ANALYSIS OF TERRESTRIAL RECORDS 



91 



a real two-year cycle analysis is a conspicuous weakness in any prediction 

 plan. 



Whether we do or do not identify the exact cycles involved in this two- 

 year variation, we must at least examine the probabilities in the case to see 

 if we can make use of them in the meantime. We can bring together some 

 statistical data on this point through studies of conservation in harmonic 

 analysis as in Dr. Alter's correlation periodogram. In working over some 

 ring records for me, he found high correlations between each value and the 

 next following. At first thought that seems not to agree with the idea of 

 the "zigzag" or two-year reversing cycle. So I have made use of scatter 



1810 20 30 40 1850 60 70 80 90 1900 10 20/^ 30 



Random data. LCFD 

 Fig. 38 — Two-year reversal tests on random data. 



diagrams to bring out the relationship between each annual value and the 

 next following. 



Figure 39 is a scatter diagram to show the relation between successive 

 values, whether quantities above a constant base or departures from a con- 

 stant mean. Here we have an obvious positive correlation in the evident 

 oval of relationship extending up to the right at 45° inclination. If, however, 

 we take the departures from a Hanned curve, such as we use in cycle analysis 

 — in short, if we take the plus and minus values discarded when we smooth 

 the curve for analysis — then we get figure 40a, which has an oval cloud of 

 dots whose major axis extends up to the left at 45° or at right angles to the 



