376 PROCEEDINGS OF THE AMERICAN ACADEMY. 



most powerfully dispersed by the vapor, and the two fringe systems 

 which it forms are shifted in opposite directions. It was hoped that a 

 method might be worked out for determining the dispersion of the vapor 

 in the immediate vicinity of the D lines by studying the behavior of 

 sodium light when dispersed in the interferometer, and a great deal 

 of time was lost in the endeavor to unravel the very complicated changes 

 produced in the appearance of the fringe system. If we were dealing 

 with strictly homogeneous radiations, lying to the right and left respec- 

 tively of the absorption band, the two fringe systems would move in 

 opposite directions at the same rate, and we should have simply rapid 

 alternations of visibility and invisibility. By the simultaneous use of a 

 helium tube we could determine the number of disappearances of the 

 sodium fringes corresponding to a shift of one or more helium fringes, 

 which would give us at once a measure of the shift of the system formed 

 by the sodium light. This would give us a measure of the refractive 

 index closer to the absorption bands than we could get by any other 

 method. An attempt was made to utilize the Zeeman phenomenon in 

 this case, the sodium flame being placed in a magnetic field and the 

 interferometer illuminated with one or the other of the D lines by means 

 of the polarizing system. There was no difficulty in finding positions of 

 the interferometer mirror for which the fringes were visible with the 

 field on, and invisible when it was off, but nothing resembling a rapid 

 appearance and disapjiearance of the fringes was observed when the 

 sodium vapor was formed. This is probably due to a number of causes. 

 In the first place the light from the flame is by no means strictly 

 homogeneous, for a pale flame containing only a little sodium and 

 yielding narrow lines does not give light enough. In the second place, 

 as is well known, the D lines do not break up into polarized triplets 

 in the magnetic field, but into a quadruplet and sextuplet. At the time 

 of trying the experiment this circumstance was not thought of, and the 

 failure may have been due in part to the fact that only Do was utilized, 

 which is obviously less suited to the purpose, for the reason that it gives 

 us a double line on each side of the absorption band when the central 

 components are cut out by the nicol. I am of the opinion, however, 

 that the finite breadth of the normal line is chiefly responsible for the 

 failure. Some experiments were made with vacuum tubes containing 

 sodium vapor, which yield highly homogeneous light, but they were 

 found troublesome to manage. It is quite possible that, by the use of 

 these tubes, results could be obtained in the case of Dj, and utilizing the 

 light emitted in the direction of the lines of force (Zeeman doublet) we 



