CONTRIBUTIONS TO BOTANY. 
53 
bably be more readily understood from the examination of the 
ovary. This is here compressed, as in the former tribe, 2-celled, 
with a precisely similar style and stigma; but the dissepiment is 
not, in like manner, parallel to its two faces, but, contrariwise, 
transverse, its ovuligerous margins corresponding with the antical 
and postical longitudinal ridges ; the cells, therefore, instead of 
being anterior and posterior in regard to the axis, stand on its 
right and left, so that the stigmata are contrary to the cells : 
the ovules, in two series, are borne upon the dissepiment (in 
Tecoma ochracea) (fig. 6), not along the axis, but, as in the 
Fig. 6. Fig. 7. Fig. 8. 
former instance, at the utmost distance from it. Now, in this 
case the ovary may be assumed to be constituted either of two 
carpels, ovuligerous on their midribs, and placed back to back, 
as in fig. 7 ; or it may be formed of four carpels arranged some- 
what as in the Bignoniea, but differently disposed, as in fig. 8, 
the sterile margins being afterwards united, and the adjacent 
faces becoming confluent. Upon comparing these figures, it 
will be seen that the former view must be rejected, because, 
under that hypothesis, the ovules ought to be found arranged 
along the axis ; and the latter view must be adopted, because it 
satisfactorily accounts for the position of the ovules in the ovary 
and of the seeds in the fruit. Owing to a somewhat different 
inclination of the carpels, as shown in fig. 9 (which will be seen 
Fig. 9. Fig. 10. 
to be a modification of fig. 8), the ovary and fruit become 4- 
locular in Heterophragma ; and the resulting form of fruit is 
shown in fig. 10, where the dissepiment is cruciform, with two 
longer and two shorter arms ; the latter, bearing the seeds 3- or 
4-serially on both sides, terminate in the middle of the valves, 
the former touching the sutural and dehiscent margins of the 
valves. When the valves open, the corresponding barren portion 
of the dissepiment becomes split across into two longitudinal 
halves, each half bearing the seminigerous portion of the dis- 
sepiment in the form of the letter T, which at the same time 
detaches itself from the middle of the valve. We see here the 
most complete verification of the hypothesis above suggested of 
