that the use of the year beginning November 1 at Flagstaff and September 1 at Prescott gave the closest agreement between growth and rainfall. At Flagstaff the majority of the trees came from a thin clay soil derived in place from decomposed lava, and so there was little depth for the storage of moisture. At Prescott the sections of group 5, shown in the solid line of figure 7, came from trees growing in a porous soil of decomposed granite in a rather flat depression with reaarded drainage, so that conservation would have a greater influence. Perhaps this explains why the year beginning September 1 gives the best results.
In the region of the great sequoias nearly all the precipitation in the mountains (and quite all in the valleys where comparative rain records are found) comes in the winter months. For these trees, therefore, the winter precipitation is compared with the growth for the succeeding year and the month of beginning annual means is in the autumn.
THE TIME OF YEAR OF RING FORMATION.
Among the problems connected with the relation of the growth of trees and the amount of rainfall, one of the most interesting was suggested by Director R. H. Forbes, formerly of the Arizona Experiment Station. This was to determine the time of formation of the red or autumn portion of the rings and the causes for the formation of double rings, which were very numerous in the Prescott group. It seems evident at once that the growth of red cells is related to the decreased absorption of moisture as winter approaches. A number of tests were made on the Prescott group. The first was designed to determine the character of the rainfall in the years producing double rings. The half-dozen most persistent cases were selected and in each of these the red ring was found double in the following number of cases: 4 out of 10 in 1896; 5 out of 10 in 1891; 7 out of 10 in 1881; 4 out of 10 in 1878, 1872, and 1871. The average width of all the rings was 1.55 mm. The mean rainfall by months for the years above selected was found and is plotted in the solid line of the upper diagram of figure 1. Six other rings showing one double in 10 trees in 1898, but no doubles in 1897, 1885, 1884, 1876, and 1874, and averaging 1.54 mm. in thickness, were then selected and the curve of rainfall by months for the year during which they grew has been plotted as the upper dotted line in figure 1. In each curve the 6 months preceding and the 2 months following the year are included. The curves seem to indicate clearly that the chief cause of doubling is a deficiency of snowfall in the winter months, December to March. This appears to mean that if the winter precipitation is sufficient to bridge over the usual spring drought, the growth continues through the season, giving a large single ring which ends only in the usual red growth as the severity of winter comes on. If, however, the preceding winter precipitation has not been entirely adequate, the spring drought taxes the resources of the tree and some red tissue is formed because of deficient absorption in the early summer before the rains begin. When these rains come the tree continues its growth.