22 



THE CHLOEANTHAL DIVISION. 



when growing separately, so that it can take its natural form, becomes curved in the same manner as the seeds of Elatine. This curvature, as in other similar cases, is owing, there appears no 

 reason to doubt, to the curvature of the seed, not of its carpellary covering. 



From the habits of the species when they become terrestrial, it might be anticipated that the affinities of the genus would be those already suggested, especially if amongst such affinities 

 genera occurred with 2-celled carpels, and this division of the cell takes place in Batidece (Trans. Linn. Soc. vol. XXII. p. 412), and also in Tetradiclis* among Crassulacece, of which perhaps 

 Elatinacece and Callitrichacece may be regarded as reduced forms. Tetradiclis differs, however, in the raphe being lateral, at least it is so in the lateral enclosed ovules in which I believe its position 

 can be better relied on than in the intermediate ovules. In Callitriche there is a tendency in the half of each cell to adhere more firmly to the half cell of the other carpel than to that of its own, 

 and this also takes place in Tetradiclis, the halves of opposite carpels remaining firmly adherent after dehiscence. 



It is therefore near the Elatinacece that I would station this genus, and it does not appear to nearly approach any other family, except the Boraginece, with which it agrees in the structure of 

 its ovary, ovule, and seed, as noticed in the Trans. Linn. Soc. vol. XXII. p. 412. In this point of view it connects Elatinacece with Boraginece, which are also allied to each other through 

 Crassulacece, which I have endeavoured to explain elsewhere. 



Batis is so near Verbenacece that the spikes of fruit and seed are identical except in the position of the two carpels (1. c. p. 412). 



POD O STE MONEiE. 



Castelnavia may be compared to Callitriche in its two stamens being posterior (v. Callitrichacese), and also shows a close resemblance to that genus and Tetragonia, in the surface of the stem 

 being covered with glands, which in the dried plant have a flattened appearance, and if examined in the fresh plant they would, it might be expected, prove identical with them. In the number of 

 their sepals, from three to eight, and of their stamens ; in the carpels being, in some genera, united by their margins ; and in the hardened texture of the capsules of some genera I believe they 

 approach Polygonacece. The numerous ovules occurring in Saurureae compared with the single one of Chloranthacese, may assist in accounting for the difference in these characters between 



Podostemonece and Polygonacece. 



Dr. Lindley's suggestion that they are near Plantaginece is I believe quite correct, and it appears beyond doubt that the true station of this remarkable family is in this Division. One of their 

 differences consists in the compound foliage, but as in Plantago Coronopus the leaves are sometimes doubly pinnatifid, this character may to some extent be accounted for. 



Phytolaccace^. 



This family is generally considered either as a part of, or a near ally of the apetalous families allied to Caryophyllacece, especially of Nyctaginece, which they approach in habit, but to this 

 alliance there appears to me to be two objections, viz., the ovary of these families is syncarpous, with a free central' placenta where any is formed, except in Molluginece, but that of Phytolaccacece 

 is always apocarpous ; and the embryo of Seguiera is large and foliaceous with the cotyledons contorted laterally, much like that of Petiveria.j* The wood of Seguiera has medullary rays 

 showing a resemblance to Polygonacece, between which and Crassulacece appears to me to be their position. 



Chloranthace^:. 



The near affinity that exists between this family and Polygonaceae on the one hand and Gnetacese on the other, I have endeavoured to explain in the Ann. Nat. Hist. Ser. 3, Vol. 1, p. 106 ; 

 and more recently the genus Welwitschia has been added to Gnetacece, the monadelphous stamens of which may be regarded as a further approach between Gnetacece and these families ; they are 

 also somewhat unilateral, the anterior stamens being longer, in which they agree with Chloranthus. 



* I have no doubt that the division of the cell into compartments in Tetradiclis is of the same nature as that of BoraginecB and Callitriche. In a very early stage the ovary of Callitriche is round or very obscurely quadrangular, having 

 no appearance ot division. In Tetradiclis a few ovules intervene between the inflected dissepiments, by protruding as it were between them. 



t In Gallesia Gorazema the cotyledons do not fold over each other, in Petiveria alliacea very slightly so ; the former has a small quantity of albumen the cotyledons curving round it. 



