REPLACEMENT OF THE ANCIENT E. AFRICAN FOREST 507 



long only by, fire. Such a tradition as that mentioned, I be- 

 lieve, by Theal, to the effect that the Kaffirs on reaching the 

 Limpopo found savages there who, at that late date, had no 

 knowledge of fire, might tell against it. But even if the tradi- 

 tion be well-founded, it does not seem (|uite inijjossible that a 

 type as low as are the present Vaali)ens of that valley might have 

 remained without fire for an immense time even with fire-using 

 Bushmen and Hottentots in the country. 



It seems impossible to say whether the retarding factors T 

 liave referred to earlier will haxe been more than counter- 

 balanced >by accelerating factors. A single, really prolonged 

 period of drought might have seen immense fores't-obstruction. 

 Again, a slow and gradual start from a few small centres is 

 possible, or, on the other hand, the fires may have come in along 

 the whole dry western front, the forests of the inland i)lateau 

 coumtry, if they existed so late, having been burned oft' first. 

 It might also be argued that tlie dense forests are mainlv a 

 mountain phenomenon, and that it is unnecessary to assume 

 that, at this latitude, they may have extended to the coast. A 

 once more or less continuous stri]) of forest on the mountains 

 and eastern foot-hills would exjilain distribution, and have 

 needed less time for its destruction. Against this we must i)ut 

 the fact that the South Melsetter and Mossurise forest-patches, 

 though occurring at from 3.000 to 4,000 feet, are essentiall>' 

 of a lowland type, that such patches occur also in the lowlands 

 and not ifar from the sea, and that dense, high forest is simi- 

 larly found in the lowlands in British East Africa. 



{b) Pyrophytic Forest. — The trees that are associated 

 with the annually-burning grass-veld are fully as intolerant of 

 dense-forest coiuhti»jns as are the grasses themselves. Ada])ta- 

 tions of the bark enable them (once well established) to meet the 

 fires with nothing worse than a loss of leaves and their seedlings 

 survive both fires and winter drought in a marvellous way ; but 

 the latter, in contrast to the true forest seedlings which will keep 

 alive for years in the undergro'wi:'h awaiting light in order to 

 grow tip, cannot long survive damp and shade. I have numbers 

 of times seen masses of Uapaca seedlings in particular, dropped 

 in the forest by Kafirs or baboons, germinate, yet fail every time 

 to persist. It is correct, therefore, to say that the factors that 

 have reduced the dense forest areas have also brought about the 

 present wide distribution of the trees of the pasture-lands. 



Derivation. — Dr. W. L. Thompson has said to me that he 

 has been much impressed with the probability that the one type 

 has arisen (from the other by the fact that the same genus so often 

 possesses representatives both in the forest and in the pastur'^ 

 Ochna, Eckebergia, Rlioicissiis, Parinarinm, Eugenia, Vangueria, 

 Stryehnos, Vitex, Bridelia. Eicns, are amongst the instances of 

 this quite striking fact. If (as I think we must) we regard 

 regidar fires as a relatively recent phenomenon and the ])resent 

 " wooded-pasture " trees as belonging to a later, specialized, defin- 

 itely pyrophytic type, it certainly seems not entirely unlikely that 

 if we could go back far enough we should find that they are. each 



