364 The Principles of Spectrum Analysis. 



rays have such, a refrangibility that our eyes cannot take 

 cognizance of them ; and so also certain rays exist in solar light 

 which are incapable of transmission through certain media. 

 Applying this to the solar spectrum, we have a clue to the 

 production of the dark lines, by supposing, with Kirchhoff and 

 Bunsen, that the sun has the property of inducing or giving* 

 out rays of a certain refrangibility, but yet cannot produce 

 others capable of filling the interspaces. Another interpreta- 

 tion has been given, by supposing an interference in the undu- 

 lations of certain rays, which produce darkness ; but this theory 

 will not meet the circumstances of the case, and we can show 

 by experiment that by making an artificial atmosphere the same 

 or parallel results can be produced. 



The natural variations in the composition of the atmosphere 

 produce similar effects, and Brewster was the first to notice 

 bands in the red and green spaces, whose appearance was not 

 constant. These appearances are usually observed when the 

 sun is not far from our horizon ; and Dr. Miller mentions an 

 instance in which he saw a group of lines during a thunder 

 shower. They came suddenly, and faded as the rain passed away. 



The readiness with which the spectrum responds to changes 

 in the atmosphere, or in the nature of the source of light, is 

 shown in the following experiment of Kirchhoff and Bunsen : — 



They threw up into the air of the apartment a small quantity 

 of chloride of sodium in very fine powder. Motion was imparted 

 to the atmosphere, to ensure an equable diffusion of the salt. 

 The spectrum in an instant demonstrated its presence, by show- 

 ing a golden-yellow band in the yellow space. This effect is 

 uniform whenever sodium is present in a state of incandescence,, 

 and is therefore called the sodium spectrum. This result 

 might have been expected, knowing that sodium in any form 

 always tinges flame an intense yellow ; but when we come to 

 the combustion of other metals, the bands produced by them 

 are such as could never have been anticipated. When silver 

 is burnt we have other coloured bands brought out equally cha- 

 racteristic, and so with every other metallic substance. To show 

 the relation between these coloured bands and the dark lines, we 

 will suppose the light of a pure white flame to be passed through 

 a yellow sodium flame, and then through the prism. Now, 

 mark the change. The spectrum is no longer continuous, and 

 having its bright yellow band in the yellow space ; but where 

 it flashed out so conspicuously is now to be seen a dark line, 

 known as " D." The rationale is obvious. The yellow atmo- 

 sphere has interfered with the yellow of our normal spectrum, 

 and by that interference darkness has resulted. 



Keeping these results in view, we have a key to the wholo 

 subject of spectrum chemistry. It can be shown that each 



