NO. 3 SIXTY-YEAR WEATHER FORECASTS — ABBOT 5 



and event, is 80 ±5.2 percent. The average epoch of the synthesis, 

 '^^77i> is 19^ years earlier than the average basis on which the periods 

 rest, 1897. 



Figures 2 and 3, representing the percentage departures in precipi- 

 tation at Peoria,* 111. (about 100 miles from St. Louis), and St. Louis 

 for the 6 years 1934 to 1939, are included with figure 4, representing 

 departures in temperature at St. Louis for the same years, in order 

 that these several results may easily be compared. The syntheses 

 were made with the same periods given in table i, except as stated in 

 the paragraph which next follows table i. The reader will see, as in 

 figure I, that even in small details the syntheses often follow similar 

 details of variation in the events. He will also note the very close 

 similarity between the precipitation at St. Louis and that at Peoria, 

 during these 6 years, as well as how closely the syntheses, based on 

 the mean year of records, 1897, 40 years earlier, follow the actual 

 events. 



It is a severe test to compute correlation coefficients, for often one 

 or two months' displacements of features, between synthesis and event, 

 make great differences in ordinates of the curves and strongly reduce 

 correlation coefficients. However, all three of these syntheses, 40 years 

 from their mean bases, give correlation coefficients between 50 and 

 70 percent, with probable errors from % to %5 of the coefficients. It is 

 to be noted as surprising that the synthetic temperature curve of 

 St. Louis averages 2° F. above the event in these 6 years. One would 

 have expected it below rather than above. But possibly if longer 

 periods, 136 and 273 months, had been used, the displacement would 

 have disappeared. This displacement of scale was taken account of 

 in computing the correlation coefficient for St. Louis temperature. 

 No such adjustment was required in figures i, 2, and 3, or in figure 5 

 to follow. The reader will note that the ranges, in all these figures of 

 synthesis and event, are substantially the same. 



Figure 5 gives the synthesis and event for St. Louis precipitation 

 from i860, which, as mentioned above, was the last of several unsatis- 

 factory years, to 1887, the first of several unsatisfactory years. As 

 stated above, these unsatisfactory intervals may be reasonably at- 

 tributed to the violent eruptions of the volcanoes Cotopaxi in 1856 and 

 Krakatoa in 1883. The good interval, 1861 to 1886, according to 

 Humphreys, was unusually free from violent volcanic eruptions.^ I 



* The Peoria synthesis is made from a new reduction, with latest improve- 

 ments over that given in Smithsonian Misc. Coll., vol. 117, No. 16, 1952. 



^ It takes several years, apparently, for a great volcanic eruption many thou- 

 sands of miles away to produce its efifects on phases of periodicities in St. Louis 

 precipitation. 



