THE INFLUENCE OF FORESTS ON CLIMATE AND FLOODS 227 



but that does not prove 

 that the forests may not 

 increase the rainfall ; a 

 man ceases to drink 

 water after he dies, but 

 that does not prove that 

 the drinking of water 

 has no relation to keep- 

 ing him alive. There 

 may be ; and no doubt 

 is, a reciprocal relation 

 between the forests and 

 the rainfall. If the 

 rainfall disappears the 

 forests, of course, will 

 die, and great climatic 

 changes may make the 

 rainfall disappear; but, 

 all the same, the forests 

 may increase the rain- 

 fall. 



As a matter of fact, 

 this subject affords an 

 excellent illustration of 

 the imperfections of in- 

 ductive reasoning in this 

 case. The rainfall is an 

 exceedingly variable 

 quantity ; the annual 

 rainfall this year may 

 be double what it was 

 last year. Even taking the average in 

 ten-year periods, Professor Moore's 

 curve for New Bedford shows an aver- 

 age of about forty-nine and a half inches 

 in the ten years from 1820 to 1829, and 

 forty-three and one half inches in the 

 ten years from 1840 to 1849. I" 1883 

 the rainfall near Boston was thirty-one 

 and two-tenths inches; in 1888 it was 

 fifty-six and ninety-three one-hundredths 

 inches ; in 1892 it was thirty-nine and 

 four one-hundredths inches. Moreover, 

 these observations of rainfall them- 

 selves are subject to large errors. A 

 rain-gauge whose mouth stands one foot 

 above the ground may collect six per 

 cent more water than if it were placed 

 level with the ground, and if placed 

 higher the difference may be still 

 greater ; moreover, the methods used for 



VIRGIN SPRUCE IN WHITE MOUNTAINS 

 But little of this once typical iorest remains 



measuring the snowfall and reducing it 

 to inches of water involves errors, and 

 there are still other sources of uncer- 

 tainty. Considering that, even if the 

 forests are untouched, the rainfall over 

 a given area may vary fifty per cent or 

 more in consecutive years, how can it be 

 expected to demonstrate by a compara- 

 tively few years of observations the 

 effect of the forests — which, in any case, 

 will be small ? Any effect of the forest 

 would very likely be less than the prob- 

 able error of the observations them- 

 selves. Under these conditions, there- 

 fore, the inductive or statistical method 

 fails. What then, are we justified in 

 doing? 



Professor Moore draws this conclu- 

 sion : "Precipitation controls forestation, 

 but forestation has li'lle or no effect 



