l6 SMITHSONIAN MISCELLANEOUS COLLECTIONS VOL. 92 



atmospheric agents, the grosser dust particles affecting them nearly all 

 alike, or with a general absorption ; the minuter ones beginning to act 

 selectively, or, on the whole, more at one end of the spectrum than 

 another ; smaller particles, whether of dust or mist, and smaller still, 

 forming a probably continuous sequence of more and more selective 

 action down almost to the actual molecule, whose action is felt in the 

 purely selective absorption of some single ray. 



" The effect of the action of the grosser particles then is to pro- 

 duce a general and comparatively indifferent absorption of all rays, 

 so that the spectrum after such an absorption would simply seem less 

 bright or less hot. The effect of the smaller ones is, as has just been 

 said, to act more at one end of the spectrum than another, with a pro- 

 gressive absorption, so that the quality of the radiation is sensibly 

 affected as well as its quantity. The effect of the molecular absorption 

 is to fill the spectrum with evidences of the selective action in the 

 form of the dark telluric lines, taking out some kinds of light and heat 

 and not others, so that after absorption what remains is not only less 

 in amount but quite altered in kind 



" The writer has demonstrated that in neglecting to observe ap- 

 proximately homogeneous rays we not only commit an error, but an 

 error which always has the same sign, and that the absorption thus 

 found is always too small. He accordingly devoted much time to the 

 construction of an instrument (the bolometer, which will be described 

 in its place) for the special study of such heat rays, and, with this, 

 observations were carried on in the years 1880 and 1881 at Allegheny, 

 with the conclusions which have just been stated. With this instru- 

 ment the heat in some approximately homogeneous ray (that is in 

 some separate pencil of rays of nearly the same wave-length) is mea- 

 sured in the pure and normal spectrum at successive hours of the 

 day, and the calculation of the absorption on Bouguer's principle 

 (justly applicable to strictly homogeneous waves) gives the heat out- 

 side the atmosphere in this approximately homogeneous portion with 

 a degree of approximation, depending on the actual minuteness of the 

 part examined. The process is then repeated on another limited set 

 of rays, and another, until the separate percentage and the separate 

 original heat is found for each heat pencil directly or by interpolation, 

 and then finally the whole heat, by the summing of its parts, the result 

 being that the solar constant is much greater than it was believed to 

 be, and the absorption of the atmosphere much greater. 



" Toward the close of 1880 it had already become clear that the 

 gain in our knowledge by repeating the observations then in prog- 



