Candolle was misled by imperfect material, and this theory was adopted and elaborated in the same year by J. J. Bennett, when dealing with a tree found by Horsfield in the "Banyumar Province" of Java, on which Bennett's genus Saccopetalum was founded (Plaht, Jav. Ear. p. 165 t. f xxxv : 1840).
Bennett remarked that since the publication of the family in De Candolle's Prodromus two new genera of Anonacere had been constituted, viz., Milium and Hyalostema. With "Hyalostemma we need not concern ourselves further than to say that it was coined by Wallich without any diagnosis, for a plant which had been duly described by Eoxburgh (Fl. Ind. ii. 660) as Uvaria dioeca. Bennett's note adds nothing to the history of " Hyalostemma " (which has subsequently been dropped by common consent), and his discussion of De Candolle's plant does not advance the history of Miliusa, because he postulates the identity of Wight's species with "Miliusa Leschenaultii", which, as we have seen, can only be accepted by making alternative assumptions, in support of which no proof has been put forward.
As it happens, a good deal of the reasoning expended on the matter is invalidated by facts since discovered. Speaking of that group of Anonaceae, of which Miliusa has been taken as the type, Bennett writes:—
"The stamina also are subject to some modifications, less extensive however than the character and description of Miliusa given by M. Alph. De Candolle would lead us to believe. Their number in that genus, according to my observation, is about 27 instead of 12, forming three alternating series, in each of which two are opposed to each of the inner, and one to each of the outer petals ". Now, in Miliusa nilagirica Beddome, Ic. PI. Ind. Or. t. lxxxviii (1876) there are but eight stamens : that this is a Miliusa no one probably will dispute, and in that case Bennett's estimate of the stamens as " about 27 " obviously calls for revision. It seems even possible that this plant, and not either of Wight's Miliusae, was the true M. Leschenaultii* The glabrate middle surface and margin thickened towards the apex of the petals, which are given as distinctive of M. nilagirica are approached in some examples of ill. montana, Gardner, and these differences are not sufficient, perhaps, to mark off Beddome's species from that of De Candolle. Beddome's illustration Ixxxv serves to contrast M. a finis, Wight MSS., — which is there figured as " M. indica", — from M. Leschenaultii. The number of the stamens in the last differs, as observed, from that M. nilagirica, but it may be questioned whether the number is so fixed in this group as was supposed by Bennett. As regards the structure of the valvate petals he has very justly noted that the gamopetaly of the inner row is less
