28 Bews. — Some General Principles of Plant Distribution as 
west, yet along the central and eastern mountain-ranges we have many 
representatives (there are twenty or so on the Drakensberg), and some even 
compete with subtropical species in the eastern grass-veld, e. g. Danthonia 
purpurea around Molteno and Pentaschistis natalensis in Natal. As 
Stapf (33) has pointed out, temperate species of Avenastrum , Agrostis , 
Melica , Poa, Festuca , Bromus , Br achy podium, form, as it were, a bridge 
across the mountains of the tropics connecting with the northern centres 
of distribution. The south-western section of the Gramineae, then, like 
most of the other characteristic south-western families and large genera, 
are not without representatives eastward and northward. 
The Gramineae of the great eastern grass-veld region and of the central 
and western regions have much more obviously invaded South Africa from 
the tropics. The genus Aristida has produced species, many of them rather 
closely connected, e.g. A. congesta and A. barbicollis , the former very common 
and widespread, the latter much rarer, which are* of supreme importance in 
the early stages of the establishing of grassland. Other similar pairs are 
A. angustata and A . junciformis, A. bipartita and A. Burkei. 
Aristida has also a very distinct section (S tip agrostis), the species of 
which are adapted to desert and semi-desert conditions of the dry central 
and western areas. No better example of differentiation could be found 
anywhere than that of Aristida in South Africa. 
Eragrostis is another great pioneer genus in grass-veld, with about fifty 
species. Many of the rare Transvaal and Free State endemics (. Eragrostis 
barbinodis , E . pattens , E. Witmsii, E. sporoboloides , E. Atherstonei , 
E. ech in och to idea , E. margaritacea , E. dura) may very well have been 
derived from the widely distributed species, E. curvula , which is so wide- 
spread and abundant everywhere. Except for the shorter valves, the 
differences are in no case very great. Numerous similar examples could be 
given from the Andropogoneae, which are characteristic rather of climax 
grassland. Andropogon hirtus , for instance, has probably produced A. audits ; 
A. nardus is in the process of breaking up into several distinct species, and 
has already given rise to A. plurinodis. The variations of the dominant 
grass, Anthistiria imberbis ( Themeda Forskalii ), should be studied from the 
same standpoint. The numerous mesophytic species of Panicum and other 
Paniceae which are characteristic chiefly of the moist marginal belt sur- 
rounding forest and scrub are of interest, since some of them show discon- 
tinuous distribution, particularly between the Natal coast-belt and the 
northern Transvaal. Many of them, however, extend into the tropics and 
have simply invaded South Africa at different points. I regret that space 
does not permit of a fuller discussion of the Gramineae from the standpoint 
of the present paper. In fieldwork, I have recently found that the presence 
of many species in certain grassland areas can be explained only on these 
lines, i.e. by taking into account their probable origin from other widespread 
