caruncula, a mistake involving a second, that of considering the seeds erect in the capsule, and which has led Labillardière into a third error, namely, describing the radicule of the embryo as pointing towards this supposed umbilical appendix.
545] The Tremandreæ are in several respects nearly related to Polygaleæ; they appear to me, however, sufficiently distinct, not only in the regularity of the flower, and in the structure of antheræ, but in the æstivation of both calyx and corolla, in the appendix of the seed being situated at its apex, and not at the umbilicus, and, I may also add, in a tendency to produce an indefinite number of ovula in each cell of the ovarium.
The greater number of Tremandreæ are found in the principal parallel of New Holland, they extend also to the south end of Van Diemen's Island, but none have been observed wdthin the tropic.
DIOSMEÆ. To this natural order, in addition to the Australian genera hereafter to be mentioned, and the south African genus from which its name is derived, I refer Fagara, Zanthoxylon, Melicope, Jambolifera, Euodia, Pilocarpus, Empleurum, and Dictamnus: and four genera of equinoctial America, namely, Cusparia of Humboldt and Bonpland, Ticorea and Galipea of Aublet, and Monnieria, if not absolutely of this order, belong at least to the same natural class.
Both Ruta and Peganum may be annexed to Diosmeæ, but neither of them are calculated to give a clear idea of the order, from the usual structure and habit of which they deviate in some important points; I have therefore proposed to derive the name of the family from one of its most extensive and best known genera. The first section of Jussieu's Rutaceæ is sufficiently different to admit of its being considered a distinct order, which may be named Zygophylleæ.
Diosmeae are numerous in Terra Australis, and form, at least in its principal parallel and more southern regions, a striking feature in the vegetation. Nearly seventy species
