V. ANALYSIS OF TERRESTRIAL RECORDS 



Multitudes of attempts have shown that here and there seemingly pur- 

 poseless climatic variations gather themselves into orderly form and give 

 cycle effects of limited duration. The obstacle which climatologists have 

 everywhere encountered is the short length of the meteorological records; 

 they are far too short to permit a general study of climatic cycles. This 

 difficulty is now being overcome by the use of tree-ring records having a strong 

 climatic correlation. We have a pyramid of ring records consisting of hun- 

 dreds of trees that cover the last century diminishing to four that cover 3000 

 years. For all this material we have a flexible and rapid form of cycle 

 analysis. 



In this study of climatic variations we seek answers to many questions: 

 What are the cycles that do appear; when do they come; how long do they 

 last; are they of small or large amplitude; do they come over large geo- 

 graphical areas at the same time or only by the acre, so to speak; have they 

 prevailed at other times in the last 1900 years; do they resemble cycles in 

 sunspot numbers or other solar phenomena and how can observations of 

 the sun contribute to the problem ; can they be used in long-range prediction? 



These questions state the field to which studies in the succeeding pages 

 will be devoted. Attention will be given to the relation between solar and 

 climatic changes and a final chapter will take up the practical summary of 

 this information in an attempt to point out its most important parts contribut- 

 ing to long-range prediction. The remainder of this chapter takes up in a 

 series of topics the most distinguishing characters and problems in ring growth 

 and climatic cycles that have emerged from many considerations of the sub- 

 ject and especially from the use of the cyclogram method of analysis, all 

 introductory to the lists of cycles found. 



CHARACTERS OF CYCLES IN RING RECORDS 



Discontinuous Periods — Sometimes called broken periods, fragmentary or 

 short-lived periods; sometimes called cycles when their variability is referred 

 to. The word cycle is used as a more general term. Reference is made to 

 their special discussion in Chapter III. 



Conservation — Persistence of similar values in two or more successive years 

 resulting in less frequent crossing of the mean in natural sequences than in 

 random sequences, is classed statistically as conservation and should be 

 examined under that heading. When three general curves of western tree 

 growth were examined by Dr. Dinsmore Alter for lag correlation expressed 

 in his correlation periodogram, high correlations were found in short lags of 

 1, 2, and 3 years. He at once recognized this conservation and asked whether 

 I had otherwise noticed it. I had used a form of it in testing relation of 



82 



