256 PROCEEDINGS OF THE AMERICAN ACADEMY. 



Unfortunately the condition is a difficult one to keep fixed, for the phe- 

 nomenon only appears when the density of the sodium vapor is just 

 right and the surrounding vacuum high. As I have shown in the 

 paper on the dispersion of sodium vapor, we can have a dense mass 

 of the metal vapor, bounded on each side by a very high vacuum, a 

 very anomalous condition from the p ant of view of the kinetic theory 

 of gases. My impression is that the green spot will show the fluores- 

 cent spectrum, and the yellow spot the lines of the principal and sub- 

 ordinate series, as found in the sodium arc, but as yet I have not found 

 time to make even a visual examination. Several attempts have been 

 made, but by the time the image of the spot was thrown upon the slit 

 in the proper direction to pass the light through the prisms, and the 

 eye brought to the instrument, the conditions in the tube changed. 



It is difficult to account for the absence of luminosity of the 

 centre of the mass and the two bright spots. Perhaps the condition 

 under which the rays excite fluorescence exists only where the vapor 

 mass is in contact with the vacuum, i. e. in the region where the h}^o- 

 thetical clusters of molecules are breaking up and flying to the cooler 

 walls of the tube. Even assuming this to be the fact, the difference in 

 the color of the two spots is still to be accounted for. Possibly the 

 cathode rays excite the green spectrum, while the canal rays travelling 

 towards the cathode excite the orange-yellow luminescence. I have 

 made one experiment with a similar tube arranged so as to deliver 

 a stream of canal rays against the vapor. The luminescence was 

 bright yellow, but the tube cracked before a spectroscopic examination 

 was made. 



On the other hand, it may be that whatever causes the green lumin- 

 escence is removed from the ray bundle by absorption, the residue 

 exciting the yellow luminescence at the point of exit. If this is the 

 case, we should expect the same amount of yellow light in each spot, 

 and I am of the opinion that the green light is much too pure for this 

 to be the case. Further experimenting will be necessary before it is 

 possible to draw any very definite conclusions. 



In the spectrum excited by the cathode rays the D lines are of 

 immense brilliancy, running together into a single band of light. On 

 each side of this are seen three or four symmetrically placed bands, 

 decreasing in brilliancy as they recede from the D lines in each 

 direction. No trace of these bands appears in the magnetic spectrum, 

 which in this region shows only fine lines arranged in narrow groups, 

 which do not coincide with the bright bands of the cathode lumin- 

 escence. 



A photograph of these bands is reproduced in Plate 4, Figure 4. 



