Second Spectrum of Hydrogen. 351 



i 



creases as A A increases, if A A > 1 — e~~$. Since, then, for a given 

 wave-length, A A is only a single function of the temperature, 

 which, of whatever form it may be, must at least possess this 

 property — that, within limits determined by its nature, it 

 increases with the temperature — we see that so long as the 



temperature is so low that A A does not attain the value l — e~s y 

 an increase in the thickness of the radiating layer and an 

 increase in the temperature both produce an increase of E ; 

 whilst for those temperatures for which A A exceeds the value 

 mentioned, a further increase of temperature produces the 

 opposite effect to the increase of S. Consequently at high 

 temperatures a spectrum will be considerably less affected by 

 change in the thickness of the radiating layer than at lower 

 temperatures, exactly as shown by the above experiments. 



But we may perhaps go further still, and in these expe- 

 riments obtain even an experimental confirmation of the 

 peculiarity of the function A mentioned above. I am the more 

 disposed to do so, since it appears to me altogether unintel- 

 ligible how a continual rise in temperature can have any other 

 effect, at least so long as no dissociation and consequent alte- 

 ration in the arrangement of atoms in the radiating system, 

 has taken place. If this is admitted, we shall be obliged to 

 assume such a dissociation as the explanation of the displace- 

 ment of a spectrum by a new one with rise of temperature, 

 and must therefore ascribe the first spectrum to a more com- 

 plicated arrangement of molecules, or to a compound of the 

 body with itself. Since, according to the investigations of 

 Wiedemann*, in the case of hydrogen a continual rise in tem- 

 perature produces first a gradual diminution of the spectrum 

 above described, and then upon reaching a certain limit its 

 almost sudden disappearance, these considerations lead us to 

 the view proposed as a second alternative by Angstrom, 

 according to which the spectrum belongs to a compound of 

 hydrogen with itself, The heat-equivalent found by Wiede- 

 mann for the quantity of energy necessary to transform this 

 spectrum into that consisting of the three characteristic bright 

 lines, would therefore be nothing else than the thermal equi- 

 valent of the corresponding work of dissociation. 



If the views thus explained of the spectroscopic conditions 

 of hydrogen are regarded as justified by facts, we have an 

 easy explanation of the fact that in the spectra of the sun and 

 most stars only the characteristic lines of this gas appear as 

 bright lines or absorption-lines, as the case may be, whilst no 



* Wiedemann's Annalen, vol. x. p. 202. 



