ON OUR KNOWLEDGE OF SPECTRUM ANALYSIS. 285 



The same idea was more fully developed by B. Wiedemann.' Even 

 Angstrom and Thalen,^ who do not adopt Pliicker's interpretation of his 

 experiments, remark : — 



' We do cei'tainly not deny that a simple body may in certain cases 

 give different spectra. We may quote, for instance, the absorption- spec- 

 trum of iodine, which does not in any way resemble the system of bril- 

 liant rays of the same body, obtained by means of electricity ; and Ave 

 may remark moreover, that in general every simple body, presenting the 

 property of allotropy, must in the state of incandescence give different 

 spectra, provided that this property of allotropy exists not only in the 

 gaseous state of the body, but also at the temperature of incandescence. 



' Supposing, therefore, that there is really allotropy, even for the 

 gaseous state, a certain absorption-spectrum must belong to every one of 

 these allotropic states.' 



An idea which has commended itself to so many different observers 

 must possess a large amount of plausibility ; but before we give various 

 reasons and facts which seem to support it we must first refer to the only 

 rival hypothesis which has been offered. This theory is founded on the 

 formula given by Zdllner, and already quoted in this report, which con- 

 nects the intensity of radiation of each line with the number of radiating 

 molecules in a line of sight. Zollner already mentioned the possibility 

 of explaining spectra of different orders by means of it, but the idea was 

 chiefly developed by Wiillner. Starting from the fact that band-spectra 

 are generally given by the brush discharge, in which a great number of 

 molecules are luminous, while the line-spectra are given by the spark, 

 which, as a rule, is thin ; also that the band-spectra often appear in the 

 wide part of a Pliicker tube, while the line-spectra are seen in the capil- 

 lary part, Wiillner concludes that the thickness of the radiating layer 

 materially affects the spectrum which is seen. The difficulty which 

 stands in the way of this explanation lies in the fact that the maxima of 

 light in the band-spectrum lie altogether in other places than in the line- 

 spectra. To overcome this difficulty, as has been pointed out by B. 

 Wiedemann,^ we must assume a change in the emissive power with tem- 

 perature different for each ray of the spectrum. Zollner,* for instance, 

 remarks : — 



' But these values (the coefficients of absorption) may have for the 

 same toave-length and continuous alteration of the temperature, similar 

 maxima and minima to those which they, in fact, possess for the same tem- 

 perature and continuous alteration of the wave-length, whereby they pro- 

 duce the phenomenon of discontinuous spectrum.' 



Wiillner,^ referring to the same point, says : — 



' Hereby it is by no means necessary or even probable, that the ab- 

 sorption power increases for all rays in the same way ; that, therefore, the 

 ratio of the values of and for different wave-lengths is the same at all 

 temperatures. As soon as such a change takes place, a displacement of the 

 maxima of light will, or at least may, be produced. But it is only such 

 a displacement of the maxima of light, if the bright lines of a line-spec- 

 trum are situated at other places than the maxima of illumination of a 

 band- spectrum. In this way lines of a line- spectrum can really disappear, 



' Wied. Ann. v. p. 500 (1878). •* PMl. Mag. xli. p. 199 (1871). 



2 Aov. Act. Uj>s. ix. (1875). ' Wied. Ann. viii. p. 594 (1879). 



' Wied. Ann. x. 202 (1880). 



