NO. 3 SIXTY-YEAR WEATHER FORECASTS — ABBOT I5 



The tendency to make mistakes, in tabulating and adding for s)m- 

 theses, will be very apparent if one reflects that, when a change comes 

 between Wolf numbers ^ 20, the computer must change from tables 

 like table 5 to their counterparts for Wolf numbers < 20; that at all 

 times he must remember to reverse the displacements indicated by the 

 arrow subscripts of displacements in those tables ; that for periods not 

 of whole numbers of months he must insert or reject a month from 

 time to time ; that he must always start a column in the correct phase; 

 that when adding the 22 or more columns, as in figure i, he must not 

 only be sure to correctly read the values, but also their algebraic signs. 

 With all the care I can muster, I have always found many mistakes, 

 of one or another of these sorts, when rechecking the whole process, 



I give in figure 8 the "departure" curves like those indicated in 

 table 5, as found for the ii^-month period at St. Louis, when Wolf 

 numbers > 20. The reader will see from table 6 relating to Wolf 

 numbers ^ 20 that, though two months displaced in phase, the two 

 series, coming from entirely different data, are identical in range and 

 nearly identical in form. These pleasing similarities between such 

 pairs of series for Wolf numbers ^ 20, representing all the 19 or 

 more periods which can be thus compared, are almost invariably 

 found. 



This in itself is evidence that the existence in weather of the large 

 family of periodicities, aliquot parts of 22f years, has a sound basis. 

 For in all tabulations of many values, leading to a periodic result, the 

 careful computer always divides the data into two or more sections, 

 to see if the several groups of data yield the same periodic result. As 

 just stated this criterion is well met in this research. 



But there is another evidence of great weight. The ranges of the 

 periodic curves in weather are not small. In precipitation at St, Louis, 

 Peoria, and Albany, the ranges of the periods run from 5 to 25 per- 

 cent. This is a very great matter, if the critic is inclined to think of 

 accidental error, and to scout my whole investigation as trifling or 

 spurious. 



In the third place, as seen over the 26 good years shown in figure 5, 

 the great features, and many of the minor features, found in the 

 event are seen to be also in the synthesis, and in very nearly the 

 same amplitudes and positions. That this does not hold true for 30 

 out of 100 years of synthesis of St. Louis precipitation has been rea- 

 sonably explained above as probably caused by tremendous disturb- 

 ances of the atmosphere by violent volcanic outbursts, by the prodigal 

 bombing during the two world wars, and the Korean war, and by the 

 tests in recent years of atomic bombs in different parts of the world. 



