ANONAGEJE. 



215 



of internal horn, which remained obtuse in the other species, is 

 flattened out and elongates inwards so much (fig. 260) that the union 

 of all the stamens forms a sort of chamber around the gyna^ceura, 

 covered in by a nearly Hat roof, only perforated in the centre to give 

 passage to tlie tops of the styles (tig. 258). The number of stamens 

 is not the same in all the species, and the apparently verticillate 

 arrangement of the fertile stamens does not prevent the presence of 



Fopowia {Clathrospermum) Barten 



Fio. 256. 

 Flower-bud {^^). 



Fig. 259. 

 Loiigitudiiuxl section of Howe 



Fig. 258. 

 Flower, perianth removed. 



a certain number of external staminodes arranged with some degree 

 of symmetry (fig. 252), and representing the shortest stamens 

 observed in P. caffra. As for the gynseceum and fruit, they present 

 every possible variation in the number and position of the ovules and 

 seeds, from a solitary erect ovule to a pretty large number in two 

 parallel rows, from the one-seeded berries of P. caffra and P. Vogelii 

 to the chaplet-like fruits of P. Heiideloti and the allied species.' 

 As known to us, this genus at present consists of about fifteen 



' In an African plant ascribed to this genus by 

 BENTII.4M & Hooker {Qen., 958) the flowers are 

 dioecious, and the carpels are about sixty in 

 number. If this is really the species of Mr. 

 Mann's collection (which we have analysed, and 

 for which a new genus should perhaps be esta- 



blished), its outer petals are alone well developed, 

 the inner ones being represented by very minute 

 obtuse scales, and the ovaries are each surmounted 

 by an ovoidal stigma, and contain at least six 

 ovules. (See Oliver, loc. cif., n. 2.) 



