far as the historian goes. They part company, however, when it comes to the question of why the invaders came into Egypt and were thus given the necessary wealth, leisure, and other opportunities which enabled them to develop their talents and become great. The geographer who believes in pulsatory changes of climate can scarcely avoid the conclusion that great movements of peoples have been induced by such changes, and that these movements have given rise to periods of invasion and anarchy. Furthermore, he is led to conclude that when the stress due to unfavorable climatic conditions has been removed by reason of another change, this time in the direction of more favorable climatic conditions, prosperity and progress have been the rule. This by no means implies that all invasions and all prosperity are supposed to be due to climatic causes, but merely that climate has been one of the important factors in producing such results. I do not propose to discuss this question here, as I have already considered it in the publications referred to above, especially in the later chapters of The Pulse of Asia and Palestine and its Transformation. I wish, however, to concentrate attention upon the question which, at the present stage of investigations, is the crux of the whole matter. If Professor Olmstead is a fair representative of the younger school of modern historians, the views of that school would coincide with those of the geographers, provided there were certainty upon one point, namely, the verity of our conclusions as to climatic pulsations. After devoting some pages to a statement of reasons for not believing that such pulsations have taken place, Professor Olmstead concludes :

We have not the space to further test by historical facts the theory that the Arabian desert with its surrounding lands [was once] more occupied, more fertile, and easier of access than it is at the present day. Further examples would only prove that it was not well grounded. And this brings us to our conclusion as regards the question of the relation of climate to history. That climate, working through the ages, has a highly important effect on the permanent population of a country is admitted by every historian. That it has effects, mostly negative, on the transient population has also been seen. At present, the theory of a more immediate influence on the details of history seems to be bound up with the theory of cyclic [pulsatory] climate changes and we have. seen that the facts of history tend to disprove this. Accordingly, the historian is not justified in utilizing climate for more than the study of the background of his history. For influence on particular events, there are many geographical facts of far more significance.

The question before us divides itself into two parts. In the first place, was the climate of the past, let us say at the time of Christ, different from that of the present? In the second place,

assuming that there has been a change, did it take place gradually or was it characterized by pulsations whereby certain periods were exceptionally dry while others were moist? The type of evidence to be employed is the same in both cases, and consists first of physiographic phenomena among which river terraces, lake strands, denuded mountain slopes, desiccated springs, and rivers whose salinity has increased, are of special importance. A second highly important type of evidence consists of archaeological phenomena, such as the location of ruins like those of Palmyra or Ilandarin. Here, in the past, great cities grew up in places whose supply of water is now not one-tenth large enough for the support of such a population as once existed. Still a third line of evidence is based upon plant life, forests, areas of cultivation where crops cannot now be grown, and the like. Finally, with all this must be joined direct historic evidence, such as accounts of famines, recorded facts as to the supply of water in places now dry, old roads across deserts which to-day are impassable, and a vast number of other matters which have never been properly scrutinized because historians have not investigated the subject.

In all these cases it is far easier to find and interpret evidence in reference to the first of our questions than to the second; for the discovery that regions which once were well populated are now uninhabitable is a comparatively simple matter, while only the most careful research reveals the reasons for believing that while the past as a whole was distinctly moister than the present, certain periods were notably drier. This would seem to indicate that if some new method of investigation is to be tried, the study of possible fluctuations is more important than that of possible differences between the past and the present. If the fluctuations should prove to have taken place as inferred from the other lines of evidence, there would be little question that the climate of the past, as inferred from those same lines of evidence, was in general different from that of the present. Hence in this article I wish to present a new type of evidence which seems to go far toward proving conclusively. that the pulsatory theory of climatic changes is correct. By this I do not mean to imply that all the details of the climatic curves which are shortly to be presented are as yet established beyond question. I merely mean that the evidence seems to indicate that pulsations of climate lasting through periods having a length of centuries have actually taken place. This, it will be seen, is in direct opposition to the statement of Professor Olmstead. "At present", to repeat a sentence already quoted, "the theory of a more immedi

ate influence on the details of history seems to be bound up with the theory of cyclic [pulsatory] climate changes and we have seen that the facts of history tend to disprove this."

་

The question cannot be settled offhand by a reference to "the facts of history". Long research in the realms of physiography, climatology, archaeology, and, as I shall shortly point out, botany, can alone determine it. In other words the problem is primarily geographical, in the modern sense of that term, and the final decision of geographers must be accepted by historians. When it comes to the study of the effect of any possible climatic changes upon the course of history, however, the case is reversed; the geographer may offer suggestions, but the final decision rests with the historian. Hence the purpose of this article is to show the grounds upon which an increasing number of geographers are becoming convinced that changes of climate have actually taken place, and then to suggest certain ways in which these changes may have been of historic importance. I realize fully that in making these suggestions a geographer is liable to error, for his view of history must of necessity be limited. Therefore in no case would I be understood as asserting categorically that such and such results have occurred because of climatic changes, but merely that certain results appear probable from the point of view of the geographer. If the changes here discussed have actually taken place, they must have had some effect upon history, and it is only by discussion of the question from both the historical and geographical sides that the truth can be learned.

Lack of space forbids any discussion of the evidence of changes of climate in Asia, and I must once more refer the reader to The Pulse of Asia and the other publications already named for a statement of the results of three expeditions to Asia during which about three years were spent in the Turkish Empire, Persia, India, the southern portion of Asiatic Russia, and the western part of China. These expeditions, extending over the period from 1903 to 1909, led me to formulate the theory of pulsatory climatic changes. The evidence which was first found indicated only the greatest pulsations, but as time went on the number was seen to be larger, or rather the details of minor pulsations became more clear. At best, however, the resultant climatic curve was no more than an approximation to the truth. Some definite, mathematical method of measuring rainfall or other climatic factors was necessary. In order to test the theory as widely as possible I accepted the invitation of Dr. D. T. MacDougal of the Department of Botanical Research of the Carnegie Institution of Washington to co-operate with the Desert

Botanical Laboratory at Tucson, Arizona, in a study of the climate of the arid portions of the United States. Two seasons of field work among the dry lakes, terraced valleys, and innumerable ruins. of Arizona, New Mexico, and the neighboring parts of Mexico, supplemented by a journey to southern Mexico and Yucatan, led to the conclusion that the climate of America has been subject to pulsations similar to those which appear to have taken place in Asia. I have discussed the matter in articles appearing in Harper's Magazine during the years 1911 and 1912, and in a series of articles in the Geographical Journal of London, and shall not here attempt to say more about it. The lines of evidence were similar to those followed in Asia and Greece, that is, they were primarily physiographic and archaeological, with the addition of historic evidence. wherever possible. They will be fully discussed in a volume shortly to be published by the Carnegie Institution of Washington under the title, "The Climatic Factor".

Leaving, now, these more purely geographical lines of research, let us turn to another type of evidence which seems to add to the conclusions already reached the final touch of mathematical accuracy which alone can lead to certainty. Realizing that my work in America was liable to error because of the danger of being influenced by a preconceived theory, I made use of a method suggested by Professor A. E. Douglass of the University of Arizona.3 Professor Douglass found that the thickness of the rings of annual growth in the old trees of the forests on the plateaus of Arizona is proportional to the amount of rainfall. If the average growth in diameter of a large number of trees be plotted for year after year, the ups and downs of the curve thus formed agree in general with the ups and downs of the curve of annual rainfall plotted in the same way.

-Evidently, then, the annual rings of old trees preserve a record of the rainfall in past times, and it is only necessary to read this record to answer the question of the reality of climatic pulsations extending over hundreds of years. Before great accuracy can be obtained it is necessary to eliminate the effects of variations in the growth of trees because of differences in age. Young trees grow faster than old, but there is a regular law for this, as is well known to foresters, and it is purely a matter of mathematics to apply the necessary corrections. Accidents such as fires or storms also affect the rate of growth, but, as I have shown in the articles already mentioned and in another which will soon be published in the American Journal of Science, these become negligible when a large

A. E. Douglass, "Weather Cycles in the Growth of Big Trees", Weather Review, XXXVII. 225-237 (1909).

number of trees from different localities are employed. Thus, after all corrections and allowances have been made, we are able to secure a curve which represents with considerable accuracy the fluctuations of climate in past times. The process by which this is obtained is purely mathematical, and no amount of theorizing on the part of the investigator can affect the results.

During the years of 1911 and 1912 I measured the rings of four hundred and fifty of the Big Trees, or sequoia gigantea, of California, which, fortunately for the purposes of historical research, had been cut in order to make fence posts and shingles. These trees grow in the Sierra Nevada Mountains under climatic conditions closely similar to those of the high plateaus of Arizona. The winter is snowy and rain falls during the spring until May or June, but the rest of the summer is absolutely dry. The forests need more rain than they commonly get. In years when the amount of winter snow is larger than normal, or when the storms of spring persist well into the summer, the trees grow much faster than usual. The trees which I measured ranged from 230 to 3200 years of age. Eighty began to grow more than two thousand years ago, and three were more than three thousand years of age.

From these four hundred and fifty trees I have constructed the curve shown by the solid line in the accompanying diagram. The

1300

1000 800 600 400 200 B.C. A.D. 200 400 600 800 1000 1200 1400 1600

1800

course of time is represented by horizontal distance, the left end indicating the date 1300 B. C., and the right end 1900 A. D. High portions of the curve denote moist conditions, which would be highly favorable in countries like Syria, Egypt, and Greece, but detrimental in such countries as Germany or England. Low places, on the contrary, indicate relative aridity, which would be disastrous in lands such as Palestine. The details of the curve will be modified somewhat when a larger body of data is available. For instance the violent zigzags of the earlier portions, where the number of trees is small, will undoubtedly be reduced. The general form of the curve, however, will in all probability remain as here indicated, although previous to about 200 B. C. the fluctuations will be less sharp and the extremes of the peaks and depressions will not rise so high nor fall so low as is here indicated.