process was recognized in 1911, the 19 sections of which samples had been preserved were compared and a complete cross-identification carried through. Thus the errors of identity in the former tabulation were found (published in 1914) and a complete new set of tables and averages made from the original measures. For a time it was thought that an error of one year might exist in the period of the great drought of 1820-23, but the various checks made upon identity lead easily to the belief that there are no errors of identity in this 500-year series.
Subgroups.—The trees of this group were divided into three subgroups consisting of (1) 6 trees from 3 miles south of Flagstaff; (2) 9 trees from 11 miles southwest of Flagstaff; (3) 10 trees from a point 1 mile west of the last subgroup. A comparison of the 3 subgroups clearly reveals the general character of the longer periods hereafter to be discussed and shows lesser variations to be common to all. Interesting differences, depending on the location in which the trees grew, have been mentioned.First suggestion of cross-identification.—Other interesting facts came to light. It was especially noticeable that a given year of marked peculiarity could be identified in different trees with surprising ease. This is illustrated in plate 2, where shavings from 5 of the Flagstaff trees have been photographed; the photographs have been enlarged to such a scale that the distance from the large ring 1898 (indicated by the upper line of black crosses) to the small ring 1851 at the lower line. of crosses is equal in all cases. The other lines of crosses indicate the noticeably broad rings of 1868 and 1878. An examination of the photographs shows that a very characteristic feature is a group of narrow rings about the years 1879 to 1884. These can be identified in practically every tree and an examination of many stumps which were not measured showed that it was easy to pick them out wherever one chose. Striking verification of this was found in the case of a stump near town which had been cut about 20 years previously. By finding this group of rings, the writer was able to name the year when the tree was felled and the date was verified by the owner of the land. In the more recent work this same group shows conspicuously among Prescott trees, and in general 95 per cent of these trees have rings so characteristically marked that the identification of the same series of rings can be made with little doubt, whether at Flagstaff or at Prescott.
- The Flagstaff 500-year record.—Figure 3 shows the Flagstaff tree record from 1385 to 1906 A. D., a period of 522 years. The table of measures from which the curve was plotted will be found on page 112. To give the record from 1503 to 1906, 5 trees are used, and comparisons showed that these 5 gave as accurate a record as a larger number whose inclusion would have shortened the record or made awkward breaks in it. The earlier part of the record is from 2 trees