206 Professor Dewar [June 6, 



By this means the characteristic lines of the volatile metals thal- 

 lium and indium may be easily reversed. 



Metallic lithium, alone or mixed with sodium, gave no results. 

 Similarly, chloride of lithium and metallic sodium, introduced together, 

 gave no better results. To a tube containing mixed potassium and 

 sodium vapour, lithium chloride was added. Now the bright-red 

 lithium line was sharply reversed, and remained well defined for a 

 long time. The lithium line was only reversed in a mixture of the 

 vapours of potassium and sodium, and it seems highly probable that 

 a very slightly volatile vapour may be diffused in an atmosphere 

 of a more volatile metal, so as to secure a sufficient depth of vapour 

 to produce a sensible absorption. This would be analogous to well- 

 known actions which take place in the attempt to separate organic 

 bodies of very different boiling points by distillation, where a sub- 

 stance of high boiling point is always carried over, in considerable 

 quantity, with the vapour of a body boiling at a much lower tempera- 

 ture. 



Sodium and potassium, when observed in such tubes, give none of 

 the appearances noted by Lockyer, " On a New Class of Absorption 

 Phenomena,'* in the ' Proceedings of the Koyal Society,' vol. xxii., but 

 the channelled-space spectrum of sodium described by Roscoe and 

 Schuster in the same volume of the ' Proceedings ' was often seen. 

 Potassium gives no channelled-space absorption, but continuous 

 absorption in the red, and one narrow absorption band, with a wave- 

 length of about 5,730, not corresponding with any bright line of that 

 metal. 



The absorption spectrum of sodium vapour is by no means so 

 simple as has been generally represented. The fact that the vapour 

 of sodium in a flame shows only the reversal of the D lines, while the 

 vapour, volatilized in tubes, shows a channelled-space absorption cor- 

 responding to no known emission spectrum, appears to be a part of a 

 gradational variation of the absorption spectrum, which may be induced 

 with perfect regularity. Experiments with sodium exhibit the follow- 

 ing succession of appearances, as the amount of vapour is gradually 

 diminished, commencing from the appearance when the tube is full of 

 the vapour of sodium, part of it condensing in the cooler portion 

 of the tube, and some being carried out by the slow current of 

 hydrogen. During this stage, although the lower part of the tube 

 is at a white heat, as long as the cool current of hydrogen displaced 

 metallic vapour, on looking down the tube it appeared perfectly 

 dark. The first appearance of luminosity is of a pui-ple tint, and, 

 with the spectroscope, appears as a faint blue band, commencing 

 with a wave-length of about 4,500, and fading away into the violet. 

 Next appears a narrow band in the green, with a maximum of light, 

 with a wave-length of about 5,420, diminishing in brightness so 

 rapidly on either side as to appear like a bright line. This green 

 band gradually widens, and is then seen to be divided by a dark band 

 with a wave-length of about 5,510. Ked light next appears, and 



