OP PLANTS CALLED COMPOSITE. 267
nearly related to Compositae. It exists also in Ernodea, rsa in which the lateral nerves, though they give out externally a few branches, observe the same course, and terminate in the same manner in the laciniae as those of Compositae. A similar disposition is observable in certain genera of Solanaceae, as Datura and Oestrum, though in these the lateral nerves are more ramified, and their trunks generally less distinct in the laciniae. It appears therefore that, in adopting M. Cassini's theoretical expression for the vas- cular structure of the corolla of Compositae, one pecu- [86 liarity actually existing is lost. 1
adhering part dilated into nectariferous processes ; or in those where, the seg- ments of the calyx not being closely approximated, the coloured corolla is visible in the interstices. In some species of Goodenia, particularly G. decurrens and bellidifolia, I find it practicable to separate not only the adhering calyx, but also the tube of the corolla from the ovarium. In the tube thus separated it appears that the lateral nerves, which preserve their parallelism to the middle nerve nearly to the base of the segment, become more evidently divergent below the point of adhesion, and in such a degree that the corresponding branches of the neighbouring segments unite with each other considerably above the middle of the tube, forming a common trunk, which is continued to the base of the ovarium; the five trunks thus formed uuiting internally with those from which the filaments originate, and externally with the axes of the opposite seg- ments of the calyx. The middle nerves of the segments of the corolla are in like manner continued below the point of cohesion to the real base of the tube.
The analogy of this disposition of vessels in the corolla of Goodenovie to that of Compositae is obvious. To assimilate entirely the two structures, it is only necessary to suppose a deeper division of the five primary vessels of Com- posite, and a continuation of the tube of the corolla below its apparent base to that of the ovarium. That this is its real origin, is rendered not improbable both from the analogous structure now described in the family of Goodenovice, and from the manifestly hypogynous corolla of Brunonia ; a genus in many respects still more nearly related to Composite, though differing in the dispo- sition of the vessels of its corolla.
The more direct proof of this origin, derived from an examination of the sur- face itself, can hardly, perhaps, be expected where the parts are generally so small, and where, as I conceive, the surface of the pericarpium in many cases depends less on that of the cohering envelopes, than on the proper figure of the ovarium itself, as seems to be likewise the case in Umbellate.
There are however a few cases in which this opinion respecting the origin of corolla in Composite may derive some additional support from the appear- ance of the surface of the ovarium, as in Marshallia and Hymenopappus, in both of which genera, but particularly in the former, it is marked with ten longitudinal strie, of which the five stronger are continued into the five nerves of the corolla, the remaining five ending abruptly at the apex of the ovarium.
1 A still stronger objection to M. Cassini's definition is, that while its appli- cation to Composite is only hypothetical, it very nearly corresponds with the
�� �
