ON OUR KNOWLEDGE OF SPECTRUM ANALYSIS. 287 



on heating changes its vapour-density and its molecular structure from 

 N2O2 to NO without showing an equally marked change in its spectrum. 

 But the exception is apparent only, for the absorption spectrum gets very- 

 much stronger as the temperature is raised, so that we have every reason 

 to suppose that the spectrum we observe is really due to the molecule NO. 

 "We ought at the same time to expect the spectrum belonging to N2O2 to 

 get weaker, and finally to disappear. As, however, it is a general rule 

 that the specti^um of a more complex molecule lies more to the red than 

 that of the simpler molecule, it is likely that the spectrum of NjOg lies in 

 the extreme red and ultra-red. Moser' indeed has found that three 

 bands which are observed in the absorption spectrum of nitrons oxide at 

 low temperatures disappear when the temperature is raised ; these bands 

 therefore we may ascribe to ISToOg, which gas very likely possesses a 

 greater number still further towards the red. 



When chemical combination takes place a change in the spectrum is 

 observed which is entirely similar to that observed in many vapours when 

 the temperature is lowered. The spectra of the oxides, chlorides, bromides, 

 and iodides of the alkaline earths are spectra of fluted bands. If a com- 

 bination of one element with another can change a line spectrum into a 

 channelled- space spectrum, it is quite a plausible assumption to suppose 

 that a combination of one element with itself can produce an analogous 

 change, and that as a rule a spectrum of channelled spaces corresponds 

 to a more complicated molecular structure. 



The question is likely to be definitely settled by a new line of inves- 

 tigation which has been started by Prof B. Wiedemann,^ and which, if 

 followed out further, will largely add to our store of knowledge on these 

 and similar points. Prof. Wiedemann has undertaken calorimetric 

 measurements, in order to see whether or not heat is absorbed by a gas 

 when a change of spectrum takes place. He has taken hydrogen gas as a 

 first example, and measured the heat produced by a spark, equaUsing the 

 same difference of potential, fir.st when the gas gives a band spectrum, 

 and, secondly, when it gives the line spectrum. The change from one 

 spectrum to the other was produced by a minute change in the length of 

 an air- break, which was introduced into the circuit. Prof Wiedemann has 

 found indeed that a certain amount of heat is necessary to change the 

 band spectrum into the line spectrum, and that this heat is independent 

 of the pressure and cross section of the tube. 



These experiments would be decisive if we were quite certain of the 

 chemical origin of the band spectrum investigated by Prof Wiedemann. 

 According to several spectroscopists this band spectrum is due to a hydro- 

 carbon ; so that if this is the case. Prof Wiedemann would really have 

 measured the heat of combination of hydrogen and carbon. Prof. 

 E. Wiedemann will no doubt follow out this most promising line of 

 research in the case of other gases, the spectrum of which has been 

 investigated with more decisive results. 



It is often found that metallic vapours near their point of condensation 

 give an absorption which is continuous through part or the whole visible 

 range of the spectrum. Even some gases like oxygen give us continuous 

 spectra at the lowest temperature at which they are luminous. In the 

 case of sulphur the appearance of the continuous spectrum is coincident 

 with the anormal vapour- density, and is therefore no doubt produced by 



' Fogg. Ann. clx. p. 177 (1877). » Wied. Ann. x. p. 202 (1880). 



