14 SMITHSONIAN MISCELLANEOUS COLLECTIONS VOL, I07 



computed as based on the date January 3.0000, 1924, which is my 

 fiducial starting point. It is useless to employ in this study any mea- 

 surements of the solar constant of radiation other than those observed 

 at the best Smithsonian station, namely, Montezuma, Chile. Results 

 of other stations are too inaccurate. To include them would defeat 

 our purpose. 



8. TABULATIONS OF MONTEZUMA SOLAR-CONSTANT DATA 



For each month of the year I made a separate tabulation of the solar- 

 constant values from Montezuma, from 8 days before to 1 1 days after 

 each zeroth date of recurrence of the period of 6.6456 days, computed 

 from January 3.0000, 1924. Unfortunately a great many days are 

 missing from the record. In many recurrences of the period the 

 missing days are so numerous that these incompletely recorded re- 

 currences must be thrown out. Only 310 cases, and these all in the 9 

 months March to November, were thought complete enough to be 

 retained. 



It remained to decide whether to fill up the dates missing in the 

 310 cases retained. After consideration this appeared necessary. For 

 the solar constant varies from about 1.930 to about 1.965 frequently, 

 and sometimes beyond these limits. The mean values of successive 

 columns of a tabulation covering numerous groups of values, of 20 

 successive days each, would not be comparable if some of these 

 columns missed a number of high values, and other columns missed 

 a number of low values, because of unfavorable observing conditions. 

 Accordingly I interpolated missing values to harmonize best with 

 values next preceding and next following them. Besides these there 

 were a few wild values, clearly incorrect when compared to those pre- 

 ceding and following. Interpolated values were substituted for them 

 also. The number of scattering interpolated values was counted, and 

 it proved to be about 3/10 of the whole number of days covering the 

 310 retained recurrences of the 6.6456-day period. Forasmuch as the 

 interpolated values could not have a part in determining the true solar 

 variation, the amplitude of the mean resulting variation is to be multi- 

 plied by 10/7, to give what it would be if the record were complete. 



9. QUESTIONS REGARDING THE SUN'S EMISSION, AND 

 METHODS OF INVESTIGATION THEREOF 



The following questions were to be solved : 



I. What is the form and amplitude of the mean curve, 1924 to 

 1944, of the sun's ])eriodic variation of radiation? 



