NO. 5 SOLAR VARIATION AND FORECASTING ABBOT 3 



All of the Mount Wilson work, excluding" altogether July and 

 August, 1912, is useful in the form of averages. It is only since 

 January, 19 19, when we have had several determinations each day 

 by a method which avoids errors from the variability of the sky, and 

 much of the time have received results from two stations, that indi- 

 vidual values have begun to deserve some confidence. Even yet, they 

 are not up to the class which we hope they will reach within one or two 

 years more. They are still most useful in the form of mean values. 

 This, indeed, is a major reason why correlation coefficients reported 

 by Clayton and by Hoxmark, in discussing their solar forecasts of 

 daily weather conditions, are still low. Very accurate solar radiation 

 data are necessary for that purpose, and we cannot yet quite reach 

 the required degree of accuracy. The methods of reduction of 

 observations for the station at Montezuma are being improved, the 

 Harqua Hala station is being transferred to Table Mountain, Cali- 

 fornia, 2.000 feet higher, and the National Geographic Society is 

 installing, in cooperation with the Smithsonian Institution, a new 

 station in the Eastern Hemisphere. We beheve that these improve- 

 ments will in about two years largely better the results. 



DEFENSIVE ARGUMENTS 



I. Our critic and I approach this matter from opposite points of 

 view. He has felt that it is necessary to be sure that our solar 

 observations are sufficiently impeccable before he can use them. 

 I was convinced five years ago by figure i that one can use them, 

 and having, in cooperation with the Argentine Weather Bureau and 

 with Mr. Clayton, tried experiments in using them, every month 

 reveals new evidences that they can be used. Consider figure i. The 

 high reputation of Mr. Clayton, whose results are here shown, 

 forbids us to doubt that the march of the curves is real. 



What then? Certain observations made on Mount Wilson, Cali- 

 fornia, in the years 1913, 1914. 191 5 and 1918 were definitely asso- 

 ciated with temperature differences of 6° F. at Buenos Aires, 

 Argentina, 10 days after the event.'' But, says our critic, this is not 

 a solar but a terrestrial correlation. I am so constituted that, if I 



^ Owing to errors often occurring at Mount Wilson because of increasing 

 or decreasing haziness during an observation (errors nowadays eliminated in 

 our "new method") no doubt some values in the high and the low groups 

 of solar constants used by Clayton were extreme because erroneous. Thus the 

 range he finds of 5 per cent, he and I now agree was probably not over 

 25^ per cent in reality. As will be seen in his present paper, such a range of 

 solar constant is large enough to produce notable effects. 



