THE BALANOPHORAL DIVISION. 



17 



that they could with difficulty be distinguished except that those of the latter have a small gland near the apex ;— the very short filaments of Axinandra depressed so as to be horizontal, and the 

 singular thickened connective with small anther lobes on its under surface, are very nearly identical with those of Memecylon, if indeed they are distinguishable ;— and the calyx and corolla are 

 very nearly alike, except that the lobes of the calyx in Memecylon are slightly imbricated, while those of Axinandra are valvate, those of the former however being thickened and closing over the 

 petals in aestivation much like those of the latter. 



But the winged seed of Axinandra is so precisely that of a Duabanga that I believe it is correctly referred to Lytliracece. The embryo is straight, and the cotyledons have their edges in 

 apposition with the raphe, but in Memecylon the flat surfaces of the cotyledons are in apposition with the raphe,* which is probably a difference of the same kind as the cotyledons accumbent and 

 incumbent in the Cruciferce, as the raphe is always in the same line as the curvature of an embryo. This difference in the position of the cotyledons I believe to be the only distinction which 

 deserves notice as separating Lythracece from Melastomacece. 



There are also some other approaches between Lythracece and Melastomacece which are very close, as that of Chcetogastra divaricata, which has the dichotomous inflorescence of Cupkea with 

 a sessile flower in each fork ; — that of Microlicia cuspidifolia in which the ovary, which is of an elongated oval form, is quite superior, being attached to the calyx only by its flat base ; — and on the 

 other hand that of Physocalymna among Lythracece has a free central placenta with numerous ovules, the ovary consisting of three carpels, traces of the dissepiments being seen on the inner 

 surface, showing an analogy with Spathandra and some species of Memecylon as far as a 1-celled compound ovary is concerned. 



Lythracece also approach Myrtacece, the genus Sonneratia having been referred to the latter family, but I find it is so near Duabanga among the former that unless it is in the embryo, 

 which in all probability is the same, they differ only in the fruit of Sonneratia being in some degree succulent, the ovary of Duabanga molluccana being more inferior than that of any species of 

 Sonneratia I have seen. The valvate calyx and other parts of the flower are stikingly alike, as also the remarkably vascular wood. The calyx in Myrtacece when fully developed is much imbricated. 



The ovary of Pleurophora has the appearance of being 1-celled, and if so the carpel would be always posterior, i.e., supposing it to consist of a single carpel, but having repeatedly 

 examined it I have no doubt it consists of 2 carpels and is always 2-celled, the 2 cells communicating above so much as sometimes to make it appear 1-celled. The anterior cell I have never found 

 to contain more than 1 ovule, but the posterior cell usually contains from 3 or 4 to 8. The anterior cell is much shorter, and the single ovule arising immediately below the edge of the dissepiment 

 where the 2 cells communicate, it is uncertain to which it belongs, but the raphe next the placenta always shows that the anterior cell contains 1 ovule. Although I have never found the anterior 

 cell empty, I have occasionally found the ovary with the two cells much less unequal than usual, and only 1 ovule in each. 



VOCHYSIACEJE. 



The irregularity of the flower of Erisma is the same as that of Cuphea, and the only distinction between this family and Lythracece is that the calyx is imbricate, the presence of stipules not 

 being without exception, as in Salvertia they are wanting. A tendency to an irregular flower occurs also in Melastomacece, so that it is probable that Lythracece and this family are two sections of 

 Melastomacece, but they should be kept distinct if they can, as in all other cases where the genera are numerous and for the most part different in appearance. Among other agreements between 

 this family and Lythracece are the large wooded fruits, occasionally verticillate leaves, quadrangular stems and a close resemblance in the wood, that of Qualea grandiflora agreeing with that of 

 Sonneratia, Duabanga and other genera of Lythracece in having a rather large but dense pith, very numerous distinct medullary rays, and vessels rather large (much larger than in Myrtacece) and 

 often in pairs. 



CUCURBITACEJS. 



The structure and affinities of this family are so well known that I have only to add that in Sicyos angulata the raphe is dorsal, or possibly lateral, but not next the placenta, which may be 

 regarded as an approach to Onagracece and Myrtacece, and as in some degree removing them from the hypogynous families with which they have been compared.f 



* For a further notice of the embryo of Memecylon see remarks on Combretacece, Table IV. 



t For a comparison between the anthers and those of Columelliacece see Ann. Nat. Hist. 3rd Ser. Vol. I. p. 109. 



