Significance of cycles.—It has already been stated that three characteristics were observed in the curves of tree-growth: (1) correlation with rainfall; (2) correlation with sunspots; (3) general periodic variation. In the first and second of these the trees are compared directly with existing records, but in the third the tree record is available over hundreds and even thousands of years during which no human observations were recorded. Thus, if previous inferences are correct, the trees may reasonably be expected to give us some knowledge of prehistoric conditions. In the first attempt to secure such knowledge, the method which promises the most certain results is the analysis of ring variations in terms of cycles.
Correlatively, the study of cycles is of special value in climatic investigations. Such studies are undertaken for the purpose of predicting the future. The basis of daily or short-distance prediction is found in the conditions existing about the country at a given moment and a knowledge of the usual movement of storm areas. A basis for long-distance prediction is now generally sought in climatic cycles. Such cycles may or may not be permanent. Perhaps they are nothing more enduring than a series of wave systems on a water surface. Yet for the navigator a knowledge of the existing system is important, and so for the purpose of weather prediction we need to know the nature of the pulsations actually operating, and each one should be studied minutely. For this purpose the very long tree records and their presumably fair accuracy seem especially advantageous, since they give us a range in centuries which the meteorological records, with few exceptions, give only in decades.
A special and rapid method of carrying on the study of cycles has been developed in the periodograph which has been used in checking fully all the results in the present chapter. But after its recent completion and trial the fact became clear to the writer that its real service will be in a complete and thorough examination of all curves obtained, in order to derive a quantitative statement of the extension in time and space shown by each cycle. This in itself is a long process. Moreover, preliminary analysis of many tree curves reveals a very complex system of short-period variations in the trees, some of evident significance and some of little-known value as yet. The study of this complex of short periods together with other problems naturally suggested in the course of the work is reserved for the future; we shall now touch upon a few of the most important results reached in the analyses already accomplished.
Predominant cycles.—With the understanding that the study of cycles is not yet complete, it may be stated at once that the more
