8 



SMITHSONIAN MISCELLANEOUS COLLECTIONS VOL. Ill 



It was now apparent from the behavior of the yearly variation of 

 the solar constant, the excellence of the 39^-, 91 -, and 68-month 

 periodic curves, and the superposition of curves of 91/6 and 68/7 

 months, as noted above, that it is quite justified to regard 273 months 

 as a master cycle in solar variation, and that many periodicities, 

 nearly or exactly integral submultiples of 273 months, exist simul- 

 taneously therein. In all my subsequent search for periodicities in 

 solar variation, as displayed in Montezuma solar-constant values, I 

 accepted the 273-month master period, and sought for integral sub- 

 multiples of it. 



Fig. 2. — Yearly march of solar constant, 1924-1947. 



Proceeding by the methods explained above, the periodicity of 54? 

 months was next sought, found, and determined. Its ampUtude is 

 0.0020 calorie, its form, like that of 91 months, comprises a rapid 

 rise and slow fall. The curve, though smoothed by 5-month running 

 means, has excrescences indicating the encroachment of a period ap- 

 proximating 8 months. Study of it was postponed, like those found 

 with the 91- and 68-month periodicities, for later determination. 



Attempts were then made to determine periodicities of 45^, 34, 

 and 3O3 months. But these proved so far dominated and obscured 

 by variations of shorter periods that they were all passed over for 

 the time. However the curve drawn when seeking a periodicity of 

 30^ months clearly indicated a periodicity of half that length, of 

 fairly large amplitude. So the next search made concerned 15^ 

 months. It will be noted that solar variations of 273, 91, 68, and 

 54^ months period had now been extracted from the monthly data, 

 and that the fourth list of residuals was now being used. 



