84 



ANNUAL REPORT SMITHSONIAN INSTITUTION, 1912. 



better exhibit the comparison. The vertical scale (fig. 2) represents 

 Mount Wilson values and the horizontal scale Bassour values of the 

 solar constant for each day when satisfactory observations were se- 

 cured at both stations. If the values observed were without ert-or, 

 it is obvious that for each day they would have been identical at the 

 two stations. Hence, if the solar radiation had values of 1.90, 1.95, 

 and 2 calories on three different days, they should have been repre- 

 sented by points at the lower left corner, the center, and the upper 

 right corner of our diagram, if observed at both stations without 



%oo-' 



O 30 31 I t 3 ^ J- 6 ~f 8 9 /o T/ 7Z 73 /f- /s- ^T 



Fig. 1. — Successive solar constant values, Bassour and Mount Wilson. 



error. 



In general all values of the solar constant would fall on the 

 line A B of the figure if the measurements were without error. But 

 we have found the Mount Wilson values consistently lower by 2 per 

 cent. If we admit a constant systematic error of this magnitude, 

 but still deny all accidental error of measurement, then all observa- 

 tions should fall on the line C D of our diagram. They must all lie 

 at a single point of C D if the solar radiation is constant, but may 

 fall anyAvhere upon that line if the solar radiation is variable. In 

 practice it is of course never possible to avoid accidental errors of 



