116 FOEEST INFLUENCES. 



Celebes, where there is no such defore>tiiig-. giv^es the normal annual 

 values : 



Celebes: 



Northern peninsula — windward, 110.6 inches (2,809 mm.); lee, 55.1 inches 



(1,399 mm.) 

 Southwestern peninsula— windward, 136.2 inches (3,458 mm.) ; lee, 100.0 inches 

 (2,540 mm.). 

 West Centi'al Java : 



Windward, 78.4 inches (1,991 mm.) ; lee, 182.3 inches (4,630 mm.) 



It appears that in Celebes the windward side has a notably higher 

 rainfall than the lee side, while in Java it is reversed. Is this not due 

 to the relative lack of forest growth on the north side of Java? Where 

 forest growth exists the rainfall is higher, even there. For instance, 

 at Lormadjang, in a densely wooded level area, the annual rainfall is 

 73.7 inches (1,872 mm.), while at the two nearest stations on the same 

 side it is only 4(3.9 inches (1,191 mm.). 



The literature of the historical treatment of this problem is very 

 great, and the most extreme and absurd opinions can be found registered 

 in it. It usually suffers under the ol)Jection, first of imperfect data at 

 the beginning, and always under that of uncertainty as to the cause of 

 any variation of rainfall which may be found. 



The argument founded on changes in regions mentioned by ancient 

 writers deserves consideration only to point out its extreme unreliabil- 

 ity. The condition of the countries in question at the beginning of the 

 comparison is based on incidental references in the classics which are 

 not only incomplete but which are usually in ambiguous terms. The 

 classical terms for forest have an extensive range of meaning, as will 

 be seen by reference to any classical dictionary, greater than that of 

 our word forest, which is used in law English to designate areas which 

 now have no woods, and in some cases there is no evidence that they 

 ever had any. 



Much more trustworthy are the more modern cases which depend on 

 meteorological observations; but in these the earlier observations are 

 not so trustworthy as to give assurance that any differences found are 

 not the result of imperfect observations. The Avriter endeavored, some 

 years ago, to show that along the parallels of latitude 40o and 42^1 iu 

 the Mississippi Valley the lines of higher rainfall were moving slowly 

 westward with the advance of settlements, but it was abundantly 

 shown that the sparsely scattered earlier observations were (luite un- 

 trustworthy. 



There is one further case, which is quoted by Dr. Brandis, for India. 

 In the part of the central provinces between the Nerbudda River and 

 Nagpur and Rajpur, embracing a part of the Satpiira range of moun- 

 tains, much attention has been paid for several years to the care of the 

 forests, and specially to protection against forest fires. In consequence 

 a large territory, with scattered tree growth or entirely treeless, has 



