THE RELATION OF FORESTS AND FOREST FIRES 397 



altogether probable. One set of facts which may ultimately be 

 used to establish this latter contention is found in the positions 

 chiefly or exclusively occupied by trees in semi-arid regions, 

 which positions are either along water-courses, and so shielded 

 from fire by moisture, or on rough and stony ground, and so pro- 

 tected by the absence of enough grass or other vegetable ground- 

 cover to carry a destructive flame. 



The same course of reasoning applies to certain kinds of open 

 glades or prairie, well named " fire-glades " by Mr Frederick V- 

 Coville, Botanist of the U. S. Department of Agriculture. In 

 the Black Hills of South Dakota, for example, these glades, sur- 

 rounded by forest-bearing land, are almost exclusively confined 

 to ground rich enough to support a crop of grass sufficiently 

 dense to burn fiercely, while the timber is restricted to rough 

 rocky or stony land, almost always higher than the glades and 

 comparatively safe from fire because of the scantiness of the 

 minor vegetation it is able to support. 



In semi-arid regions where fire-glades of this kind occur, there 

 is an interesting alternation, by years or series of years, of the 

 presence and absence of the moisture which makes forest repro- 

 duction possible. In the same way the occurrence or absence 

 of burning gives or denies an opportunity for young seedlings to 

 reach a size at which they are reasonably safe from the attacks 

 of ordinary surface fires. It must be clearly borne in mind that 

 it is only the average effect of the class of causes of which fire 

 and rain are the chief with which we are concerned. Young- 

 trees sometimes succeed through, combinations of temporary 

 immunities in establishing themselves in the midst of fire-glades 

 of old date, and the rocky refuges where some seedlings usu- 

 ally escape the fire are not uncommonly burned over, as the 

 fire-scarred trunks abundantly testify. But these facts do not 

 obscure the effective working of the averages, although they do 

 tend powerfully to lengthen the time required for the average 

 to work itself out. Thus reproduction around the fire-glades 

 of the Black Hills is extremely slow. 



Perhaps the most remarkable of the regulative effects of forest 

 fires relates to the composition of the forest — the kinds of trees 

 of which it is composed and the proportion of each. This effect 

 depends upon the action of fire in combination with the various 

 qualities of resistance which trees possess. . These qualities are 

 of two chief kinds; one, adapted to secure the safety of the in- 

 dividual tree directly through its own powers of defense, the 



