508 REPLACEMENT OF THE ANCIENT E. AFRICAN FOREST 



one, eventually thus derived from a forest-inhabiting ancestor. 

 But the present wide distribution in Africa of many of them sug- 

 gests that for these we should need to go back very far indeed, 

 and that the replacement of any particular great piece of forest 

 has probably proceeded (as it does under our eyes to-day) far 

 more by immigration and spread of already-specialized forms 

 than by the trans'formation of its own trees. It is likely enough 

 also that some of the forms came in, already ombrophobous, from 

 an equivalent to our present grass-veld that may already have 

 existed in drier areas long before tire, in the hands of man, 

 became an important factor. This would entail a less violent 

 re-adaptation than is presupposed in the transformation of a 

 forest-tree into a pyrophyte and, especially, of its seedling into a 

 seasonal xerophyte. and the indirect palaeontological evidence, such 

 as there is of it, seems to point to the existence of open country in 

 early Tertiary times. Against the view that such country may 

 have existed in potential forest areas before man and his fires 

 commenced to destroy, may be placed an argument which anyone 

 who has lived long in this country — in this part of it at any rate 

 — will have seen growing u]> under his eyes. The fact is that 

 land from which fire is excluded tends to go back to dense bush. 

 Even the more open grass-veld here is full of stumps that seldom 

 get further than a one season's shoot.* The very fires that have 

 rendered their existence on that ground possible, by driving the 

 (forest ofif it, keep them from growing up until, some year, a |xoor 

 'burn, it may be, or no burn, allows of a second season's growth 

 being superadded to the first and gives a more fire-resisting bark 

 to the latter — just as the shade that secures the survival of the 

 forest seedlings may also keej) them back till the sun, some year, 

 gets in. Keep the fire from such a piece of ground^-or 

 burn too soon — ^for several years, and these shoots grow up and 

 eventually in places become so dense as to reduce the grass and 

 the severity of the fires and to allow semi-forest types 

 as Marklmniia lanata and Albissia chirindensis to spring up 

 amongst them as I shall describe below, and eventually to replace 

 them. The result, when this occurs so far from high forest as 

 not to obtain seeds from it, is a form of dense thicket, that is not 

 uncommon in this section. I am not at all sure that very large 

 parts, at any rate, of the great rubber-forests of the Portuguese 

 lowlands may not be of this type. I was much struck, when 

 there, by the frequent intermingling seen of the trees usually 

 ifound in pasture with the shrubs and climbers of the denser 

 growth and the smothering by these that was actually in pro- 

 gress. 



The semi-forest types referred to just now, and especially 

 such a one as Rainvolfia inebrians, at home even in Chirinda yet 

 capable of holding its own outside with only a little protection 

 from fire, may represent the sort of form through which our 



* Cf., also, Neave ("A Naturalist's Travels on the Congo-Zambezii 

 Watershed." Geog. Jul., Feb. 1910, p. 138"):" The tree-stumps . 

 have to undergo a very keen struggle to survive the annual 1)ush hres." 



