CHAPTERS ON THE STARS. 501 



the sun light of certain definite wave-lengths is wholly or partly want- 

 ing. This fact has been observed for more than a century, but its true 

 significance was not seen until a comparatively recent time. 



If, instead of using the light of the sun, we form a spectrum with the 

 light emitted by an incandescent gas, say hydrogen made luminous by 

 the electric spark, we shall find that the spectrum consists only of a 

 limited number of separate bright lines, of "various colors. This shows 

 that such a gas, instead of emitting light of all wave-lengths, as an in- 

 candescent solid body does, principally emits light of certain definite 

 wave-lengths. 



It is also found that if we pass the light of a luminous solid through 

 a sufficiently large mass of gas, cooler than the body, the spectrum, in- 

 stead of being entirely continuous, will be crossed with dark lines like 

 that of the sun. This shows that light of certain wave-lengths is ab- 

 sorbed by the gas. A comparison of these dark lines with the bright 

 lines emitted by an incandescent gas led Kirchhoff to the discovery of 

 the following fundamental principle: 



Every gas, when cold, absorbs the same rays of light which it emits 

 when incandescent. 



An immediate inference from this law is that the dark lines seen in 

 the spectrum of the sun are caused by the passage of the light through 

 gases either existing on the sun or forming the atmosphere of the earth. 

 A second inference is that we can determine what these gases are by 

 comparing the position of the dark lines with that of the bright lines 

 produced by different gases when they are made incandescent. Hence 

 arose the possibility of spectrum analysis, a method which has been ap- 

 plied with such success to the study of the heavenly bodies. 



So far as the general constitution of bodies is concerned, the canons 

 of spectrum analysis are these: 



Firstly, when a spectrum is formed of distinct bright lines, the 

 light which forms it is emitted by a transparent mass of glowing gas. 



Secondly, when a spectrum is entirely continuous the light emanates 

 from an incandescent solid, from a body composed of solid particles, 

 which may be ever so small, or from a mass of incandescent gas so large 

 and dense as not to be transparent through and through. 



Thirdly, when the spectrum is continuous, except that it is crossed 

 by fine dark lines, the body emitting the light is surrounded by a gas 

 cooler than itself. The chemical constitution of this gas can be deter- 

 mined by the position of the lines. 



Fourthly, if, as is frequently the case, a spectrum is composed of 

 an irregular row of bright and shaded portions, the body is a compound 

 one, partly gaseous and partly solid. 



It will be seen from the preceding statement that, in reality, a mass 

 of gas so large as not to be transparent cannot be distinguished from a 



