of the Centres of Emission of I) Lines of Sodium, 1023 



We finally substituted a large constant deviation spectro- 

 graph for the smaller instrument, as the latter barely resolved 

 the lines, and it was often difficult to be sure of what we had 

 on the plate. 



This instrument showed very clearly that the D 2 centre of 

 emission could be set in vibration without disturbing the D x 

 centre, in other words we can have sodium vapour emitting 



one D LINE ONLY. 



Photographs made of the spectrum of the resonance 

 radiation excited by both sodium lines (with the polari zing- 

 prisms and quartz block removed) showed that the D lines 

 had the same intensity, in some cases Dj even appearing 

 slightly brighter than D 2 . It was found that if the amount 

 of. sodium in the flame was reduced to the least amount 

 consistent with having resonance radiation of sufficient 

 intensity to photograph, the D 2 line came out stronger than 

 D x in the spectrum of the latter, as is always the case with 

 the sodium flame. Exciting the vapour by the light of a 

 bright soda-flame gave a resonance radiation in which D x 

 came out stronger than D 2 , which is never the case with 

 the flame. This is due to the circumstance that with a 

 bright flame, D 2 is more or less reversed, hence it is less 

 effective in exciting the resonance, for the vapour in the glass 

 bulb responds only to the wave-length forming the core of 

 the line. Dx is less easily reversed, and may consequently 

 be more efficient in exciting resonance. 



On PI. XVII. fig. 1, we have a photograph of the single line 

 (D 2 ) emitted by the resonating vapour, the greater intensity 

 at the top being due to the light from the horizontal narrow 

 strip of magnesia. Immediately below this (fig. 3) we have 

 the two D lines as emitted by the flame. Fig. 2 shows the 

 spectrum of the resonating vapour when excited by both 

 sodium lines from a strong flame, and we find D x , which is to 

 the left, slightly brighter thau D 2 . In fig. 4, the resonance 

 was excited by a feeble flame and D 2 is brighter than D x . 



This change in the ratio of intensity of the two D lines of 

 the resonance radiation leads to some curious results which 

 were somewhat puzzling at first. 



It will be remembered that the central patch of light 

 furnished by the polarizing separator results from the super- 

 position of two images, and, other things being equal, will 

 have double the luminosity of the two lateral patches adjacent 

 to it. If the light of the D 2 line, which is brighter than D t 

 in the flame, is thrown into the central patch, we should 

 expect it to be more than twice as bright as the lateral 

 patches adjacent to it. It was observed, however, that the 



