in Incandescent Sodium Vapour. 55 



it. If the spoon is brought exactly into the middle of the 

 flame of the Bunsen burner, it is easy to maintain the flame 

 above it as a cone of intensely yellow brightness. Now this 

 cone acts like a prism with horizontal refracting angle turned 

 upwards. Therefore, if the incandescent sodium vapour ex- 

 hibits a dispersion, this cone of rays, which pass horizontally 

 through it, must give a vertical spectrum (impure, it is true, 

 on account of the conical shape). If the rays pass simulta- 

 neously through a glass prism with a vertical and the sodium 

 prism with horizontal refracting angle, a spectrum is obtained 

 which, if dispersion is present in the vapour, must have the 

 shape above delineated*. As the refracting angle of the 

 sodium prism lies above, the index of refraction of the vapour 

 must be the highest for those rays which are most deflected 

 downwards. The drawing shows that, in accordance with my 

 investigations on solid bodies and liquids, the index of refrac- 

 tion rises much as the absorption-bands of the red side of the 

 spectrum are approached, is lower on the green side of the 

 dark line than on the other, and then rises again rapidly. 



After the phenomenon was once recognized, I very often 

 repeated the experiment ; and when a very regularly conical 

 sodium-flame of great intensity can be obtained, the anomaly 

 in the refraction is very considerable. I have also, instead of 

 sketching the phenomenon objectively on a screen, observed 

 it subjectively with the telescope. 



The above experiment, however, is successful only when the 

 intensity of the sodium-flame is very great, such as is obtained 

 by burning metallic sodium, and that for the following 

 reason : — While the sodium-flame obtained by introducing a 

 salt of sodium into the flame of a Bunsen burner, examined 

 spectrally, shows two bright lines (the two D lines), the phe- 

 nomenon is changed when a piece of sodium of the size of a 

 pea is put into the burner. At first the two D lines come out 

 distinctly; then, when the sodium begins to be vaporized in 

 greater quantity, these lines widen considerably; with still 

 greater density of the vapour they blend into one; and finally, 

 upon this broad yellow band with fainter margins, there usu- 

 ally appear two fine black lines corresponding to the D lines. 

 These dark lines are produced by the absorption of the cooler 

 sodium vapour surrounding the bright sodium-flame. These 

 phenomena have already been observed by Hankel f and Cia- 

 micianj, and perhaps also by others. 



The absorption-power changes correspondingly to the emis- 



* Pogg. Ann. vol. cxliv. pp. 128-137 (1871). 

 t Berichte der Leipziger Akademie, 1871, p. 307, 

 % Wicn. Bar. lxxviii. (1878), p. 887. 



