NO. 3 



SOLAR VARIATION AND WEATHER — ABBOT 



53 



Fig. 44.- 



-Drought prediction from forecasts of precipitation, St. Louis, Mo, 

 Peoria, III. 



, and 



Referring to Fig. 1 of my just cited paper, here reproduced [here figure 46], 

 the reader will find there two curves for each month of the year showing 

 departures from normal temperature at Washington, D.C. In each month the 

 curves show a well-marked opposition like the right and left hands. The sep- 

 arations of the curves in the months January, February, March, April, May, 

 June, August, October, November, and December range from 14° to 24° F., 

 and evidently constitute major departures from normal temperatures. Similar 

 results, showing in almost all cases opposition like the right and left hands, 

 but differing widely in actual march of the pairs of curves, are shown for 

 St. Louis, Helena, and Potsdam, in other illustrations in the cited publications. 

 The curves are computed for all these cities starting from identical dates, 

 320 in number, scattered over 12 years. Some 10 to 20 cases combined are in each 

 curve shown. The data of temperature departures in each case cover 16 days 

 following the starting date selected. 



How were these 320 dates selected? They are chosen as dates when solar 

 variations commenced. As shown in Fig. 1 and Table I of "The dependence 

 of terrestrial temperatures on the variations of the sun's radiation," they 

 comprise all the dates during 12 years when good consecutive solar constant 

 observations, made mostly at Montezuma, Chile, began to indicate rising or 

 falling sequences of the sun's output of radiation. The range of these sequences 

 was small, rarely exceeding 1 percent, and their mean range is only about 0.7 per 

 cent of the solar constant. Owing to the interference caused by changes in 

 atmospheric transparency, superposed on the inevitable accidental errors of 

 measurement, it is highly probable that some of these 320 cases are spurious. 

 If the spurious cases could be eliminated the average temperature departures 

 would doubtless be increased above their already large magnitudes. 



