24 



THE CERATOPHYLLINAL DIVISION. 



C/OMBRETACEiE. 



This family differs so much from Myrtacece and their allies, that if retained in the Epigynous Division they would show anomalies in structure which, I believe, could not be accounted for. 

 On the other hand they are very distinctly connected with Lauracece through llligeracece, by the genus Hernandia afterwards noticed ; and the wood of some species of Combretacece is referred to by 

 Prof. Oliver as presenting that kind of irregular structure which occurs in Malpighiacecc and Bauhinice, yiz., of being partitioned into numerous vascular masses by intruded plates of cellular 

 tissue. (Trans. Linn. Soc. vol. XXII. p. 290.) They differ from Myrtacece in their winged fruit, the wings being produced by the tube of the calyx in the same way as those of Margyricarpus 

 among Sanguisorbacece, not s by the ovary, as in Begoniacece and Umbelliferce ; — in their style, which is furrowed like that of Amygdalece, and in some instances of Lauracece, which does not 

 occur in Myrtacece, although the ovary may be reduced to a single carpel; — in their fruit, which is occasionally umbilicate, a character which occurs in Lauracece ;— and in the position of the 

 carpel which is irregular, as in Illigeracece. In their placentation, also, they differ from the epigynous families. In Combretum (R. Spruce, No. 1523*), the ovules are 10 or 12, in part pendulous 

 from the apex of the ovary, but the greater part of them are attached to the sides, suspended in a ring round the apex. Supposing the ovules attached round the orifice of an inverted funnel 

 would give a correct idea of their placentation. The style is quite pervious in its lower half, and the canal opens into the centre of the ring in which the ovules are attached, so that the placentation 

 of nearly all of them must consequently be regarded as dorsal. The uppermost ovules are attached so close to the stigmatic canal, that their funiculi extend into it. Where the ovules 

 are two, they are attached to opposite sides (or nearly so) of the apex of the ovary and stigmatic canal, and the raphe in most instances is dorsal, so that the raphes of the two ovules are 

 sometimes in apposition with each other, but the funiculi being long it is sometimes possibly from twisting lateral, or more rarely next the wall of the ovary ; but whether the placentation in these 

 instances is dorsal remains a question, as the ovary may be dicarpous. In the Combretum above referred to, the raphe in most of the ovules is also dorsal. The position of the carpel is inferred 

 from the furrow in the style, which is occasionally deepened into a fissure, not from the placentation. 



In their spirally convolute cotyledons they have been referred to as agreeing with the § Memecylece in Melastomacece, but the convolution of the cotyledons of Memecylon is of a very different 

 character. In Memecylon the embryo is curved in the direction of the flat surfaces of the cotyledons from apex to base, like those of Lepidium and Heliophila, and the cotyledons are curved 

 probably to a less extent than those of the latter genus, being simply incumbent in their relation to the radicle, and not in any degree spirally convolute. The embryo is curved in the same way in 

 other genera of Melastomacece, but to a less extent, the curvature in the cotyledons being from apex to base, whereas in Combretacece and Calycanthacece it is from side to side ; — in Granaiece 

 and V ochysiacece, however, it is also from side to side, so that this character throws no light on affinities in the present instance. The peculiarity of the adhesion of the style to the long tube of the 

 calyx, with which it is confluent for more than three-fourths of its extent, occurring in the genus Quisqualis, I regard asanalogous to the adhesion of the ovary to one side of the tubular calyx in 

 Chrysobalanece, and in Bauhinia and Jonesia, among Leguminosce, and as therefore tending to show that Combretacece do not belong to the Epigynous Division. In Jonesia the tube of the calyx is 

 much elongated, the stamens arise from its orifice and the stipes of the ovary adheres to the tube up to the orifice.f 



The 6-furrowed embryo of Sparattanthelium and the 3-furrowed cotyledons of Hernandia may show an analogy to the 8 or 10 plates of the albumen of the nutmeg (v. p. 23), and admitting 

 the very near affinity of Illigeracece to Combretacece, this would supply another argument for placing the latter with the perigynous and hypogynous families. 



Illigeracece. 



This family, originally described by Dr. Blume, was subsequently admitted into systematic arrangements as distinct from Lauracece, but the genera have more recently become regarded as 

 only apetalous forms of Combretacece. The genus Hernandia agrees in all important characters with Illigera and Gyrocarpus, and I would refer it, and without doubt, to the Illegerece of Blume, as 

 constituting with them and Sparattanthelium Mart* a very distinct section of Combretacece or a distinct family intermediate between them and Lauracece. It is the generally received opinion that 



* Coll. R. Spruce, Maio, 1851. Ad oram meridionalem Rio Negro, &c. 

 t In Chrysobalanem and Bauhinia the adhesion is to the anterior side of the tube, but in Jonesia I believe it to be posterior. 



% This genus is very near Gyrocarpus, but besides the stamens being alternate the lobes of the calyx, it differs in the embryo being deeply plaited in addition to being convolute, which gives it the appearance of being 6-lobed like that of 

 aernanaia; and, as tar as I have seen the species, it differs also in the fruit being wingless, the segments of the calyx not enlarging. A part of the species are described in the Botanisehe Zeitung, 1841, vol. II., p. 40. 



