FLORA OF SOUTH AMERICA. 357 



without breaking through the only fixed characters by which 

 convolvulaceous genera have as yet been distinguished. They 

 are allied, on the one hand, to Maripa, on the other to 

 Erycibe, but abundantly distinct from both. The latter 

 genus has been raised by A. De Candolle to the rank of a 

 distinct natural order, but its connection with the true Con- 

 volvulacece appears to me far too close to justify the separa- 

 tion. De Candolle relies chiefly upon three grounds: the 

 aestivation, the quinary stigmate, and the entirely unilocular 

 ovarium. The first, the aestivation, is in Erycibe as in the 

 two new genera now proposed, induplicate-valvate, whilst in 

 ordinary Convolvulacece it is plicate ; but the only difference 

 between the two is, that in the former case the lobes are 

 divided to the bottom of the folds, whilst in the latter, the 

 folds extend below the divisions ; a distinction of very little 

 importance. The apparently quinary stigma does not 

 appear on a careful examination of several Erycibes, to be 

 organic, but the 5 or 10 oblique ribs and furrows in that 

 genus, as in Maripa, (as described by E. F. W. Meyer,) 

 are probably merely the impressions made on the stigma 

 in the bud by the quinary external parts of the flower, the 

 stigma being, in fact divisible into two lobes, as in most 

 other Convolvulacece ; the shortness of the style is but a cha- 

 racter of degree, and may be observed in Cuscuta. The third 

 distinction, the unilocular ovarium, is again of very little im- 

 portance in Convolvulacece, where the dissepiments are often 

 very incomplete, and always independent of the position and 

 insertion of the ovules. 



GeSNERIACEjE. 



Since the publication of the yth volume of De Candolle's 

 "odronius, the number of Gesneriacece known has nearly 

 doubled, the forms assumed by the flower and fruit have 

 been found to be much more varied than was supposed, the 

 characters assigned to the genera are no longer sufficient to 

 distinguish them clearly, either from one another, or from 

 the new ones since proposed, and the characters assigned to 



