78 ANNUAL REPORT SMITHSONIAN INSTITUTION, 19 3 6 



casting must await Government support. The necessary expense will 

 approximate $200,000 a year. It is necessary to observe the complete 

 solar radiation of all wave lengths nearly every single day to an 

 accuracy of l^ of a percent at mountain solar-constant stations, or else 

 to observe daily the intensity of a certain band of ultraviolet solar 

 rays from sounding balloons at an altitude of 100,000 feet or more, 

 with an accuracy of 2 percent or better. Either j^lan involves diffi- 

 culty and expense, but the probable advance in weather forecasting 

 seems to justify it. 



In August 1935 Dr. Abbot published a study of the variation found 

 in the monthly mean values of the solar constant of radiation since the 

 year 1920. This was entitled "Solar Radiation and Weather Studies." 

 The following conclusions are quoted from the summary : 



1. The output of radiation of tlie sun varies, as proved by simultaneous 

 observations at three stations remote from each other. 



2. The solar variation, seemingly irregular, really comprises 12 or more regu- 

 lar periodicities, which support successful predictions of solar changes for years 

 in advance. 



11. Forecasts based on these relations having been made to cover the years 

 1934, 1935, and 1936 for more than 30 stations in the United States, these fore- 

 casts are fairly well verified both as to temperature and precipitation in 1934. 



Although, as stated in conclusion 11, fair success was achieved in 

 the predictions for the year 1934, the success was less complete in 1935 

 and 1936. Weather details, though present, were found to be displaced 

 in time so much as to invalidate the method for detailed predictions 

 more than a year in advance. This partial failure is believed to be 

 caused by the changes of phase of the subordinate periodicities which 

 make up the 23-year cycle. It is hoped that further study may give 

 means of predicting these changes of phase, and so of perfecting the 

 method. 



But while the prediction of details is thus unsatisfactory, certain 

 large and prolonged features, like the great drought in the Northwest, 

 seem to be very clearly provisioned by the cycle. This appears plainly 

 in the accompanying figure 1, a reproduction of figure 26B of the 

 article. It is very plain that the first decades of each of the five 23-year 

 cycles shown there were marked by the fall of the level of Lake Huron, 

 but that this depression in the first, third, and fifth cycles is outstand- 

 ing. Thus there seems to be a 46-year cycle of great droughts, and 

 while moderate drought conditions may be expected in the decade 

 1950-1960, a severe drought is probable in the decade 1975 to 1985. 



The present drought seems comparable to those of the decades fol- 

 lowing 1840 and 1890, and may tlierefore be expected to last for a year, 

 or possibly 2 years more before the initiation of recovery. But it 

 would be rash to make a more definite prediction. 



