20 SMITHSONIAN MISCELLANEOUS COLLECTIONS VOL. 65 



dust still remained distinguishable by pyrheliometry in the United 

 States up to near the end of the year 1913. No indication of its 

 presence above Arequipa in either 1912 or 1913 seems to be shown. 

 The volcanic dust from Katmai, though general in the northern 

 hemisphere, seems not to have crossed the equator. 



In the last line of the table the mean monthly radiation values 

 for the whole period of observation have been reduced to what they 

 would have been if the sun's distance had remained uniform at 

 its mean value. The close connection between solar radiation at 

 the earth's surface, and atmospheric humidity is brought out graphi- 

 cally in fig. 1. Ordinates are mean monthly values of e 12 reduced 

 to mean solar distance, abscissae are corresponding mean monthly 

 values of water vapor pressure (p). The smoothness of the curve 

 defined by these points is remarkable. It is perhaps to be ascribed 

 to the great altitude and inland location of Arequipa. Apparently 

 the degree of atmospheric humidity at the earth's surface there is 

 a good index of the total quantity of humidity existing between the 

 station and the limit of the atmosphere. 



It is obvious, of course, that fluctuations of atmospheric trans- 

 mission coefficients must also produce their effects on the observed 

 intensity of solar radiation at the station. Such fluctuations are 

 of two kinds : First, those associated with changes of water vapor. 

 Second, those associated with changes of dustiness, such as those 

 produced in the northern hemisphere by the Katmai eruption. The 

 influence on the solar radiation of fluctuations of the first type, 

 which are a function of the humidity, may be generally (for a 

 high-level station like Arequipa) much greater than those associated 

 with dust alone. But it might well be expected that for certain 

 months of the year the dust fluctuations would be by no means 

 negligible. However, restricting our thought to a high-level station 

 like Arequipa, and remembering the powerful true absorption pro- 

 duced in the infra-red spectrum by water vapor, and the large 

 changes in this true absorption attending changes of humidity when 

 the humidity and air mass are both small, it is easy to see after all 

 why the observed radiation at M—1.2. at Arequipa seems to be so 

 well represented as a function of water vapor alone. For both the 

 true absorption and a large proportion of the variable elements of 

 the general scattering are functions of water vapor. Compared 

 to these, the variable scattering produced by dry dust alone is gen- 

 erally small. 



In figure 2 the radiation, e x (not reduced to mean solar distance), 

 the vapor pressure, p, and the transmission, a 2 , are all given as func- 

 tions of the time of the year. 



