10 



SMITHSONIAN MISCELLANEOUS COLLECTIONS VOL. y"] 



This relates to the past seven years. I am not prepared to insist that 

 it runs back of 1918. Possibly, like the hot and cold side of the 

 sun, which holds sometimes for several revolutions, it may have 

 disappeared in process of time. 



In Professor Marvin's 12-month curves from Mount Wilson 

 observations of from 3 to 7 months duration, it is necessary for him 

 to exterpolate over half the year. He combines 15 years of observing. 

 Surely he should have omitted July and August of the year 191 2 

 when the sky was so very turbid from the Katmai eruption that its 

 skylight reaching the pyrheliometer may very likely have led to 

 higher values than its relation to sun-spot minimum would have led 

 us to expect. 



i^-hiVfnf.iri ; riiriit;i'(iiriitiitM-iiiiTr' 



f.S40 



'930 



'920 



JUJUU 



i^miijjjjJ. 



Fig. 6. — Eleven-month periodicity in solar variation. The monthly means 

 for solar radiation given in the upper curves are smoothed by a usual 

 process, and show minima eleven months apart, as indicated by the arrows. 



The Mount Wilson values of the years 1918, 1919, and 1920 agree 

 in detail with Calama, as I shall show directly, so that they need not 

 be considered as indicating spuriously high summer values. The 

 monthly mean values for the other 11 years are plotted in figure 7. 

 Maxima occur in every month observed except May, and minima 

 occur in every month observed except August. The run of the curves 

 is so varied that one cannot safely conclude how the other unobserved 

 months of the year would have turned out. Professor Marvin has 

 thought that their maxima agree with minima of Chile, northern 

 summer agreeing with southern summer, and northern winter with 

 southern winter. His Mount Wilson and Chile data refer to different 

 years. There is no fair comparison of one year with another when 



