l6 SMITHSONIAN MISCELLANEOUS COLLECTIONS VOL, I07 



ference in the amplitudes of the temperature effects of the cycle. 

 Naturally, therefore, I separated the solar-constant data into two 

 groups, one corresponding with occasions when large amplitudes of 

 temperature change appeared, and the other for small amplitudes. 

 This segregation, I hoped, would indicate whether solar or terrestrial 

 causes produced the differences of amplitude. 



Regarding query 4, I made two reductions of the solar-constant 

 data. In one I arranged the tabulation so that corresponding com- 

 puted dates ran vertically, without phase shifting, in the monthly 

 tables. In the other I used the same phase displacement corrections 

 of the data which I had used so successfully with the temperatures, as 

 illustrated above in tables 3 and 4. If now, the second tabulation should 

 give larger and more consistent mean amplitudes of solar change than 

 the first, it would appear that the solar periodic variation, like the 

 temperature variation, was but quasi-periodic after all, and truly 

 periodic only in the mean over long intervals of time. In such a case, 

 though the period 6.6456 was found to persist as a general mean, the 

 individual recurrences of it in the sun's radiation would appear subject 

 to displacements of phase. 



With these explanations given, I now proceed to the results. 



10. RESULTS OF TABULATIONS OF SOLAR VARIATION 



In figure 7 I give a photographic reproduction of my tabulation of 

 solar constants for the month of September, 1923 ^ to 1944. The data 

 used, when available, are "Preferred Solar-Constant Values" from 

 table 24 of volume 6, Annals of the Smithsonian Astrophysical Ob- 

 servatory. For accuracy's sake Montezuma values only are to be used. 

 Hence if more than one station contributed to the "preferred" value, 

 it is discarded, and the mean of the Montezuma values for the day is 

 substituted. The reader will understand that the solar-constant value 

 is the number in the text and in figure 7, prefixed by i .9. The values 

 from October 1939 to December 1944 are from unpublished results, 

 kindly communicated by the Director of the Astrophysical Observa- 

 tory, L. B. Aldrich. In figure 7 all interpolated values are under- 

 lined, as for example, "1923, —8, 5^." Parentheses about a value in- 

 dicate that, though observed, it is so wild that its inclusion would 

 falsify the result. Thus: "1925, +2, (34)." In such cases an inter- 

 polated value was used. In the case just cited, 46 was interpolated. 



From the mean values at the bottom of the tabulation the graph, 



^ Several months of the year 1923 are included in table 24 of volume 6 of the 

 Annals and are used here. The year 1944 yielded no favorable cases. 



