripetal, but points to the apex of the 
fruit. 
Having brought together these facts, we are better prepared 
to inquire into the construction of the carpellary arrangement 
in the Bignoniacea. We find here, generally, an ovary formed 
of two cells separated by a complete dissepiment, a simple style, 
and a stigma consisting of two dilated lamellae, corresponding 
most frequently in position with the cells. Upon this prima- 
facie evidence, the inference would naturally be that the normal 
number of its carpels must be two j but when we come to in- 
quire a little further, we find great difficulty in accounting for 
the mode of dehiscence of the capsule and the position of the 
seeds on the dissepiment, upon this hypothesis. We must bear 
