184 FOREST iNFLUENCEg. 



and the irregularities iu liorizoutal distribution are presumptively greatest at that 

 time. During the winter the extended layers of clouds give us no a priori reason 

 to expect large irregularities in the geographical distribution of snow fall and rain. 

 Hellmann's records show that the geographical irregularities in the catch of his 

 gauges is really least in summer and greatest in winter, thus confirming our convic- 

 tions that on the average of the year the precipitation is uniformly distributed and 

 the variations in catch depend on the geograi)hical distribution of the wind at the 

 gauges during the fall of rain and snow. 



The eleven gauges here selected from Hellmann's data were unprotected and uni- 

 formly 1.07 meters above ground, and it is evident that they would not have neces- 

 sarily shown a similar discrepancy of 6 per cent among themselves had they been 

 placed at some other altitude. As the absolute deficits of each gauge increase like 

 the wind with the square root of the altitude, so also should the apparent irregu- 

 larities in geographical distribution. But this rule should not be so far stretched 

 as to assume that gauges at the ground surface would therefore show uo irregulari- 

 ties in the horizontal distribution of rain, the fact being that there is even for them 

 an outstanding uncertainty of 2 per cent, which is the total combined effect of all 

 the irregularities of measurement and the drifting of snow or rain. 



In general, then, we conclude that in the case of a number of gauges placed within 

 a few miles of each other, and of which we know nothing as to the height and exposure, 

 except that in general the observers have placed them in fairly open situations, there 

 is no reason to give a preference to the reports of one gauge rather than that of 

 another, since if the observers are equally reliable the irregularities of catch are 

 likely to far exceed the errors of careful observers. Again, the probable error of 6 

 per cent, due to unobserved and uncontrollable irregularities in the action of the wind 

 on these ordinary cylindrical gauges located 1.07 meters above the ground, indicates 

 the utmost limit to which any attempt at refinement in drawing annual isohyetal 

 lines should be carried at present, at least in the climates such as thatof Berlin, and 

 until the data are corrected for wind effects. Finally, any attempt to deduce from 

 such gauges the relative rainfall over the forest, the cleared land, the hill and the 

 valley, can only be successful in so far as we make due allowance for the influence 

 of the wind and the character of the precipitation. 



CHRONOLOGICAL VARIATIONS OF RAINFALL. 



What has just been said with regard to geographical distribution holds good 

 equally with regard to the chronological variations in rainfall. Undoubtedly there 

 are years of large and of small precipitation, but if we analyze these years we shall 

 see that they differ, not only in the quantity, but at the same time in the quality of 

 the precipitation and in the forces of the winds. Until we are able to correct the 

 measured rain or snow for the wind eff^ect we must include this large source of uncer- 

 tainty in the catalogue of errors to which our measurements are subject ; thus, in some 

 years, there may be a heavy snowfall of very light snow flakes falling during strong 

 wind, and in spite of all our eff"orts to estimate we get too small a record. Again, if 

 we confine ourselves to the summer rains only, namely, those that directly affect the 

 growth of plants, we shall find that in almost every long-continued series of observa- 

 tions at any locality trees, houses, and other obstacles have gradually grown up in 

 the neighborhood so that the average wind force at the gauge has undergone a steady 

 progressive diminution and the gauge, therefore, catches a larger percentage at the 

 close of the series than at the beginning, unless the obstacles were always so near as 

 to shelter the gauges. I have computed the departure of each annual total precipi- 

 tation (rain and snow) from the mean of forty-six years at Fort Leavenworth, Kans. 

 (using post-surgeon's record only) ; of twenty-two years at Spiceland, Ind. (observa- 

 tions by H. R. Dawson), and forty-two years at Washington, D. C. (observations at 

 the Naval Observatory). From the mean of these departures it is easy to compute 



