Appendix 1. 



DETERMIN\TIONOF THE TRUE AMOUNT OF PRECIPITATION AND ITS 

 BEARING ON THEORIES OF FOREST INFLUENCES. 



Ly Clkvklani) Anr.E. 



The accurate raeasnrement of rainfaH, or the confidence to he placed 

 in the rainfall data that are 'oeing- accumulated in such great quanti- 

 ties, is a matter of profound importance in climatology, in engineering, 

 and even in storm predictions. This subject excites much considera- 

 tion in the public mind, since the comparison of the ancient and pres- 

 ent condition of Asia, Europe, and America leads to many forebodings 

 as to our own future; it is therefore eminently worthy of full treat- 

 ment and critical discussion. We must be satisfied as to the degree of 

 reliabihty of our data before Ave use them in attempting to elucidate 

 such Ojueries as are suggested by thoughtful men and such as were in 

 fact discussed at a recent symposium of the Philosophical Society at 

 Washington upon the question '<Do forests affect rainfall?" 



On that occasion extensive statistics were presented by several mem- 

 bers of that society, but it seemed to me that certain other fundamen- 

 tal questions which are discussed in this paper nuist be considered if 

 we desire to get at the exact truth in regard to this matter. 



The questioD, "Whether forests affect rainfall f" can apparently he attacked 

 from several .sides. Some are satisfied to appeal to deductive reasoning, therel)y 

 proving that the forests ought to affect raiulall, and their arguments have much 

 plausibility; others are satisfied to appeal to the historical evidence and quote the 

 dried-up streams of Europe, Asia and America, and the deserts that were once gar- 

 dens. But the instrumental mcteorok)gist and the observer who knows that rain 

 gauges have been carefully observed here and there for the past 300 years, or since 

 the first gauges were used by Leonardo da Vinci, still naturally look to the 

 records of the many stations that are accessible to us and expect these to give a 

 very definite statistical answer that can be relied on quantitatively as well as quali- 

 tatively. 



Such raini'all records miist be the nllimafe test of the truth of any hyl)()th(^sis, 

 hut the (luestion is one of considerable difficulty since we have todetcniiine a small 

 quantity by means of observations that may be liable to large errors, and it is nec- 

 ■Bssary to be on our guard against fallacious reasoning and against the liability to 



173 



