NO. 4 PERIODIC SOLAR VARIATION — ABBOT 3 



for weather features, the ranges in precipitation found for these same 

 individual periods run from 5 to 25 percent. Ranges in temperature 

 for individual periods sometimes reach over 2° Fahrenheit. 



In an earlier paper, "Periodicities in the Solar-Constant Measures" 

 (Smithsonian Misc. Coll., vol. 117, No. 10, 1952), I used nearly all 

 of the observed solar-constant values, from August 1920 to Decem- 

 ber 1950. This interval included the unique depression of the solar 

 constant of about 5 percent, which was found in 1922 and 1923. It 

 also included the widely varying values found from 1920 to 1922. 

 In this present research I desired to obtain the highest attainable 

 accuracy. Both in length and range I aimed to determine the best 

 average values of periodic variations in solar radiation. I therefore 

 purposely omitted those exceptional years, and began with Sep- 

 tember 1923, ending with December 1952. 



There is a weightier reason than this improved selection of data 

 for revising my former study of solar radiation. I have become more 

 and more convinced, by my researches on precipitation and tempera- 

 ture, that a simple tabulation of a suspected period, in a long series 

 of monthly or lo-day values, is unconvincing as to the reality of peri- 

 odicity. For there is a large family of periods that have been found 

 in various phenomena, all members of which are aliquot parts of 22| 

 years. Not only does a long succession of monthly or lo-day means 

 contain numerous members of this family, but the individual mem- 

 bers, if segregated merely by simple tabulations, each contains, in its 

 mean expressions, several shorter periods which are aliquot parts of 

 the period being investigated. 



Unless these shorter periods are removed, the mean expression of 

 the desired period is so marred in form as to be unconvincing. No 

 elimination of these overriding shorter periods, aliquot parts of the 

 periods determined, Avas done in the publication just cited. 



In the present publication I shall disclose a considerable number 

 of new members of the family, found overriding known periods, and 

 all integrally related to 22| years. Many of these new periods are so 

 short as to be better expressed in days than in months. They include 

 some of the periods discovered in basal pulse rates by my friend 

 Dr. F. P. Marshall. 



After a short statement regarding the solar-constant data, speci- 

 mens of the treatment of it will follow, and after that the conclusions 

 reached will be given in tables and graphs. 



In my former publication (cited above), I gave, in table 4 in the 

 appendix, a list of solar-constant values of the years 1920-1950. For 

 convenience of tabulation these values were given in the form of 



