62 CORRELATION OF PHYSICAL FOECES. 



low light of such purity that other colours exposed to it ap 

 pear black a phenomenon shown by the familiar experiment 

 of exposing a picture of bright colours, other than yellow, to 

 the flame of spirits of wine with which common salt is mixed : 

 the picture loses its colours, an.d appears to be black and 

 white. When the prismatic spectrum of such a flame is ex 

 amined, it is found to exhibit two bright yellow lines at a cer 

 tain fixed position. If a source of light be employed which 

 gives no lines in its spectrum, and this, being at a higher 

 temperature, be made to pass through the sodium flame, two 

 dark lines will appear in the spectrum precisely coincident 

 in position with the yellow lines which were given by the so 

 dium flame itself. The same relation of absorption to radia 

 tion is therefore shown here : the substance absorbs that 

 light which it yields when it is itself the source of light. 

 The same is true of other substances, the spectra of which 

 exhibit respectively lines of peculiar colour and position. 

 Now, the solar prismatic spectrum is traversed by a great 

 number of dark lines ; and Kirchoff has deduced from con 

 siderations such as those which I have shortly stated, that 

 these dark lines in the solar spectrum are due to metals ex 

 isting in an atmosphere around the sun, which absorb the 

 light from a central incandescent nucleus, each metal absorb 

 ing that light which would appear as a bright line or lines in 

 its own spectrum. 



By comparing the position of the bright lines in the spec 

 tra of metals with that of the dark lines in the solar spec 

 trum, several of them are found to be in identically the same 

 place : hence it is inferred, and the inference seems reason 

 able, that the metals which show luminous lines in their spec 

 tra, identical in position with dark lines in the solar spec 

 trum, exist in the sun, and are diffused in a gaseous state in 

 its atmosphere. It does not seem to me necessary to this 

 conclusion to assume that the sun is a solid mass of incandes 

 cent matter : it may well be that what we term the photo- 



