WOOD. — ANOMALOUS DISPERSION OP SODIUM VAPOR. 387 



column of sodium vapor terminates in a vacuum, if the expression is 

 allowable. If the sodium were heated in a tube provided with end 

 plates which were not acted upon by the vapor, so that the whole could 

 be brought to a constant temperature, it is probable that agreement with 

 the results obtained by Jewett would be obtained. As it is, we have 

 a cold condenser at each end of the column, and only obtain vapor of 

 any considerable density through the circumstance that the rate of 

 diffusion is comparatively slow. 



I hope in the near future to determine the actual density under the 

 conditions of the experiment. 



The density at a given temperature was found to depend to a large 

 extent upon the previous treatment of the tube. If the tube was heated 

 for the first time to a temperature of say 350°, a shift of say ten helium 

 fringes would be observed. If the tube was then raised to a red heat 

 and allowed to cool, on reheating it to the same temperature (350°); a 

 shift of perliaps fifteen fringes would be found. It appeared as if 

 the density at a given temperature was much greater if a sodium "dew " 

 had been formed over the entire inner surface of the tube, by the con- 

 densation of the dense vapor given off at the high temperature ; in other 

 words, the density at a given temperature was proportional to the surface 

 of molten sodium which was exposed. 



The determinations were made in glass tubes for the reason that in 

 steel tubes the transition from the heated to the cool portion of the tube 

 was not sufficiently sudden to insure uniform density in the cylinder of 

 vapor. The Jena glass tubes could, however, be raised to a full red heat 

 by means of the coil of iron wire, without showing any trace of sodium 

 vapor in the portions protruding from the helix. 



In order to obtain consistent results it was found necessary to heat 

 the tube very gradually, in order to allow the thermo-element time to 

 take the temperature of the vapor. One observer watched the scale uf 

 the galvanometer, the other counted the fringes as the tube was heated 

 or cooled. The helium fringes disappeared after a shift of something 

 over a hundred fringe-widths liad occurred, but by dropping a plate of 

 plane parallel glass into the air path of the interferometer they could 

 be made distinct again and the count continue<l. In this way it has been 

 possible to record shifts as high as four hundred fringe-widths. Beyond 

 this point it is impossible to go with the the helium light, the components 

 of helium light becoming too much spread out by the dispersion of the 

 vapor to be brought together a^ain l)y going out any farther in the 

 fringe system. To extend the observations to higher temperatures and 



