WOOD. — ANOMALOUS DISPERSION OP SODIUM VAPOU. 375 



the vapor in one of the two paths, but on covering up the other path it 

 v/as found that the field of the instrument was still brightly illuminated. 

 On turning the screw of the instrument the fringes appeared again, 

 and a little further experimenting showed that the effect of the vapor 

 was to alter the visibility curve, its {)eriodicity becoming less as the 

 density of the vapor increased. No satisfactory explanation of the 

 altered appearance of the fringe system could be found, and an attempt 

 was made to get rid of one of the D lines in the illuminating light, 

 thus simplifying the conditions. Prisms and gratings were ti-ied, but 

 the loss of light was so great that even with an oxyhydrogen-sodium 

 flame the fringes could be made out only with the greatest difficulty. 

 The desired result was finally obtained by mounting a quartz crystal 

 between two crossed nicols. ~ A number of crystals were examined with 

 polarized light perpendicular to the axis, and one selected which showed 

 dark bands in the spectrum, separated by a distance about double the dis- 

 tance between the D lines. The nicols were arranged so that the emer- 

 gent light vibrated in such a plane as to be most copiously reflected by the 

 interferometer mirror, i. e. the first nicol was mounted with its short 

 diagonal horizontal, while the one next to the instrument was placed 

 with its long diagonal horizontal. Between the nicols was mounted 

 the (juartz crystal with its axis making an angle of 45 degrees with the 

 vertical. The light from an oxyhydrogen-sodium flame, after passage 

 through the polarizing system, was brought to a focus on a cardboard 

 screen by means of a lens, forming a series of bright bands separated by 

 dark intervals. A narrow slit in the screen allowed light corresponding 

 to Dj or Dj to enter the interferometer, according as it was set on one 

 edge or the other of one of the bright bands. This arrangement worked 

 admirably and yielded an abundance of light. No trace of the periodic 

 invisibility was found in the fringe system, such as is always observed 

 when ordinary sodium light is used. As soon as the smallest trace of 

 sodium vapor was formed in the tube the fringes were found to have 

 become quite invisible in certain parts of the system. The visibility 

 curve changed rapidly as the density of the vapor increased, the period 

 becoming smaller. The general behavior of the fringes did not differ 

 materially from that observed when both sodium lines were used, prov- 

 ing that the presence of the two wave-lengths was not essential. The 

 appearance of fringes under the conditions of the experiment depends 

 upon the fact that the emission line is broader than the absorption line. 

 The absorption is restricted to the centre of the line, the edge light only 

 getting through the sodium tube and forming fringes. This edge light is 



