4 SMITHSONIAN MISCELLANEOUS COLLECTIONS VOL. 95 



but all Other dates found in Table i of my former publication were 

 used as dates of incipient solar changes. 



To obtain departures from approximate daily normals at Potsdam, 

 the values of the daily means of temperature at y^, 14*', 21^ were 

 employed. Monthly means of these daily means were available for 

 the years 1921 to 1934, inclusive.^ Average monthly values were com- 

 puted for this interval and plotted on a sufficient scale. A smooth 

 curve was drawn from these monthly averages, and the dififerences 

 between it and the observed daily mean temperatures were used as 

 temperature departures. 



It is recognized that the normals thus obtained are only approximate. 

 But this will not impair the value of the departures for the purpose 

 proposed. For instance, if the normals thus obtained were a little 

 low for a given month, the departures for all years during that month 

 would be prevailingly positive, but the curve of their fhictuation over 

 an interval of 16 days would l)e almost precisely the same as if the 

 normals had been perfect. 



Proceeding then in exactly the same way as described in my former 

 publication, cited above, we arrive at the results shown in figure 2. 

 Except for the months of May and June,^ the results seem to be as 

 emphatically favorable for Potsdam as for Washington to the former 

 conclusions, as follows : Opposite changes of solar radiation are asso- 

 ciated for at least 2 weeks after their commencement with opposite 

 marches of temperature departures in weather. These average effects 

 are of the order of several degrees Centigrade, although the solar 

 changes on which they depend are only of the average range of about 

 0.7 percent. If daily excellent values of the solar constant of radiation, 

 trustworthy to 0.2 percent, were available, certain weather features 

 might probably be predicted for 2 weeks or more in advance. Since 

 theory and observation indicate that the solar radiation changes frac- 

 tionally perhaps ten times as much near 3. too angstroms as in the solar 

 constant, there is great hope that automatic records of ultraviolet solar 

 radiation at great altitudes may be of first rate importance for weather 

 forecasting. 



'Beob. Obs. Potsdam, Deutsch. Meteorol. Jahrb., Teil 4, Heft i. 

 ' In these two months considerable temperature changes are found at Potsdam, 

 but they run parallel for opposite variations of the sun. 



