including a new Arrangement of Phanerogamous Plants. 13 



as in Nelumbium, and is always more or less distinctly visible in 

 the early stages of the flower* (PI. III. fig. 5). From this cha- 

 racter th€ nearest affinity of Ceratophijllum may be, as first sug- 

 gested by Dr. Asa Gray, with Nelumbium rather than with Pipe- 

 racese, which, from the position of its carpel, I formerly supposed 

 might be its true station. Its habit however is more that of 

 Cabomba, with which it agrees in the stamens being turned out- 

 wards instead of inwards, as in Hydropeltis ; it may also be re- 

 garded as having some analogy with Hydrocharidese in its ortho- 

 tropal oviile, exalbuminous seed and unisexual flowers, and pos- 

 sibly with Cryptocoryne in its many-leaved plumule, and it 

 further agrees with Hydropeltidefe in its cellular leaves. 



4. Chloranthus. The ovary of this genus agrees with that of 

 Ceratophyllum in always having the appearance of a funiculus 

 arising from its base which is constantly attached to the posterior 

 side. That the posterior is the dorsal side of the ovary is the most 

 probable, as the stigmatic tissue always descends on its anterior 

 side, having first obliquely crossed its thickened summit ; and 

 that this is the true structure of Chloranthus is placed almost 

 beyond doubt by the carpel in Piperomia and in Houttuynia 

 when single being always posterior. (See the figure of Chlo- 

 ranthus accompanying Part III.) 



5. Arum maculatum. Ovules five or six ascending, always 

 attached to the posterior side of the caq)el (its lower half) ; 

 stigma having its anterior surface only stigmatic as in Piperaceae, 

 the posterior being not unfrequently almost vertical and rounded 

 (PI. III. fig. 8). iUthough it may not be considered as fully 

 demonstrated that in Arum the carpel is posterior, yet it is ob- 

 vious that it must be either always anterior or always posterior, 

 and as it is variable and more frequently posterior in Typhacefe, 

 it is very improbable that it is always anterior in Arum, and a 

 comparison also with Cryptocoryne may perhaps be adduced as a 

 further argument. In genera nearly allied to Ai-um the placeu- 

 tation is however not dorsal, but the ovules being partly sutural 

 in Cabomba and entirely dorsal in Hydropeltis, shows that this 

 difiFerence of placentation may take place in genera very nearly 

 allied, if not in the same genus, as in Mesembryanthemum. 



Cryptocoryne, From Mr. Griffith's figures of C. ciliata 



* The following circumstance seems also to show the placentation of Ce- 

 ratophyllum to be the same as that of Nelumbium. Having three immature 

 fruits of Ceratophyllum demersum, I allowed them to remain growing on 

 the plant to produce seeds. But some time afterwards having observed an 

 altered appearance in one of them, I found on examination that the ovarj' 

 had entirely decayed away, leaving the ovule suspended from the apex of a 

 filiform perfectly entire funicidus which was posterior, and both funiculus 

 and ovule remained attached to the torus quite free from decay for two or 

 three davs afterwards. 



