﻿SMITHSONIAN MISCELLANEOUS COLLECTIONS [vol. 47 



It should not be inferred from this table that the values of the 

 solar radiation outside the atmosphere, determined here, are too low. 

 In solar constant determinations the smoothed curve alone is used, 

 and. as just indicated, the deficiency in plates 11 and in is probably 

 marly wholly due to the errors of observations in the bands. 



The determinations of the transmission coefficients by this method 

 are tabulated in the next section. 



Si 1 ond .Method of Determining the Transmission Coefficient 



The second method of applying and testing the formula is to 

 observe the sun at the same altitude or air mass on days of different 

 densities, d, of water vapor. Since it has been impracticable to 

 1 ibserve the sun always at such a standard altitude or air mass, which 

 in the present discussion has been taken as one and one-quarter 

 atmospheres, the observations have required a small correction by 

 means of Bouguer's formula, in order to reduce them to this standard 

 air mass. The corrections, in general, have been small. The observa- 

 tions, reduced according to this second method, are plotted in plates 

 iv and v, where the abscissas are the equivalent layers of water vapor 

 present in the path of the beam, and the ordinates the logarithms of 

 the fractional transmission. As these observations have been taken 

 with a comparatively high sun, they are not so much subject to the 

 sources of error mentioned in the previous section, although without 

 doubt they are still affected by them. 



The observations seem to conform remarkably well with the 

 formula, if we except the group including April 17, 28, 29, and Sep- 

 tember 11. It may be seen from Table 11, that on April 17 there 

 were many cumuli ; on April 28 a change in the wind during the latter 

 part of the observation ; and on September 11 there were many cirri. 

 The constants of the apparatus might have been different on those 

 days, rendering the purity of the spectrum less, but although the 

 appearance of the plot for the very deep and narrow A line may seem 

 to support this view, that of <•>.. does not. Possibly during the 

 months of increasing- water vapor in the atmosphere there is some 

 lag of the upper air in conforming with Hann's equation for the 

 iun1 of water vapor present. In drawing the straight line rep- 

 resenting the data, this discordant group has been neglected. 



I RANSMISSION coefficients 



1 he vain.- of die t rai i si i ii ssion coefficient a, of one centimeter of 

 liquid water in vapor form, may be determined in two ways: First, 

 referring to plates n and in. if </, is the transmission coefficient ob- 



