METHOD OF ESTIMATING RAINFALL BY GROWTH OF TREES. 103 



located military camps at various places, and from that time records of rainfall and tem- 

 peratm-e were kept. The record at Whipple Barracks, near Prescott, which was begun in 

 1867, has been continued to the present time. It is the longest consecutive record in the 

 pine forest, and is therefore made use of below. Aside from rainfall, other meteorological 

 elements, especially temperature, must have an effect upon tree-growth, but I have not 

 attempted to include them in this work, for it seemed desirable to ascertain the degree of 

 relationship of the growth of trees to one single element before going on to others. 

 Moreover, it is probable that the various climatic elements have so distinct a relation to 

 each other that the investigation of one will throw light on the rest. 



The plateau and the climate are not the only features of northern Arizona which favor 

 an investigation of the sort here contemplated. The yellow pine itself is favorable, because 

 of its conspicuous annual rings. The differences between the soft, rapidly growing white 

 tissues of the spring and summer and the hard, reddish layers formed in the fall are much 

 less conspicuous in most of the common trees than in the pine. The sharp, outer edge where 

 the growth of the hard, red layer is checked by the cold of winter gives a precise point 

 from which to measure. The chief growth of the tree consists of a wide, white, pulpy, 

 summer ring, whose cells are round and well-shaped. As conditions of growth become 

 less favorable, the cells become lean and emaciated and take on a red color. The autumn 

 ring thus formed is thin, hard, and pitchy. On the inner side it merges gradually into 

 the summer ring, but on the other side it is sharply limited by the spring growth of the 

 next year. Where a double ring is formed, the white portion of the secondary ring is 

 usually narrow and poorly developed. 



THE COLLECTION AND MEASUREMENT OF SECTIONS. 



At the beginning of the investigation it was foreseen that enough trees would have to 

 be measured to give a real average. The trees would have to spread over enough country 

 and be sufficiently numerous to eliminate accidents of grouping and other minutely local 

 conditions, and yet they must not extend into other meteorological regions; they must be 

 numerous enough to be susceptible of division into groups, which show common char- 

 acteristics and thus testify to the genuineness of whatever variations appeared. Work 

 was begun in January 1904, when I visited the log yards of The Ai-izona Lumber and 

 Timber Company, Flagstaff, and spent several hours in the snow, measuring the rings of 

 section No. 1. For subsequent measurements Mr. T. A. Riordan, president of the company, 

 most kindly came to my assistance by having thin sections cut from the ends of logs or 

 stumps and sent to me in town, there to be measm-ed more conveniently. Sections VII 

 to XXV were cut at my direction on the spot where the trees grew, and where I was able 

 to mark the points of the compass on the sections and otherwise identify and describe 

 their location. These 19 sections were freighted to Tucson, where the work on them was 

 done. The method of measurement consists in determining the radial thickness of each 

 annual ring in millimeters. The average age of the trees was 348 years. The total number 

 of individual measurements reached over 10,000. 



In the first comparisons between tree growth and rainfall the measures from six sections 

 only were used and comparison was made with the Prescott weather records, for the Flag- 

 staff station had been in existence only 6 years. At that time there was no thought of 

 any such remarkable relation between yearly growth and yearly rainfall as has since been 

 found; therefore, such relationship was not even tested until later. For purposes of com- 

 parison, smoothed curves were used, "the nine-year smoothed" being the one chiefly em- 

 ployed. Inasmuch as we were then attempting to study the general condition of the country 

 rather than the individual year, and as the influence of good or bad conditions of rainfall 

 lasts some years, the average of the eight preceding years and of the year in question was 

 plotted in place of the rainfall of any single year. From such smoothed curves a connection 



