30 



SMITHSONIAN MISCELLANEOUS COLLECTIONS VOL. I46 



Fig. 23. — Precipitation, Helena, Mont., 39-month period, cleared of 

 shorter harmonics. 



tion for the year 1956 in P. 4211 is a "60-year forecast," as claimed 

 by the title of P. 4211. 



I will conclude these illustrative examples by graphs (see figs. 28- 

 37)' taken from P. 4390 and P. 4471 on precipitation and tempera- 

 ture forecasted or backcasted from data smoothed by 3-month con- 

 secutive means using all the records from about 1870 to 1956. It 

 will be seen that forecasts and events have about equal amplitudes. 

 They evidently exhibit the same principal features. Principal and even 

 minor features in prediction and event prevailingly coincide on the 

 same months. But sometimes there are displacements of 1,2, 3, or 



5 Figure 33 was prepared from the 1,032 months of records used in P. 4390, 

 "A Long-range Forecast of United States Precipitation." These records cov- 

 ered the years 1870-1956 and centered on the year 1913. All 1,022 months of 

 these records had equal weight in the forecasts. The observations quoted in 

 figure 33 were not available till late 1960. Figure 33 was used as a slide 

 at my National Academy paper of April 1961. Whatever success it has 

 is for being a verification of forecasts of precipitation for 14 cities from a 

 zero date of 1912, 46 years preinoits to 1959. 



