The original means of 11 trees are given in the upper line of figure 27. In the middle line these are smoothed by Hann's formula and in the lower line is the sunspot curve, displaced three years to the left in the portions since 1810 and one year in the same direction before that date. The tree crests anticipated the solar crests by three years when the trees were large and making good growth, but when small this anticipa-

Fig. 29.—Correlation curves of solar cycle, rainfall, and tree-growth at Windsor, Vermont, 1835-1912.

tion of the sunspot maximum was considerably less. A correlation is evident, but it is hard to give a satisfactory explanation of the phase displacement. Figure 28 gives details of the time relation between tree-growth, rainfall, and solar activity. In figure 29 the curves of figure 28 have been summated on an 11.4-year period, as was done in figure 25. At the bottom is the sunspot curve from 1834 to 1912 inclusive; directly above it is the curve of rainfall for the vicinity of Windsor, compiled chiefly from records at Hanover and Concord and covering 1835 to 1912; above that is the tree-growth from 1834 to 1912, and in the upper line the sunspot curve is repeated with a displacement of — 3 years.