[PLATES 1-3.]
Experimental proof of the dispersion formulae of Sellmeier, Helm- holtz, and Ketteler has for the most part been based upon observations made upon substances in the solid state or in the state of solution, where the range of absorption extends over a considerable portion of the spectrum, and the molecular condition is in all probability vastly more complicated than in the gaseous state.
In liquid oxygen and in the salts of erbium and didymium we have the only examples, so far as I know, of substances other than gases which exhibit narrow absorption bands, and in these cases the absorp- tion is scarcely strong enough to influence the dispersion to any marked degree. A careful study of the dispersion of some substance with absorption bands as narrow and strong as those of sodium vapour would be of great value in aiding us to pass judgment on the various modifications of Sellmeier's original theory.
The anomalous dispersion of sodium vapour in the immediate vicinity of the D lines was first observed by Kuridt, and has since been studied by Becquerel and Julius, but none of these investigators has traced the effect over a range of the spectrum greater than two or three times the distance between the lines.
In a previous paper* I have shown that a spectrum closely resem- bling the bright-line spectrum of incandescent sodium can be pro- duced under certain conditions by the anomalous refraction of white light in a non-homogeneous atmosphere of sodium vapour, in precisely the manner in which Julius very ingeniously suggests that it may occur in the reversing layer of the sun, giving rise to the phenomenon of the " flash spectrum." In view of the seeming importance of the subject in its bearing on the subject of the dynamics of dispersion, and its possible connection with solar-physics, I have undertaken to determine whether sufficiently accurate quantitative data can be obtained, to make a test of the dispersion formula, as applied to sodium vapour, possible.
Though the absolute values of the refractive indices which have been found are not as satisfactory as is desirable, very interesting relative values have been obtained, and >vhat is of especial interest, the dispersion has been traced and measured throughout the entire range of the visible spectrum, the refractive index for all waves of
- 'Phil. Mag.,' May, 1901.
