new genus Saccopetalum; but as is shewn in the Genera Plantarum (Vol. I, p. 28), the only distinction between Saccopetalum and Miliusa lies in the saccate petals of the former, whereas A. De Candolle in his description no less plainly than in the accompanying figure has given this very character as belonging to the only species of Miliusa known up to the time when Bennett wrote, because what is now Miliusa Roxburghiana, Hook. fil. & Thorn, was excluded and referred to the factitious Hyalostemma. In the Flora Indica, Vol. I, p. 92, the generic character was amended by restricting Miliusa to those species that have not more than two ovules, but as further species were examined this distinction seems to have been found untenable, though retained in a modified form in the Flora of British India, due warning being at the same time conveyed that Saccopetalum had been reduced to Miliusa by Baillon (Hist, des Planfces I 244).
Hook. fil. & Thorn, at the same time pointed out that Saccopetalum tomentosum is intermediate between Saccopetalum and Miliusa, i.e. presumably between Miliusa, affinis of Wight MSS. and the Javan Saccopetalum Horsfieldii of Bennett. Baillon (I.e.) and Prantl. (Nat. Pilanzenfam. Ill 2.29) do not even admit Saccopetalum to the rank of a subgenus. Comparing the number of the ovules in the different genera included in the "Genera Plantarum" under Miliusete we hardly find sufficient ground for maintaining Saccopetalum as an independent genus ; in Orophea, it is true, it is said that there are never more than four ovules, and in Alphonsea always more than four, with eight as a maximum; but in Bocagea, which is admittedly closely allied to Miliusa, they are given as 1—8. In the critical note now attached to a sheet from his own Herbarium, written up at Kew (by Sir J. D. Hooker probably) as ' Miliusa affinis, E. VvV, mentioned above—Wight has observed "On examining the ovaries of Miliusa somewhat advanced, I find them containing 2-3 and, once or twice, 4 ovules superposed; also probably there are two or three species distinguished by the stamens being few or numerous and the carpels glabrous or hairy". Whether therefore we assume that the number of maturing ovules differs in the species, or that it varies with the individual, little weight can be attached to this character in discriminating Miliusa from its allies. It appears, moreover, that in Saccopetalum the number or the ovules is not always six or more ; in the ' Flore Forestiere de la Cochinchine ' at plate 38 (1881) Pierre has figured and described as Miliusa Bailloni a tree so closely allied to Saccopetalum Horsfieldii, that the two could not properly be referred to different genera, but in this plant it is expressly noted that the ovules are not more than four in number.
In attempting to frame an amended definition of the genus Miliusa
