ANALYSIS OF TERRESTRIAL RECORDS 85 



Similarities over Areas — In establishing the chronological identity of rings 

 in different trees, we are at the same time proving the climatic similarity 

 over the terrain involved. Climatic similarities, however, must be supported 

 not merely by a corresponding deficient ring here and there but by the aver- 

 age degree of resemblance in all rings used. In addition to the closeness of 

 resemblance, we are interested in the size of the area over which it exists. 

 Such areas of similarity in ring growth constitute meteorological districts. 

 Experience has shown that when cross-identity is perfectly evident within a 

 group of trees, we can be confident that the correlation coefficients between , 

 each one and the others are high. A mutual correlation coefficient of 0.85 ± 

 0.02 was found in five Prescott trees having hundred-year records (see fig. 2). 

 Climatic similarities have been tested also by similarities in cycles. (See 

 Reports of Conferences, 1929.) Similarity in three cyclograms, shown in 

 1928 (vol. II, Plate 9: 1, 2, and 3), joins Flagstaff to the areas near Aztec, New 

 Mexico, 225 miles away. The trees of these regions we already knew had 

 strong cross-identity. In the Flagstaff area the cycle similarities were suffi- 

 ciently pronounced to date correctly unknown fragments, 125 years long, of 

 records in local trees, in three cases out of four by cycles only. Such dating 

 depended on a well-defined distribution of certain cycles in succession. In 

 trying to date unknown fragments, 200 years long from Flagstaff, by com- 

 parison with 2000 years of sequoia records, 50 per cent were dated correctly 

 by cycles. This 50 per cent is far higher than would be the case in random 

 results. It would be conservative to say that the chance of dating in random 

 sets would have been not over 5 or 10 per cent. 



Now that we have extended the Arizona ring sequence back to A.D. 11, 

 it has become possible to match one locality against another 600 miles away 

 through many centuries, and we perceive resemblances which will be described 

 below. 



Different sub-groups of the coast redwood have been compared by periodo- 

 grams which show evident similarity and a correlation coefficient of 0.56 ± 

 0.08; sub-groups of the Central Pueblo Area compared in this way gave a 

 correlation coefficient of 0.76 ± 0.05; random data compared in similar man- 

 ner show no correlation (0.02 ± 0.12). 



Recurrence of Cycles in Past Time — Fully dated tree-ring records have 

 been extended, in California sequoias, 3200 years; and in Colorado Plateau 

 pines and firs, 1900 years. These will be discussed later in the chapter. 



Periodogram Resemblance to Solar Cycles — The last item to be listed here 

 forms the subject of the following chapter. 



PROBLEMS 



Representation of Cycles — Cyclogram analysis was first used in 1913 by the 

 multiple plot and in 1914 and 1915 by cyclograms. Since then its use has 

 been continuous. Until 1918 the best method of representing a group of 

 cycles was supposed to be the periodogram. In the reconstruction of the 

 analyzing instrument in 1918 the automatic periodogram was the goal. 

 Photographs which realized that end were made and published in Volume I. 



