﻿10 SMITHSONIAN MISCELLANEOUS COLLECTIONS [VOL. 47 



phere by employing' Bouguer's formula. The absorption still remain- 

 ing is indicated by the difference between the transmission so found 

 as given in the sixth column, and unity. As would be expected from 

 the sources of error indicated in discussing the first method of obtain- 

 ing the transmission coefficients, the value of the coefficient obtained 

 by varying the air mass, //;, is generally greater than that obtained 

 by varying the density, 3, of the water vapor. 



Several reductions with the air mass, m, varying are included in 

 the tabic for April 17 and 29 of the discordant group of days pre- 

 viously mentioned. It will be observed that the difference between 

 the values in the third and fourth columns is here greater than usual. 

 Now, assuming that the values of the transmission coefficients of 

 these different bands obtained from the observations of all the davs 

 but those of the discordant group by varying the density are correct, 

 these values (a) and the corresponding values of a 1 and a 2 may be 

 substituted in equation 5, page 9, to determine new values of 8 for 

 the days in question. These, reduced to an air-mass of 1.25 atmos- 

 pheres for comparison with plates iv and v, are : 



April 17, 1903, 1.3 1, 1.24, 1.36, mean 1.30 

 April 29, 1903, 2.41, 2.41, 1.94, mean 2.25 



If these values had been used for the corresponding points in plates 

 iv and V, the data for these dates would not then have been dis- 

 ci irdant. 



Jn plate v are plotted the values for the transmission for the bands 

 />' and A, due to oxygen, and for the bands at 1.285 /j, am ' w 2> or 

 unknown atmospheric origin. There are also plotted the ratios of 

 tlu- ordinate's at the maxima between <I> and W, and between W and 

 Q , to those between p and 0. All of these plots are apparently inde- 

 pendent of water vapor. It is improbable that any conclusion can 

 be drawn as to the origin of the bands at I.285 ju, and w 2 from com- 

 parison with the plots for the oxygen bands. The irregularities in 

 ./ arc probably due to its great depth and narrowness, so that the 

 ordinate at its bottom depends very greatly on the instrumental con- 

 ditions. 



RELATION OF ABSORPTION TO DENSITY OF AliSOKliENT 



\ .ry little is yel known of the change of absorption of a vapor 

 in passing to the liquid form, except from measures of the total 

 absorption over the whole spectrum. Those oxygen bands which 

 have an absorption varying with the square of the density, still seem 



