1879.] 



Quantitative Spectroscopic Experiments. 



487 



or if the amount of sodium present was not too nmch, D came 

 -out bright. On letting off the pressure, the phenomena recurred 

 in the reverse order, and the whole could be repeated several 

 times. After compression as long as the pressure was sustained the D 

 absorption remained permanently narrowed, but did not continue 

 bright. When there was only a very little sodium in the platinum 

 tube and the pressure diminished, the D band was seen to expand and 

 become diffuse, contracting again as the pressure was restored, but 

 not generally contracting to quite its former width or sharpness. 



We used sometimes amalgams of sodium in order to have a very 

 dense gas to retard diffusion, and for the purpose of getting various 

 quantities of sodium diffused through equal spaces. With these 

 amalgams the general course of the phenomena was much the same as 

 with the metal alone, except that we could not by compression to 

 three atmospheres make the D lines show out so bright as with 

 sodium alone, although the absorption was reduced to extremely 

 narrow dimensions. 



One set of experiments was made with a series of amalgams con- 

 taining decreasing percentages, all less than 1 per cent., of sodium, 

 equal quantities of each were put into the hot tube in succession, the 

 vapour of each being blown out before the next was put in. The 

 thickness of the layers observed was thus very nearly constant, but the 

 density of the sodium vapour variable. It was calculated that the 

 mercury vapour would, at the temperature of the tube, fill a height 

 of about 20 centims. The D absorption was very broad, with 

 wide diffuse edges, with an amalgam of \ per cent., and did not 

 sensibly diminish in width until the percentage of sodium was re- 

 duced to \. 



The results of the foregoing experiments may have been complicated 

 by the sodium vapour which diffused into the cool part of the vessel. 

 We have attempted to overcome this complication by passing down 

 into the bottle, when full, or nearly full, of sodium vapour, a platinum 

 tube closed at the top with a glass plate, and filled with nitrogen, and 

 observing the absorption through this tube. The nitrogen in the tube 

 prevents for a short time the entry of the sodium vapour into the 

 tube, and so by passing the tube to different depths, the thickness of 

 the layer of sodium through which the observations were made could 

 be varied. It was found in this way that a layer of sodium vapour, 

 about 4 centims. thick at the atmospheric pressure at the tem- 

 perature of our furnace, gave the D absorption sharp and very narrow, 

 but as the sodium diffused into the tube the absorption extended until 

 it produced a broad band with diffuse edges. Without a long process 

 of washing, it is impossible to clear the tube of sodium, so that the 

 repetition of observations in this way is a very tedious business. We 

 .are looking for some transparent substance which will stand the high 



VOL. XXIX. 2 L 



