118 CLIMATIC CYCLES AND TREE-GROWTH 



taining the cycle, and twice drawing them out by lot; thus producing 

 three curves, of which one is genuine and two spurious. If the genuine 

 one can be distinguished from the others by the cycles alone, without 

 other marks of identity, then the cycles are there. We can hardly 

 yet make application of this to rainfall or tree-growth curves, because 

 we do not know (or are just learning) what cycles ought to be there. 

 On this account a half dozen criterion tests have resolved themselves 

 largely into solving the question of the existence of cycles over 20 

 years, for that was the only known mark of identity. That in turn 

 depended vitally on the length of the curve under test, for a cycle 

 does not carry conviction unless it is repeated five or ten times in the 

 record. So the trials on short curves of 50 or 75 years were not suc- 

 cessful, while those on curves of 200 years were. It is probable that 

 there will be extended use for this criterion, but in the absence of 

 better knowledge of the cycles to be expected it has not been thoroughly 

 tried and another method of judging reliability has been applied, 

 namely, identifying similar cycles in many trees and over wide areas. 



Cycle identification in small areas — In the early use of the cyclo- 

 graph it became a matter of interest to know whether cross-identifica- 

 tion could be done by cycles. To test this, an early general curve of 

 the Flagstaff region was prepared as a standard. An assistant selected 

 125-year portions of other Flagstaff trees without letting me know 

 to what tree or to what part of the 500 years they belonged. By 

 cycles alone each unknown was compared with the standard 500-year 

 curve. In the first trial of 10 unknowns, 7 were dated correctly, and 

 in the next trial of 10, 8 were dated correctly. In other words, the 

 cycles in any given tree in the region specified bear 75 per cent resem- 

 blance to a good average cyclogram of that region. Dating by size of 

 individual rings is considered to have a reliability of 95 per cent or 

 more. This decreased reliance in cycles is due in part to over- 

 importance given in those tests to short-period cycles, before their 

 unreliability was recognized. 



Cycle identification at 200 miles — Two groups of 8 or 10 trees each, 

 one from 40 miles north of Aztec, New Mexico (BMH), and the other 

 from 18 miles east (AE), were compared with the Flagstaff records. 

 The resemblance in the cycles is extremely close. Periods of 14, 17, 

 and 21 years appear in all three groups in practically identical form. 

 In this comparison cross-identification by cycles was carried over 225 

 miles of country (see Fig. 19 and Plate 9, page 132). 



Cycle identification between Arizona and California — A still more 

 difficult test was made between the Flagstaff area and the big-tree 

 area. A selection of California trees was made in the following manner : 

 The last 500 years of each of 34 trees were plotted and the resemblance 

 of the cycles to Arizona and New Mexico cycles was reviewed and 



