2 SMITHSONIAN MISCELLANEOUS COLLECTIONS VOL. 89 



Meteorologists generally do not unreservedly accept the reality of 

 solar variation. Much less do they admit the conclusiveness of the 

 studies of periodicity just referred to. Definitive monthly mean values 

 of solar variation having been published up to the end of 1930 in 

 Table 45 of the Annals of the Astrophysical Observatory, Volume V, 

 and preliminary values being now available including February 1933. 

 1 have made a new analysis based solely on the latest results beginning 

 with January 1924. During this recent period, excepting since June 

 1932, when A'lontezuma work is not yet available, the daily observa- 

 tions are of the highest weight yet reached in our solar-constant 

 research. 



JAN.. MAR. MAY JULY SEPT. NOV. JAN. MAR. MAY JULY SEPT. NOV. JAN. 

 1931 1932 



Fig. I. — Forecast made November 1930 and verification. 



In the meantime, the periodometer, an instrument for discovering 

 and evaluating periodicities," has been invented. By means of it a new 

 analysis of the definitive solar-constant values of 1920 to 1930 inclu- 

 sive was made by Mrs. A. M. Bond, as described in pages 5 and 6 and 

 Figure 3 of a paper entitled " Periodicity in Solar Variation." "* In 

 addition to the five periodicities found in my earlier work, she dis- 

 covered two more of 7 and 21 months period, respectively. 



Assuming, then, that seven periodicities of 7, 8, 11, 21, 25, 45, and 

 68 months period, respectively, are requisite and sufiicient to comprise 

 the variation of the sun, I sought to evaluate them as accurately as pos- 

 sible from the data of January 1924 to February 1933, inclusive. To 

 minimize error I preferred to use the method of computing explained 



^Smithsonian Misc. Coll., vol. 87, no. 4, 1932. 



^Smithsonian Misc. Coll., vol. 87, no. 9, 1932. Bj^ mistake the side scale num- 

 bers on curves B to H are on twice the proper scale. 



