28 



THE CERATOPHYLLINAL DIVISION. 



from three to eight imperfect carpels (some of which have cavities but contain no ovules), in which it may be regarded as approaching Ranunculacece ; these surround the original ovary and are 

 adherent to it, some of them being as high as the ovary itself and others not half its height. The ovules are obliquely ascending, with the raphe lateral, as in Francoa ; and in its long loose testa 

 the seed closely approaches that of Pyrolacem ; and Droseracece and even Ericacece, to some extent, also agree in this character, the seed having in some species of Drosera a loose membranous testa 

 prolonged at the base and apex into acute membranous processes. 



Ceratophyllace^;. 



The affinities of Ceratophyllum appear to me, as originally suggested by Dr. Asa Gray, to be entirely with the Nymphal Alliance, except in the conical male flowers, which have in some 

 degree the appearance of those of Schizandracece (V. Ann. Nat. Hist., ser. 2, vol. xii., p. 12). Salvinia differs from other vascular Cryptogams in being almost rootless, thus approaching Ceratophyllum. 



Ericace^. 



The genus Diapensia is so near Azalea agreeing with A. procumbens in habit, in the number of stamens and cells of the ovary, that it appears to me that Diapensiacece are not distinct from 

 Ericacea, differing from them only in a more complete adhesion between the stamens and petals. The flower is pedunculate, with no bracts near the calyx, except two immediately under it, which 

 are so much like the sepals that they appear as if they formed a part of the calyx, but they are distinctly external to it. In this character it agrees well with Erica and Calluna y E. cinerea having 

 only two bracts which are equally close to the calyx and look like sepals. It is probable that the monopetalous corolla is formed for the most part by the adhesion of the flat filaments to the petals 

 as in one instance the margins of a petal could be distinctly traced down to the base outside though usually they form only ribs. The flat filament narrows at the apex like the broad filament of 

 Cyrillacea, the anther projects inwards bringing to mind Epacridece, and has appendages not unlike those of Ericaceae. The ovules are as numerous as in Erica, attached to a shield-like 

 placenta occupying the upper half of the cell, and the raphe is lateral with some variations especially in the upper and lower portions of the placenta, which may be regarded as the normal 

 position where the ovules are numerous and horizontal. 



Sapotaceje. 



Among species arranged as undetermined Sapotacece, which are probably not distinct from the genera at present known, is a Brazilian plant collected by Mr. Spruce, having a calyx consisting 

 of four sepals, the external two being valvate or very nearly so, and completely enclosing the two inner, which are much thinner and imbricated, and a monopetalous corolla with four segments.* 

 This so closely agrees in the calyx with Clusiacece, that Sapotacece may perhaps be as near Clusiacece as Ebenacece are to Termtromiacece. (V. Linn. Proc. vol. V., p. 54.) 



Brexiaceje. 



The more remarkable variations in the structure of this family are, that in Ixerba the stamens are distinctly perigynous, although it is in a low degree ;— that in Argophyllum the ovary is not 

 less than half inferior ; — that the flower of Ixerba has no barren stamens, while in Argophyllum they are more numerous than in Brexia, and adhere in parcels, which terminate in a fringe; — and 

 that the seed of Ixerba brexioides has a fungoid aril adherent to the raphe and covering about a fourth part of the surface of the seed. The last character I believe to be a more certain indication of 

 their affinity than any other, bringing them near Clusiacece ; and the minute embryo of Argophyllum and its small cotyledons may be compared to those of Marcgraviece. The flattened peduncle 

 of Brexia brings to mind the 2-edged and compressed stems of Hypericacece. The seed both of Ixerba and Argophyllum contain a small quantity of albumen in which the embryo is enclosed. 



* The number and disposition of the parts of the calyx and corolla are the same as those of Imbricaria reduced to half; the segments of the corolla are, however, not subdivided as in that genus. 



