abbot] studies of solar constant of radiation 77 



the accuracy of the extrapolation. In order to give an impression 

 of the weight which should be assigned to the solar constant values 

 shortly to be given, I call attention to plate xx, which contains the 

 plots for deducing atmospheric transmission at several wave-lengths 

 for two days, March 25, 1903, and March 26, 1903, observations for 

 the two days being represented by circles and crosses respectively. 

 The tangent of the angle of inclination of the plotted lines is the 

 logarithm of the coefficient of transparency of the atmosphere for 

 vertical transmission of a ray of the given wave-length. Plots i and 

 II represent a wave-length of 1.027 ^ ; iii and iv, 0.656 p.', v and 

 VI, 0.468 IX ; and vii and viii, 0.395 p,. In connection with this 

 branch of the subject it is well to remark what the experience of 

 meteorologists generally no doubt confirms, that the afternoon hours 

 are found far more uniform in transparency of the air than the morn- 

 ing hours, so that the observations of atmospheric transmission for 

 use in computing values of the solar constant are obtained chiefly in 

 the afternoon. Forenoon observations are distinguished in plate xx 

 by being connected by dotted lines. 



In order to fix our ideas both of the magnitude and the variability 

 of the absorption of the earth's atmosphere, the following table, show- 

 ing the percentage of transmission at numerous different wave- 

 lengths for the days indicated, is given. The computations upon 

 which the table is based were made at wave-lengths specially selected 

 to avoid large terrestrial absorption bands, and thus the table gives 

 values of the general absorption only. A few reductions have been 

 made to determine the selective absorption within the numerous 

 atmospheric bands of water vapor and oxygen, but while their dis- 

 cussion has gone far enough to show that equation (i) apparently 

 holds good in these bands, these results are not yet far enough ad- 

 vanced to be included in the tables. While, as another criterion 

 of the accuracy of the method of extrapolation, it is found, in accord 

 with what has just been said, that the employment of these observed 

 values of transmission within the water-vapor bands would prac- 

 tically fill up these bands in computations of the form of the solar 

 energy curve outside the atmosphere, yet in determining the solar 

 constant they are smoothed over and the general transmission con- 

 stants corresponding with the smoothed curves are employed in the 

 computation. 



The days included in Table I were all nearly cloudless, and thus 

 the results represent the transmission of the atmosphere in better 

 than average conditions. In order to bring out clearly what seems 

 to be a marked decrease in the transparency of the air for the present 



