6 SMITHSONIAN MISCELLANEOUS COLLECTIONS VOL. Ill 



to give the complete solar constant in calories per square centimeter 

 per minute. 



Figure i shows graphically in curve A the march of the monthly 

 mean values given in table i. Curve B, on the same scale, gives 

 departures from 1.945 calories remaining after 14 periodicities speci- 

 fied in table 2, below, have been removed from the original data 

 given in column 2, table i. 



Table 2 also gives the yearly mean values, and numbers of days 

 entering into them. It gives also smoothed-curve values derived from 

 these yearly data, after plotting them as shown in figure 2. In the 

 statistical search for periodic variations reported below, the smoothed- 

 curve yearly mean values of table 2 were first to be removed by sub- 

 traction from the original monthly means. In order to do this 

 the smoothed yearly means were first expanded graphically into a 

 plot of smoothed monthly means. I do not take space to publish these 

 smoothed monthly means, as their simple derivation will be easily 

 understood, and as it makes no appreciable errors in the periodicities, 

 to be given in table 2, whether these smoothed monthly means for 

 eliminating yearly changes of the solar constant are the best that 

 could be found or not; for these periodicities are found as means 

 from statistical tables including many repetitions of the periods, and 

 local errors are smoothed out. 



In previous analysis of solar-constant values ^ numerous periodici- 

 ties in solar variation were found to proceed simultaneously, all being 

 approximately Integral submultiples of 273 months in length. I did 

 not wish to adopt this master period of 273 months in this present 

 research without independently confirming it from Montezuma data 

 alone. Figure 2, however, itself seems to indicate that a period of 

 about this length would fit the yearly variations of the solar constant. 

 There are researches of other authors which support the validity of 

 a period approximating two ii-year sunspot cycles, as being in evi- 

 dence in various solar and terrestrial phenomena. Thus G. E. Hale 

 discovered that magnetism in sunspots reverses its polarity in a re- 

 markable way with each successive sunspot cycle of 11 years, so that 

 the sun's magnetic condition is restored only after two 11 -year cycles 

 pass, or about 22f years, A. E, Douglass has remarked a 23-year 

 period in tree-ring widths. Various meteorologists have found it in 

 terrestrial data. I myself pointed out that Wild's meteorological 

 studies of the Russian Empire, when supplemented by later data, 

 showed very clearly a 23-year cycle in weather at St. Petersburg. 



* Ann, Astrophys. Obs., vol. 6, p. 181, 1942; Smithsonian Misc. Coll., vol. 107, 

 Ne. 10, 1947. ■ 



