30 



THE CERATOPIIYLLINAL DIVISION. 



LlNACEiE. 



The affinities of this family are so fully explained in Bentham and Hooker's 6 Genera Plantarum,' that any further comment at present is scarcely required, but it may be interesting to notice 

 that the genus Radiola, which is the farthest removed from Ternstrdmiacece, shows an agreement with Pentaphylax in its filament dilated in the lower half. The stamens of Pentaphylax are 

 figured in the Trans. Linn. Soc. vol. XXI., PL 12, and those I have seen are more like those of Radiola than there represented. The stamens of Radiola are not monadelphous, but interpetalous, 

 and with the claws of the petals form a very short tube.* Erythroxylece have been placed with families having a dorsal raphe, and Prof. Agardh (Theor. Syst. Plant, p. 296), thinks it probable that 

 the raphe is dorsal; it is however so indistinct externally that no very decided opinion could perhaps be given, but the foramen which is external to the funiculus curves slightly inwards so as to-be 

 above it, and in one species it is directly on the outside of the funiculus (i.e., not inclining to either side) covered by a small process of the placenta or funiculus to which it is partially adherent, so 

 that the raphe must be next the placenta, and in fact a transverse section of the ovule, apart from the position of the foramen, leaves scarcely a doubt of it. 



DlPTEROCARPEiE. 



The ovary of Ancistrocladus I believe to consist of two or three carpels united by their margins, their number being shown by the styles. The parts which Mr. Thwaites regards as stigmas, 

 I believe to be styles, and the apex only of each to be the stigmas ; and the part he describes as the subglobose persistent style I believe to be the apex of the ovary incompletely covered by the 

 calyx. The flower of Lophira agrees well with Clusiacece, scarcely differing from Calophyllum, as the ovary of that genus may, I believe, be safely regarded as compound with the carpels united 

 by their margins, having one ovule erect from the base, that of Lophira consisting of two carpels, and having eight or ten, which are in both cases anatropal, and it thus assists in bringing to 

 light the nearest affinities of Dipterocarpete. 



In Ancistrocladus, as in Calophyllum, the raphe is variable in its relation to the axis or rachis from which the flower springs, and therefore, whether the ovary consists of two or three carpels, 

 the fertile carpel will be variable in its position in accordance with the raphe. 



Chl^enace^e. 



In the sepals being fewer than the petals, this family may perhaps be compared with Clusiacece and Papaveracece ; the leaves of Sarcolcena grandiflora are oval, with veins like those of 

 Endogens, agreeing with some genera of Menispermacece. 



* There are no intermediate barren stamens, from which it might be supposed that the barren stamens in Linum might be the consequence of the junction of the angles of the broad filaments ; this, however, is certainly not the case. 



