NO. 3 SOLAR VARIATION AND WEATHER — ABBOT I5 



The smoothed curves representing the periods, as will appear from 

 figure 10, are nearly sine curves. Their regular shapes, of course, 

 make them stronger evidences of real variation than their individual 

 points. But these points, as just said, have probable errors ranging 

 from 1/60 to 1/200 of 1 percent of the solar constant. 



4. PERIODICITIES IN SOLAR VARIATION 



Over 25 years ago I noticed periodic repetitions of dentlike depres- 

 sions in graphs of the solar-constant record. These periods appeared 

 to be harmonics of 273 months. Notably appeared 273/7, 273/21, 

 273/39. In my paper "Periodic Solar Variations" of June 1955 

 (P. 4213) I amplified what I had said earlier (P. 4088) in May 1952. 

 For convenient tabulation I gave in the two papers together the 

 departures from 1.90 calories of all mean monthly solar constants 

 from August 1920 to December 1952. These departures are ex- 

 pressed as percentages of the solar constant. Thus the computer uses 

 small positive numbers exclusively, and his results are expressed as 

 percentages of the solar constant, taken as 1.944 calories per square 

 centimeter per minute. 



P. 3902 graphs the march of the solar constant from 1920 through 

 1951 and shows that it is closely represented by the summation of 23 

 periods, all exact harmonics of 273 months. To be sure, in 1922 and 

 1923 there is a depression far exceeding any observed since. Per- 

 haps it represents one member of another family of solar changes of 

 long periods. Unfortunately, since 1955 no measures of the solar 

 constant have been made so far as I know. In P. 4462, of 1961, I 

 suggest that the solar constant might be observed daily with higher 

 accuracy than ours from a satellite. That might reveal solar changes 

 of great importance in future years. Figure 8a, p. 12, shows the solar 

 constant repeating after 273 months. 



Figure 10 graphs 26 harmonic periods in solar variation. They are 

 cleared of overriding subharmonics. Small type numbers give the 

 length numbers of the 26 harmonics as fractions of 273 months. All 

 the curves have approximately sine form. Table 3 of P. 4213 tabu- 

 lates the amplitudes of the periods and of their submultiples which 

 were removed for the sake of clearness. Altogether 31 harmonic 

 periods are given in this table with amplitudes ranging in solar radia- 

 tion from 0.18 down to 0.02 of 1 percent. Small as they are, all these 

 amplitudes are far above the probable errors derived in Section 3. 



To show the importance of clearing away the overriding submulti- 

 ple periods, figure 11 is a graph of the period of 39.0 months recently 



