NO. 9 FORECAST OF U. S. PRECIPITATION — ABBOT J 



condition varies not only with locality but with time of the year, 

 prevalence of sunspots, and march of population. To partially meet 

 these difficulties, I tabulate separately for three periods of the year : 

 January- April ; May-August; September-December; also with Wolf 

 sunspot numbers above and below 20; also with lapse of time before 

 and after the midpoint of the record. These divisions of the available 

 monthly data lead to computing 220 tables at each station before 

 undertaking a forecast. 



Forecasts by periods. — My forecasts are made by adding the effects 

 of 27 regular periodic cycles in precipitation. These cycles, like the 

 harmonics of musical sounds, proceed simultaneously, and are in- 

 tegrally related to a fundamental cycle. This fundamental is 273 

 months. The harmonics employed are as follows : 



Table 2. — Periods used for forecasting 



The harmonic family referred to was discovered in the variation 

 of the measures of the solar constant of radiation. Figure 4 shows 

 26 of over 60 periods discovered in solar variation. 2 Identical cycles 

 were later found in precipitation and temperature by study of long- 

 continued weather records. While the periods of the harmonics are 

 invariable, both in the sun and weather, and their phases are invari- 

 able in solar radiation, their phases shift in weather, depending on 

 atmospheric influences, as will be described below. On account of 

 these phase changes, depending on several variables discovered in my 

 studies of precipitation begun with Peoria, 111., about 10 years ago, 

 the harmonic family in weather is obscured and hidden, and is as yet 

 unrecognized by most meteorologists. Nevertheless it is verified by 

 an enormous mass of evidence, as will appear below. 



No observations required. — Many meteorologists and others sup- 

 pose that my method of long-range weather forecasting depends on 

 solar observations, but this is not so. The harmonic family referred 



2 See in reference, footnote 1, e, above, figure 3 and table 3. 



