8 SMITHSONIAN MISCELLANEOUS COLLECTIONS VOL. 128 



give in table 2 the mean values of the average deviation in figure 5 

 for seven 4-year intervals, and the general averages of these seven 

 quantities, which may be compared to the range of precipitation, 130 

 percent of normal. 



The range of precipitation, i860 to 1887, in 5-month running 

 means, was a little over 130 percent of normal precipitation. 



I cannot but believe that if similar results to figure 5 were computed 

 for perhaps 50 selected stations over the United States, areas of equal 

 departures from normal precipitation could be mapped years in ad- 

 vance, which, unless vitiated by world wars or violent volcanic erup- 

 tions, would be of much practical use. 



I complete Part i with a very ambitious prediction of precipitation 

 at St. Louis and Peoria shown in figure 6. Counting from 1897, the 

 middle of their common interval of preparation, the predictions, which 



Table 2. — Average deviations, in percentages of normal precipitation, bctvoeen 



prediction and event for 28 years' synthesis of St. Louis 



precipitation. Means of individual monthly differences 



Time interval 1860-63 1864-67 1868-71 1872-75 1876-79 1880-83 1884-87 General 



Average deviation .. ±17 ±14 ±22 ±13 ±17 ±17 ±17 ±I7 



Scale displacement. 

 Event minus pre- 

 diction -fs +6 — 8 -I -f8 — 2 -f6 +2 



end with 1957, extend 60 years from their mean base, and 18 years 

 beyond December 1939, the last month used in the preparation of 

 that base. 



Since I have not employed the long periods of iif, 22|, and 45^ 

 years in these syntheses, I feel warranted in adjusting the levels of 

 the two predictions to suit the prevailing conditions of precipitation, 

 1952 to 1954. This involved lowering the St. Louis curve of synthesis, 

 1952 to 1957, bodily, by 20 percent of the normal precipitation. I 

 made no change for Peoria. 



Although the features are similar, it will be noted that observed 

 precipitation precedes synthesis at St. Louis in 1952 by four months 

 and follows synthesis at Peoria by four months in 1952 and 1953. Yet 

 almost throughout the 6 years of prediction, St, Louis and Peoria 

 syntheses are nearly in step, although resting on two independent 

 series of observations 80 years long, which ended in 1939. These dis- 

 placements, between prediction and event, similar to those remarked 

 in earlier papers, are puzzling. Perhaps the solar mechanism that 

 operates has slips, making events early or late, as has long been noted 

 in the i if -year sunspot period. 



