APPENDIX 



151 



placing of emphasis on different portions of the cycle at different times. In 

 many cases, two distinctly alternative settings may be made, only a portion 

 of the individual dots being common to each. The average difference of these 

 alternative settings was found in 31 cases to be about 5 or 6 per cent. 



2. The second kind of difference is that resulting when the same material is 

 subjected to analysis at two different times by the same observer. To make 

 the test as rigorous as possible, the original analysis was compared with a 

 second set in which the same data had been replotted on an unknown scale. 

 Extensive analyses of this type in the 42 Western Groups by Dr. Douglass 

 and in the California Coast Redwoods by the writer were available. After 

 applying the proper correction factors to the second set, the average differ- 

 ence when using the maxima was almost exactly the same as that when 

 the plots were inverted and the minima were used. The total number of 

 cycle pairs compared was 694. An average difference of 2.5 per cent, the 

 largest found, is not very great when we consider that we are dealing with 

 highly variable phenomena containing fluctuations due to some minor ex- 

 tent at least to random causes. 



Table 2 — Anomalous cycles 



3. The third kind of difference considered, that between two observers, is 

 the most important. The only available data that could be used were in 

 the 42 Groups; the analyses had been made some eight years apart and were, 

 of course, quite independent. The average difference of 3 per cent is grati- 

 fyingly small; in this value there are included cycles, the identifications of 

 some of which, as corresponding to the ones paired with, were highly doubt- 

 ful. When only the strong cycles were used, the average difference is no 

 larger than the difference of the second kind (see number 2 above). 



4. In ordinary procedure the maxima of a curve are isolated for analysis. 

 What happens when the minima are isolated instead? Comparison of 

 analyses using maxima with those using minima shows that the average dif- 

 ference is of the same order as before. The total of 407 pairs of cycles is 

 certainly large enough so that the result may be viewed with confidence. 

 No apparent change was found in the number of emphasized cycles. There 

 was also no evidence for any systematic increase or decrease in cycle length 

 on inversion. 



