Mr. n. Claikf on Columelliaceae. 101) 



side, forming a inonadi.-lphous bundle; and each of them has 

 also in some degree the beaked connective of that of Hedyosiimm, 

 the beak of the central perfect anther being longer than those 

 of the sides. In one species of Sarcandra, the two cells of the 

 anther are attached exclusively to the upper half of the broad 

 thifkened connective, and this well represents the central anther 

 of Cliluranthus. 



It deserves notice, however, that Sarcandra has sometimes an 

 additional half-anther on one side of its stamen (very rarely on 

 both sides), which is imperfect, the line of dehiscence being 

 faintly or not at all marked; and this, not being partially sepa- 

 rated, as in CIdoranthus, certainly gives it an anomalous appear- 

 ance ; but a comjiarison with the anthers of Cliluranthus will, I 

 think, leave no doubt as to its origin ; — where no trace of a line 

 of dehiscence exists, it may consist of only a quarter of an an- 

 ther, i. e. of half a cell divided vertically, which occasionally 

 does occur accidentally in ordinary anthers; its rudimentary 

 condition, however, would pcrhai)s better account for its imper- 

 fection. Tiiis additional half-anther also sometimes presses on 

 the lower part of the lobe above it, occasioning it to bifurcate, 

 which mav serve as a lusus in the structure of anthers to assist in 

 explaining other irregularities. (PI. VI. fig. 16.) 



In some observations relating to dorsal placentation, I some 

 time since suggested that Chloranthus might be an instance of 

 that mode of attachment of the ovule ; but the structure of the 

 ovary of Iledi/osmum, in which there are three perfectly distinct 

 cord-like ])laccnt8e, alternating witli the three angles of the 

 ovary, and with the lobes or angles of tlie stigma, shows that 

 there is nothing unusual in the placentation of this family, and 

 that the carpel in Chloranthus is consequently always anterior. 

 And that the fertile carpel in Hedi/osmum is also anterior, seems 

 placed beyond doubt by the two placentae which are attached to 

 the anterior carpel being thicker than that on the posterior side 

 of the ovary ; the anterior carpel also projects more, which would, 

 however, be a character of doubtful value, unless confirmed by 

 the difference of the piacentse. (PI. VI. fig. 19.) 



XI. — On the Anthers o/ Columelliaceae a7id Cucurbitacete. 

 By 13. Clarke, Esq., F.L.S. &c. 



[With a Plate.] 



The affinities of Columelliacese remaining hitherto in much 

 obscurity, owing probably in part to some remarkable peculiari- 

 ties in the anthers having been imperfectly described, an analysis 



