including a new Arrangement of Phanerogamous Plants. 445 



fertile one, and on this accbunt Maranta is considered as Hete- 

 rocarpous. 



Arace^. Tn this and in the allied families the position of 

 the carpels when two appears variable; thus, in a species of 

 Pathos all are right and left ; in a Wallichia they are anterior and 

 posterior, with very rare exceptions; and in dicarpous ovaries 

 frequently occurring in Sparganium ramosum, they are variable 

 and sometimes oblique. 



In Typha and Sparganium the single carpel is very frequently 

 posterior, and in Arum I have endeavoured to show always so. 

 (See Part IV. Arum maculatum.) 



Heterocarpous Exogens. 



This division, although several nearly allied families are ex- 

 cluded, is yet in itself a natural assemblage ; the connexion how- 

 ever between the Gentianal and Nymphal Allian :es is weak, but 

 as some of the genera of Oi'obanchacese have the placentation 

 dorsal, there may be a direct affinity existing between them. 

 (See also Part IV. Hydropeltis.) Villarsia nymphaoides also cor- 

 responds with Nymphcea alba in the ovules being horizontal with 

 the raphe for the most part on the upper surface. 



GENTiANACEiE. In Leiantkvs the carpels are anterior and 

 posterior, and also in Erythrcea littoralis, but in E. centaurium 

 occasionally right and left, and Chlora perfoliata agrees with the 

 latter species. Chironia and Villaisia also form partial excep- 

 tions to the lateral position of the carpels in this family in pre- 

 senting an iiTcgular arrangement, some of the carpels being an- 

 terior and posterior and others oblique. 



Ceratophyllace.e. In Ceratophyllum a transverse section 

 of the ovary presents no appearance by which the position of the 

 carpel can be determined, but the stigma is unilateral and 

 curves posteriorly with but little variation, sometimes curving 

 down behind the ovary ; in addition to this it is grooved more or 

 less deeply, and this groove deepens towards the base in front. 

 The posterior surface is rounded and not stigmatic. (PL XIV. 

 figs. 1 & 2.) 



But the attachment of the female flowers and young branches 

 is somewhat different, each bud or young branch arising from 

 the axil of a leaf, but the female flowers rather from the stem 

 between two leaves : whether this should be taken into con- 

 sideration in determining the position of the carpel remains a 

 question. 



Chloranthace^. The only indication of the position of the 

 carpel in Chloranthus is that the stigmatic tissue obliquely crosses 

 the thickened summit of the ovary and always descends on its 



