56 SMITHSONIAN MISCELLANEOUS COLLECTIONS VOL. 139 



At the end of text are 27 maps of the United States with the 32 cities 

 as listed above, and each accompanied above the circle by a number 

 identical with the appropriate number in the column headings in 

 table 13. Below is the predicted departure from normal. Each group 

 of three maps covers the three intervals per year of four months each 

 named with table 13. Large areas of approximately equal departures 

 from normal precipitation are clearly noticeable on the maps. These 

 area similarities may aid farmers remote from the 32 cities to estimate 

 the precipitation probable in their locations. 



APPLICATIONS 



Periods control long-range weather. — I have sought to present to 

 meteorologists evidence of two important propositions. First, that 

 there exists in weather a family of periods, all exact submultiples of 

 273 months. These periods are hidden from immediate recognition 

 because their phases are shifted according to the state of the atmos- 

 phere. When, however, the long monthly records are grouped and 

 reduced with reference to time of the year, sunspot activity, and 

 march of population, the family of periods is clearly disclosed with 

 constant length, and with approximate sine-curve forms. 



Second, long-range precipitation is to nearly 60 percent governed 

 by this family of periods. By evaluating the average forms and 

 amplitudes of these periods from thousand-month records, precipita- 

 tion and temperature may be forecasted for years in advance, with 

 considerable approximation to the event. 



Whether these forecasts will appear to interested parties as trust- 

 worthy guides to help in planning their future operations must depend 

 on the agreement attained between forecasts and events, 1950- 1958. 

 I therefore prepared table 14 which gives for 32 cities the 4-month 

 forecasts and observations, 1950- 1958, and the differences in per- 

 centages of normal precipitation, A, in the sense observed minus fore- 

 casted. Their means are given disregarding signs. 



Agricultural requirements. — For agricultural purposes a foreknowl- 

 edge of seasons rather than of individual months is most desired. 

 Hence I give in table 14 4-month mean values computed from table 10. 

 But it is the difference between forecast and event which would be 

 the controlling factor in estimating the value of the forecasts. 5 The 

 average differences, A (observed minus scale-corrected forecasts) are 



6 As differences in level of observed precipitation, 1950-1958, from the averages 

 of 1,000 months, are disclosed in table 9a, I refer to that table for possible cor- 

 rections of level which might be applied to values for some stations in table 14. 



