r 



aud ^[l),>!0)'ption Spectra of SodiKni Vapour. 371 



below 4()00, which as we have shown is all that is transmitted 

 bv the vaponr when it is very dense. 



The Ahiiovptio)i Spectrum of Sodiuui Vapour. 



The tinted absorption-spectrum of sodium vapour was first 

 observed and studied by Roscoe and Schuster in 1874. 

 Subsequent investigations were made by Liveing and Dewar 

 in connexion with their work on the reversal of the lines of 

 metallic vapours, and also by one of the writers. 



Previous experiments by one of us having shown that it 

 was impossible to secure photographs of the fluted spectrum 

 with the concave grating, using the arc as a source of light, 

 that were not contaminated by bright lines from the vapour 

 of the lamp, it was necessary to find a source of light of 

 great intensity and having- a continuous spectrum. After 

 experimenting with various sources of light we finallv 

 adopted the Nernst lamp, which was found to fulfill the con- 

 ditions specified. 



In the first series of experiments the sodium was vapourized 

 in an atmosphere of hydrogen, generated by electrolysis, and 

 dried by passage over calcium chloride and phosphorus pent- 

 oxide. In order to remove traces of oxygen w^hich caused 

 the tube to smoke, the gas was finally passed over red-hot 

 copper gauze. The metal was heated in tubes of thin steel, 

 the ends of which were closed with plate-glass. In the 

 later experiments the tubes were exhausted with a mercury- 

 pump, and the metal volatilized in a vacuum. The latter 

 method w^as found to be most satisfactory. The spectra were 

 found to be identical in the two cases. 



The tubes w^ere either heated by means of Bunsen burners, 

 or by a coil of I^o. 20 iron wire, insulated from the tube by 

 a thin layer of asbestos, board. The vapour is more uniform 

 when the tube is heated electrically, for if the upper side of 

 the tube is colder than the lower, a non-homogeneous medium 

 results, the density being greatest along the floor of the tube. 

 A Bunsen burner is better, however, for some experiments, 

 where a very dense vapour is required. With it the metal 

 does not distil so rapidly to the colder parts of the tube, and 

 with careful regulation of the flow of gas to the burner, it 

 was used very successfully in many of the experiments. 



The absorption-spectrum w^as photographed with a 14-foot 

 concave grating in the first order, the time of exposure vary- 

 ing from twenty minutes to an hour, according to the density 

 of the vapour. The second-order spectrum was observed from 

 time to time during the exposure in order to keep the density 

 of the vapour properly regulated. 



It "was found that the best results were obtained with a 



2 B 2 



