mirror of molten sodium, I find that the spectrum consists of two exceedingly brilliant narrow green bands with a narrow dark line between them, and a band in the violet which is partially filled with flutings. This spectrum I have observed but once, and have given it no careful study as yet.
The flutings in the green-blue portion of the spectrum are similar in appearance to those in the red, but they make their appearance when the vapour is less dense, this part of the spectrum being blotted out entirely when the red bands are at their best.
I have not yet investigated the question of whether an increase in the thickness of the layer of vapour is the equivalent of increasing its density, but I feel quite sure that such is not the case. There remains too the question of pressure. All of the experiments that I have made thus far have been at atmospheric pressure, the variations in density being in all probability merely dilution of the metallic vapour with hydrogen.
Sodium vapour .possesses another property which requires further investigation, namely, the power of emitting light in virtue of tem- perature alone. This was first observed by Evershed.* Iodine and a number of other substances behave in a similar manner, the tem- perature necessary to produce the luminous radiations being but little above a red heat. On repeating Evershed's experiment with iodine, my first impression was that anomalous dispersion might be at the bottom of it. The vapour was heated in a tube, and the purple lumi- nosity or glow hung close to the inner wall, which was red hot. It seemed quite possible that the phenomenon might be a mirage effect, the rays of suitable wave-length coming from the red-hot wall moving in curved paths through the non-homogeneous vapour. To test this I devised a method by which this possibility was entirely eliminated, and found that the glow was more brilliant than ever. A spiral of platinum wire mounted in a small flask in which iodine was vaporised was raised to a bright red heat by an electric current. The convection current of iodine which rose through the spiral was luminous to a height of about 2 cm. above the spiral, waving from side to side in the form of a reddish-purple flame. Focussing the image of an electric arc on the spiral increased its luminosity ten-fold, with- out, however, altering the intensity of the glowing vapour. Probably the luminescence of sodium could be shown in the same way, though the one experiment that I tried with it gave negative results.
I expect in the near future to investigate all of the questions which I have touched upon in the latter part of this paper, for it appears to me that, with proper coaxing, sodium vapour may furnish us with much information regarding molecular dynamics. The results obtained up to the present time are only a small beginning, but appear to indi-
- ' Phil. Mag.,' vol. 39, 1895.
