286 BEPORT— 1880. 



while at their place in the band-spectrum an even illumination or even a 

 decreased brightness exists compared to that of surrounding places.' 



We have introduced these quotations in order to show that both 

 Zollnerand Wiillner are aware that a mere alteration in density or thick- 

 ness of the radiating layer is insufficient to account for the changes in 

 spectra, but that a change in the absorptiA'^e and emissive properties of 

 the gas is necessary. But it is further evident that once we assume this 

 change we need not any more have recourse to any effects of increased 

 number of radiating molecules ; for the change itself would be sufficient 

 to account for the phenomena, which could at most be affected, but not 

 produced, by an increased thickness of the radiating layer. We may 

 therefore jjass over the experiments which Dr. Goldstein ' has made in 

 order to show that an increased thickness of spark does not produce the 

 effects required by Wiillner's theory ; and, consequently, also over the 

 answer which Wiillner ^ has given to Dr. Goldstein's remarks. 



The difference between the two explanations really comes to this : 

 Zollner and Wiillner assume that the i-adiation for a given wave-length is 

 a continuous function of the temperature, which may be different for 

 different wave-lengths, and for each may have maxima and minima ; in 

 other words, that the spectrum is a continuous function of the temperature. 

 Those who adopt the rival hypothesis hold that the spectrum is within 

 wide limits independent of the temperature, but that at certain points 

 sudden changes in the forces which bind the atoms together take place, 

 and that these changes are accompanied by a sudden change in the 

 spectrum. It is likely that these changes are produced by a different 

 number of atoms bound together in one molecule. 



The change of the channelled-space spectrum of sodium to the line 

 spectrum seems indeed to take place within narrow limits of temperature. 

 If the absorption of that vapour is observed at low temperatures, so that 

 the channelled-space spectrum is observed, and the temperature is 

 gradually raised, little change is seen for some time. The D lines, though 

 present, are sharp and faint. When a certain temjjerature is reached, 

 however, these lines seem suddenly to get blacker ; the bands at the same 

 time weaken, and as they become dimmer and finally disappear, the whole 

 energy of the motion seems to be thrown into the D lines, which now are 

 widened into a broad black band. The change is exactly such as would 

 be produced by a change in the molecular structure of the gas at the given 

 temperature, and it is very likely that a change of density will be proved 

 to take place at that temperature. 



Those who adopt this view of the cause of multiple spectra, support it 

 by the fact that in many cases where we know a change of density to occur, 

 changes in the spectra are observed perfectly analogous to those which 

 we want to explain. Thus, for instance, sulphur vapour near its boiling 

 point has an anormal vapour-density ; it then shows a continuous absorp- 

 tion ; as the temperature is raised and the density becomes normal, the 

 spectrum changes into a channelled-space spectrum. Iodine, bromine, 

 and chlorine give us spectra of fluted bands at low temperatures ; but if 

 we pass an electric spark through them, line spectra are obtained. The 

 recent experiments of Victor Meyer and others seem to show that the 

 vapour density is different in the two cases. 



An apparent exception occurs in the case of nitrous oxide gas, which 



' Bcrl. Btr. 1874, August, p. 593, and Phil. Mag. xcix. p. 333 (1875). 

 ' Berl. Ber. 1874 (December) and Phil. Mag. xlix. p. 448 (1875). 



