726 



cither, so that the direction of the issuing beam of light with respect 

 to the liglit arc was a quite different one, I think 1 may conchide 

 that the arc lines are not to be attribuied to anomalous dispersion. 



Nor can for analogous reasons this be the case witli the lines 

 whicli were observed after the bght liad passed through sodium 

 vapour in a uniformly heated glass tube. Here too light lines would 

 have to be observed at some distance from the dark ones, of which 

 there was however, no question. 



On the other hand — as I already remarked just now — when 

 the unsymmetrically heated Wood tube was used. I saw a sharp 

 light line by the side of the dark i-egion, which latter became 

 blacker then at tlie same time; in fact besides the absorbed light, 

 also the anomalously dispersed light has vanished from this region. 



Everything considered I am therefore of opinion that anomalous 

 dispersion has had no iidluenee on my final results. 



I will mention here another phenomenon, which at first made its 

 influence felt in a peculiar way. In my first experimeids I had 

 placed the glass tubes in which the sodium vapour was generated, 

 before the entrance slit of the spectrometer, so that the whole beam 

 of white light passed through it. The measurements whicli I then 

 made of the distances of the components for different tubes, which 

 were distinguished by the thickness of the radiated layer of vapour, 

 w^ere not in harmony; at the same temperature the distance of the 

 components was found larger as the radiated layer was thicker. 

 This peculiar phenomenon must undoubtedly be a consequence of 

 the presence of fluorescence light, which the sodium emits under 

 the influence of the incident white light. According to Wood's 

 researches^) it is just the two D-lines which are very prominent 

 in the fluorescence light. This light will be tlie stronger as the 

 traversed layer is thicker. In this way it is explicable that the 

 absorption spectrum can be subjected to a modification which will 

 become greater with increasing thickness of layer. 



When, however, the distance of the absorption maxima increases 

 in consequence of the superposition of the fluorescence light over 

 the absorption spectrum, which is greatly the case at higher tempe- 

 ratures (see the curves (Dj' and {DJ), it is easy to see that the 

 maximum, resp. the maxima, of the fluorescence light must be 

 situated between the absorption maxima so that the curve representing 

 the intensity of the fluorescence light, exhibits a rise at the place 

 of the absorption maxima, when we mo\e to a point lying halfway 



1) R. W. Wood. Phys. Opt. p. 444; 1905. 



