150 



CLIMATIC CYCLES AND TREE GROWTH 



regular standard cycleplots, any scale error becoming immediately evident; 

 the instrument is then adjusted, or if the error is small it is applied as a 

 correction to the results. 



Since the scale is calibrated in tenths of units, cycle values are at best 

 correct only to the nearest tenth unit. In rare cases only, of outstanding 

 cycles or in special study, is interpolation to hundredths of a unit made. 



ERRORS IN CYCLOGRAM ANALYSIS 



The matter of precision in assigning cycle lengths has received much 

 attention. The material used consisted on the one hand of a large number 

 of cycle analyses of the 42 Western Groups carried through by Dr. Douglass 

 in 1926-28, and on the other hand of a considerable series on the same 42 

 Groups and also on the California Coast Redwoods made during 1933-34 

 by the writer. An intensive study of this material over a period of some 

 months resulted in a set of statistical "constants" and is summarized in 

 table 1. 



Table 1 — Variations in cycle settings 



1. Standard procedure in cy olograph use demands an analysis over the 

 entire range not only when the mirrors are moved out, but also when they are 

 brought in. As far as possible, the second analysis is made independently 

 of the first. Some remembrance of the pattern may, however, influence 

 the assignment of a horizontal setting the second time, in many cases, and 

 for that reason differences were expected to be small. 



As in the other cases discussed below, differences are chiefly due to 

 confusion in the pattern because of weak cycles which make one setting 

 hardly better than several others in the immediate vicinity. Sometimes only 

 one or two dots (crests) will determine a significant alternative setting. When 

 the pattern contains a strong cycle, alternative settings sometimes depend on 

 the accepted duration of the cycle. Ambiguity arises in part also from the 



