4 SMITHSONIAN MISCELLANEOUS COLLECTIONS VOL. I28 



departures from i .90 calories per square centimeter per minute. They 

 were given, not in calories, but in fractions of our mean values of the 

 solar constant, which is 1.946 calories per square centimeter per 

 minute. Thus an observed value, for instance 1.950 calories, would 



be computed for tabulation as a departure, -^ — -^ — , or 2.58 per- 

 cent of the mean solar constant. To save printing, the decimal point 

 is omitted in the table. When the range of a period in solar variation 

 is found (if, for example, it is 7 on this scale) we know at once that 

 its range is 0.07 percent of the mean solar constant. In table i of the 

 present paper I add to the former table mentioned above the de- 

 partures in solar-constant values from volume 7 of the Annals, which 

 represent the years 195 1 and 1952. 



As I said above, comparisons of daily solar-constant measures, 

 made in different hemispheres over many years, show that they go 

 along together without systematic departures. Their accidental de- 

 partures indicate a daily probable error of only ^ percent when two 

 stations cooperate. It is not material to our investigation of solar 

 variation that our results should be expressed on precisely the true 

 scale of heat units. Hence, though interesting, it is not disturbing 

 that our accepted mean solar-constant value differs from the new 

 determination of the solar constant by F. S. Johnson.- He used, 

 in part, recent rocket observations, and his result is 2.00 ±0.04 calo- 

 ries. It differs by 2.7 percent from ours, which is 1.946 calories. But 

 our work on solar variation is not thereby prejudiced. 



I now proceed to illustrate, by examples of a short and a long pe- 

 riod in solar variation, the work presented. For the short period I 

 take 7 months, which is 1/39x273 months. By tabulations and graphs, 

 this 7-month period and its superriders will be given in considerable 

 detail. For the long period I take 68:| months. To save printing, the 

 68^-month period will be given mainly by graphs. The method for it 

 will have been apparent from the treatment of the 7-month period. 



Table 2 gives the means of three groups of tabulations with 21 

 1 0-day mean values in each group, making an average 7-month pe- 

 riod. They therefore summarize 51 repetitions. They include, respec- 

 tively, 17 repetitions covering the interval September 1923 to Decem- 

 ber 1932; 17 repetitions January 1933 to November 1942; and 17 

 repetitions December 1942 to October 1952. The next step with each 

 group is to compute the average march of thirds of the mean sum- 

 maries. The justification for this will be apparent from the triple 



Journ. Meteorol., vol. 11, No. 6, December 1954. 



