NO. 9 FORECAST OF U. S. PRECIPITATION — ABBOT 23 



be of value to other investigators of periodicity I first give in table 9 

 the two sets of normals for the 32 cities I have investigated. 



The cities are in alphabetical order. The months in the first column 

 apply for all cities. Precipitation is given in inches. Columns A and 

 B give monthly normals for times when Wolf sunspot numbers are 

 respectively less and more than 20. 



Departures; observation minus forecast 1950-1958. — There are 20 

 cities showing (1950-1958) departures in level of 4 percent or more 

 from the values given in table 9. This is to be expected. One could 

 not suppose the mean precipitation, 1950- 1958, would be identical 

 with the average precipitation, 1870- 1958. Table 9a gives all the 

 cities where such differences of 4 percent or more occurred. 



When I come to give tables and maps of forecasts, 1959- 1967, I 

 shall not use table 9a to correct the maps, but shall quote the results 

 as they are determined from table 9. Persons interested may apply 

 the values of A, table 9a, as corrections in level to the forecasts, using 

 them in reverse of the signs given in table 9a. 



Sunspot effect on normals. — Lest readers think the differences 

 between mean precipitation values attending high and low sunspot 

 frequency are merely due to the sparsity of evidence, considering the 

 irregularity of precipitation, I call attention to the numbers of months 

 entering into the mean values of table 9. For nearly all of the stations 

 approximately a thousand months participated. That indicates about 

 600 for high sunspot frequency, about 400 for the low. Dividing by 

 12, there were about 50 values per monthly mean for sunspots ex- 

 ceeding 20 Wolf sunspot numbers, and about 33 per month for the 

 low sunspot frequencies. 



Referring to table 9, the yearly sums show seven cities where sun- 

 spot frequency makes no more than 1 percent difference in the totals. 

 For seven other cities low sunspot activity brings more precipitation, 

 with an average difference of 5 percent. For the remaining 18 cities 

 precipitation averages 5-1/2 percent higher at high sunspot frequency. 

 While the discovery and elimination of these differences by computing 

 new normals was of importance in my forecasting, seasonal differ- 

 ences made the elimination of the sunspot effect imperative. Thus at 

 Salisbury, N. C, precipitation averages 17 percent higher with low 

 Wolf numbers, January- April ; 9 percent lower, May- August ; and 

 11 percent higher, September-December, for Wolf numbers below 20 

 than for those above 20. 



Credibility of forecasts. — It is difficult to compress within the limits 

 of a paper, aimed to be available at moderate price to all who desire 



