81 



mistake (?) that probably originated in De Beauvois' descrip- 

 tion of them " Semina in pulpd carnosa nidulantia." Finally, 

 I myself, feeling that these could not be its true affinities, 

 placed it in the Campanul alliance, with marks of great doubt. 



It is obvious, from the foregoing description, that Napo- 

 leona has nothing to do with any of the orders to which it 

 has been referred. From Cucurbitaceaj it differs utterly in 

 its hermaphrodite flowers, axile placentation, highly developed 

 corolla, and whole habit ; it has in fact no resemblance to 

 that order. Passifloracese seem at first sight to claim a much 

 nearer relationship ; because of the triple-rowed corolla of 

 Napoleona, which much resembles the coronet of a Passion- 

 flower; but there the resemblance ceases. The tendrils, parietal 

 placentae, free ovary, distinct styles, polypetalous corolla, im- 

 bricated calyx of Passifloracea^, are all most essentially at 

 variance with the genus. Symplocaceaj were a far better 

 guess, for the monopetalous corolla, indefinite epipetalous 

 stamens, axile placentae, adherent calyx, and definite seeds of 

 Napoleona find there a parallel ; but the ovary of that genus 

 is wholly adherent, with a great epigynous disk, the calyx is 

 valvate, and the seeds have no albumen, to say nothing of the 

 lacerated condition of the corolla, which is not to be wholly 

 disregarded in a consideration of this kind. 



To me it appears that the true affinity is in the neigh- 

 bourhood of the Mangroves (Rhizophoracese); for the follow- 

 ing reasons. The ovary is in both inferior, few-seeded, with 

 axile placentae ; both have a coriaceous valvate calyx; both 

 have large amygdaloid seeds without albumen. The placenta 

 of Kandelia is almost the same as that of Napoleona, and in 

 the former genus the petals are broken up into numerous 

 fringes quite analogous to those of the genus in question. To 

 this may be added the great resemblance that exists between 

 the wood of Napoleona and of young Rhizophora, in conse- 

 quence of both consisting in part of slender acicular tubes, 

 which give the wood, when broken across, the appearance of 

 containing slender bristles. Finally, the ribbing, which is so 

 conspicuous in the outer corolla of Napoleona, is repeated in 

 the calyx of Bruguiera gymnorhiza. It is true that the one 

 genus is monopetalous and the other polypetalous, but I cannot 

 attribute much importance to that character in a case where 

 the stamens adhere so slightly to the corolla. 



While, however, there is this reason to believe that Rhizo- 



