71 having the overy 10-celled, vith two ovules in each cell, but does 

 not figure the seeds. Now, strgnges es it may seem, I heve since 

 discovered that all this is quite v/rong, and we all have been mis- 

 taken. Because I have found that the ovary appears to be composed 

 of a solid mass of cellular tissue v/ith many small cells immersed 

 in it and arranged in pairs in a circle, each filled v/ith one ovule. 

 This tissue, v/hich looks so solid and uniform, is really composed of 

 many carpels fused together into a m?ss, and in the process of ripen- 

 ing the carrels separate from one another, and the tissue composing 

 them dries up and forms a broad, reticulated wing around each seed, 

 ^o tha"*", when rine, the senarate carpels, erch tightly enclosing one 

 seed, have all the appearance of a whorl of winged seeds around a 

 central axis, as described in the key. -'•'heref ore , if the above 

 quoted paragraph in the key is altered to read — "Fruit apparently 

 one-celied; apparent seeds (really 1-seeded carcels) large," flat and 

 broadly winged all around, arranged in a single v.horl around the 

 central axis," it v.dll be correct, -^-s this seems to be a very in- 

 teresting and unique type of structure, it is remarkable that ^outh 

 African botanists have not discovered its true nature long ago. In 

 "Surone such a structure would have received much attention. 



Key to the "Genera. 



Restricting I'"esembryanthemum to species having the seme type of 

 floral and fruit-structure as that possessed by M. umbellatum/ 

 Linn., the follo^"ing ey defines the chsracter.s of that genus and some 

 other genera that have been or are nov; proposed fcr species pre- 

 viously considered as vegetative groups of one genus with a variable 

 fruit structure, or of v.'hich the fruit was then unknown to me. V/ith 

 the exception of some srecies of ^-esembryanthemum, '-"-'richodiadema 

 and I'^itrorhyllum, all are plants hav ng distinct, elongated inter- 

 nodes between their leaf-r>airs on the '^rimary branches; and their 

 placentas are on the floor or outer wall o-^ the cells in all the 

 genera here mentioned. 



1« Leaves of two kinds, all opposite; those on the plant v;hen 

 at rest, or the lovi-est of the tv/o pairs on a grov/th 

 dissimilar from others produced when in active growth or 

 frora the next pair developed; flowers solitary, pedici- 

 llate or sessile; stamens erect. 2 



Leaves all similar at all times, all opposite; flov/ers 



cymose or solitary, usually pedicellate, but occasionally 

 sessile; stamens erect. 6 



2, One pair of leaves when in active grov/th free to the base, 



follov.'ed by a pair united into e cone or cylinder for half 

 or more than half of their entire length, with spreading 

 tips; intern odes of stem evident or none, not bead-like; 

 stamens collecteci into a col\imn; stigmas, valves and 

 cells o-f the fr it 5-7; exrandingk-keels of fruit broad 

 and flat, v/ith marginal v/ings; cells vdth or v/ithout mem- 

 branous cell-wings; Placental tubercle none. C^'yi^e. 

 1'. grande, N. 3. 3r.O. i'-itrophyllum. 



One pair of leaves small and closed together .^ t first in 

 ellipsoid or globose form and with or without short, free 



