8o SMITHSONIAN MISCELLANEOUS COLLECTIONS VOL. 94 



Fair, 1 1 : New Haven, Albany, Philadelphia, Washington, Charles- 

 ton, Peoria, Galveston, Santa Fe. Denver, San Fran- 

 cisco, Spokane. 



Half and Half, 8: Key West, Cincinnati, Omaha, Helena, Salt Lake, 

 San Diego, Red Bluff, Portland. 



Bad, 5 : Hatteras, Mobile, Nashville, Abilene, Cheyenne. 



In order to give the reader a fair idea of this system of ranking 

 these forecasts, I present in figure 35 a sample prediction and verifica- 

 tion during 1934 from each group named above. 



As a further comment on the basis on which these predictions rest, 

 I refer again to figures 33 and 34 which show precipitation and tem- 

 perature departures arranged in 23-year cycles. It is observed, as 

 illustrated in figure 34 and as might be expected in view of caption 

 14-g, above, that frequently the resemblance is closer between cycles 

 separated by 46 years, than by those separated by 23 years.^* Changes 

 of phase and of amplitude certainly exist between repetitions of the 

 characteristic features which comprise a 23-year cycle. These must 

 indeed have been expected in view of the discussion given above of 

 the periodicities in the departures at Berlin and other stations. Never- 

 theless, in the preparation of nearly 70 three-year predictions, above 

 mentioned, the conviction was steadily deepened that many features 

 may nearly always be recognized in successive 23-year cycles. 



Owing to the great financial importance which these predictions 

 would assume if they could be regarded as trustworthy, it has seemed 

 improper to publish them until the lapse of another year, or even 2 

 years, shall have proved to what extent they may be relied upon. 



Employing only weather data previous to and including 192 1, fore- 

 casts have been made, first for Bismarck, N. Dak., in one continuous 

 interval from 1922 to 1932, and then by successive steps for Vienna, 

 Austria, and North Platte, Nebr., in 11 intervals of i year each from 

 1922 to 1932. These forecasts and their verifications are shown in 

 figures 36 and 37. 



26. Causes 



Evidence has been presented which seems to show that the radiation 

 of the sun varies in a complex mode comprised of the summation of 

 12 or more periodicities, all of which are integral submultiples of 

 23 years. Corresponding periodicities have been traced in weather, and 

 several other weather periodicities have been found which are also 

 integral submultiples of 23 years. Inversions, or at least major changes 

 in form, phase, or amplitude, have been disclosed in the periodicities 



" Compare the general swing of curves 2 and 4 in figure 34. 



