i6 



SMITHSONIAN MISCELLANEOUS COLLECTIONS VOL. I28 



We now come to a fourth line of evidence even more convincing. 

 It is held that the large family of periodicities are all, to within i 

 percent, aliquot parts of 22| years. If so, periods of ^, ^, and ^2 of 

 22f years, and similar groups of periods of various lengths, are also 

 related by integral numbers to each other and to 22f years. Accord- 



FiG. 8. — The iiA-month periodicity in St. Louis precipitation as averaged 

 from six independent determinations. A, Wolf numbers above 20; B, Wolf 

 numbers below^ 20. 



ingly, when a tabulation is made to determine the form and range of 

 one of the longer periodicities, we may often find several shorter 

 periods, harmonics of it, riding on the long curve. If these shorter 

 periods are not removed, the longer ones are often unrecognizable. 

 This is the case. I shall give several illustrations of it. The first is 

 from the period of 68^ months, shown in figure 9. A and B are 



