November 23, 1888.] 



SCIENCE. 



243 



Europe, concerning the supposed influence of woodland upon pre- 

 cipitation. 



Whiile criticising tlie metliods used in tlnese investigations, Dr. 

 Fernow did not. if I understood him right, give his own views upon 

 the main question. 



The question is, Does the presence of woodland influence rain- 

 fall, does its increase increase rainfall, and does its destruction 

 reduce it ? I know no theoretical grounds upon which an affirma- 

 tive belief can be based. 



While the question is an interesting one to science, it is also a 

 particularly important economic one to this country. The future 

 value of a large part of our arid region, and our policy in regard to 

 it, depend upon the decision. If the presence of woodland in- 

 creases rainfall to an economic extent, we should begin at once to 

 plant trees all over our Western plains, and supply them with water 

 until they in turn supply themselves and the adjacent land with 

 moisture. If they do not increase the rainfall, then perhaps the 

 land vv'hich is now being covered with woods can be more profit- 

 ably used for wheat. Again : the welfare of much of the eastern 

 United States, now well watered, may turn in the future upon the 

 decision of this question. In some parts the forests are being cut 

 away, and thereby the rainfall may be reduced to such an e.ttent 

 as to make the soil unproductive, in which case the timber-cutting 

 should be stopped in time. 



From this the economic point of view, it must be recognized that 

 to be of any value, the influence of woodland upon rainfall must be 

 considerable in amount. A minute influence is equivalent to none 

 at all. so far as economic effect is concerned. If it should be found 

 that woodlands induce only a trifling modification in rainfall, the 

 solution of the question is substantially in the negative, viewed 

 economically. In our examinations of recerds and other tests of 

 the comparative amount of rainfall under differing conditions of 

 forest-covering, we are, then, to look for changes of considerable 

 magnitude. The variations in rainfall from year to year and from 

 place to place are great, — so great as to mask, in a limited series, 

 or in observations at a few stations only, any general change. It is, 

 of course, understood that the difficulty in the way of detecting the 

 general movements of rainfall lies in these temporary and local 

 fluctuations, and it is apparent that to eliminate them it is necessary 

 to use what Dr. Fernow aptly calls the wholesale method, to 

 bring together into the investigation large numbers of observations, 

 from many stations, scattered widely over the territory under ex- 

 amination. I cannot conceive any retail method that will yield a 

 result worthy of any confidence, as is shown by the fact that it is 

 very easy so to select stations and years of observation as to ob- 

 tain any desired result. 



Of the retail methods of investigation in use in Europe, cited by 

 Mr. Fernow, that of pairs of stations, one situated within the forest, 

 the other loo metres outside it, seems to promise no result ; first, 

 because it is a retail method, and, second, because if the forest 

 has any influence, it must, in order to be of any value, be felt 

 more than 325 feet away from the margin of the forest. We 

 cannot afford to cover the land with woods in order to in- 

 crease the rainfall. We must have some land to cultivate. The 

 conclusions from the observations in Bohemia, cited by Mr. Fer- 

 now, can only be misleading. To compute from the rainfall in the 

 open valley, and from an estimate of the rate at which rainfall in- 

 creases with elevation, the theoretical rainfall upon cleared moun- 

 tain-sides, and then to conclude from the discrepancies between 

 these results and the observed rainfall upon the timbered moun- 

 tains that the forests have had a certain effect upon the rainfall, is 

 a case of theory run riot. 



I know of but two attempts to use the wholesale method, both 

 of which were mentioned by Mr. Fernow, — that of Mr. Harrington 

 and my own. Mr. Harrington's method consists in a comparison of 

 two rainfall maps made from data of different dates, — the Blodget 

 map, made in 1857; and the Denison map, made in 1844. The two 

 maps are not strictly comparable, as the first purports to show 

 areas of equal rainfall, while the last shows lines of equal rainfall. 

 Nevertheless, the former may be made rudely comparable with the 

 latter by means of certain assumptions regarding the relative posi- 

 tions of these lines and areas. Mr. Harrington's examination was 

 confined to the supposed increase of rainfall on the plains. Find- 



ing that the isohyetal lines of 20, 25, and 30 inches were in some 

 places slightly farther west on the Denison than on the Blodget 

 map, he concluded that the rainfall has increased. 



What is the value of this evidence, and, first, of what authority 

 are the maps? Upon the Blodget map I find only five stations in 

 the entire area of the plains, north of Texas ; viz., Forts Riley, 

 Leavenworth, Atkinson, Arbuckle, and Kearney. The only data in 

 this area of nearly half a million square miles consists of the ob- 

 servations at these five stations. It may be safely said that the 

 rainfall-curves in this area are at least 99 per cent hypothetical. 

 They might as well be drawn a hundred miles on either side of the 

 position assigned them by Mr. Blodget, without contradicting the 

 observations. Their position is necessarily based almost entirely 

 upon Mr. Blodget's judgment, and not in any appreciable degree 

 upon observational data. The Denison map is better. The worst 

 that can be said of it is that it is a popular map, made to sell. 

 But the weakest link in a chain limits the strength of the chain, and 

 the Blodget map is the weakest link in Mr. Harrington's chain of 

 evidence. With it his conclusions must stand or fall, and, as has 

 been shown, this link is most absurdly weak. 



Let us look at the matter from another point of view. If Mr. 

 Harrington's conclusions regarding the rainfall on the plains, 

 drawn from a comparison of these two maps, are correct, similar 

 conclusions regarding the rainfall of other parts of the country must 

 likewise be correct, especially as the data upon which the map is 

 based are elsewhere more abundant, and the maps correspondingly 

 more reliable. Let us see what other changes are shown by the 

 maps to have occurred. About Cape Hatteras the rainfall has 

 apparently increased from 48 to over 70 inches ; in southern Lou- 

 isiana, from 45 to 60; in northern Florida, from 50 to 65; in 

 the mountains of North Carolina, from 36 to 48 ; and so on. 

 It is unnecessary to specify further changes, as there is scarcely 

 any part of the country in which, if this method of reasoning be 

 correct, great changes in rainfall have not occurred between 1857 

 and 1884. 



The method employed in my investigation of this question, and 

 the results obtained, are set forth in an article in Science for Jan. 

 6, 1 888. The explanation there given seemed to me to be suffi- 

 ciently clear for the average reader. It appears, however, that it 

 admits of being misunderstood, and has been misuriderstood by 

 Mr. Fernow. I will therefore state it once more, and with greater 

 fulness. The method used is a wholesale one. Certain areas in 

 this country, of great extent, in which the changes in respect to 

 forest-covering have, within recent years, been radical, were se- 

 lected, and an examination was made of the rainfall measurements 

 in these areas during the time of foresting or deforesting, in the 

 hope, not of obtaining a quantitative expression for the influence of 

 forests, but of learning whether they, have appreciable influence. 

 One of the areas selected was the prairie region, where it is well 

 known that during the past fifty years the wooded areas have 

 greatly increased, — so greatly as to change the whole aspect of the 

 country. This increase of woodland has been a progressive one, 

 going on gradually year after year. Now, if increase of woodland 

 increases the rainfall, it follows necessarily, that, barring its spo- 

 radic fluctuations, the rainfall also has increased progressively in 

 this region. The following, then, is the proposition to be proved 

 or disproved by the rainfall records : that the rainfall has in- 

 creased, and that progressively, in the prairie region during the past 

 fifty years, as foresting has gone on. Within this region I had ac- 

 cess to the records of twenty-four stations scattered widely over the 

 area, each station having a series of records of considerable length, 

 ranging from ten to forty years. These series are scattered over 

 the past fifty years in an irregular manner, and no attention was 

 paid to the particular years which each series embraces, as it 

 is not believed that it is a matter of any importance. The series 

 from each station was cut in halves, and each half added, giving 

 the total rainfall of each half. Now, were there no sporadic 

 fluctuations, — in other words, were the rainfall regular in amount, 

 — the comparison between the sums of the halves of each series 

 would be sufficient to base a conclusion upon. If the rainfall had 

 increased, the earlier half series would be less than the later half. 

 As a matter of fact, however, these individual results are very dis- 

 cordant, owing to the irregularities of rainfall ; and it is necessary. 



