on Spectrum Analysis. 101 



wide, and the glass 1*6 millim. thick. The tube rested in a 

 horizontal position on glass supports, depending from wooden 

 consoles attached to the wall of the room. The connexion 

 with the air-pump was made by means of a Kundt's glass 

 spiral. At the same time a spectral tube was put up in the 

 same way in the Physical Laboratory of the University, by 

 the kind permission of Prof. Lem«trom. 



In the tube in the Polytechnic Institute the following gases 

 were successively examined : — air, hydrogen, oxygen, air, 

 nitrogen, hydrogen, air, oxygen. The apparatus was unaltered 

 all the time ; the new gas was introduced, in the manner 

 peculiar to my pump*, after the former gas had been removed 

 as completely as possible. I give here only the results for 

 air, regarding the results obtained with the other gases merely 

 as preliminary, since I was not able to work with pure gases. 

 I therefore reserve the complete examination of these gases 

 for a future occasion, and give here the results already obtained 

 with them only briefly. 



I wish to express my thanks to Herr G. Melander for the 

 help he gave me during the whole investigation. 



The conducting-globes had a diameter of 27'8 millim. ; the 

 sparks were generally 5 millim. long. It appeared that a 

 great length of spark might be injurious to the tube. On 

 one occasion a tube of thin glass broke whilst a discharge 

 passed between the tinfoil coating and the layer of air 

 clinging to the inner wall of the tube. I have further 

 remarked that of the capillary canals, which always exist 

 in the walls of the tube parallel to its length, those which lie 

 nearest the inside easily burst and discharge the air which 

 they contain into the tube, when the discharges are powerful. 

 It therefore often happens that the pressure in the tube sud- 

 denly increases very much, and the luminosity becomes \evj 

 vivid, if, after having produced a great exhaustion, the air 

 has been driven from the inner wall of the tube by electric 

 discharges. In one case I was able, after such an occurrence, 

 to detect the small holes in the capillary channel which had 

 burst, and the little splinters of glass. 



I have made the following observations with regard to the 

 mercury-lines which are to be expected when the mercurial 

 pump is employed. In tubes containing pure air, nitrogen, 

 or oxygen, the mercury-lines appear only at high exhaustions. 

 Only the strongest line 546 appears in tubes containing air; 

 but in nitrogen and oxygen tubes also the lines 579 (double), 

 49*2, and 43b*. In hydrogen tubes, and in tubes with impure 

 air (containing carbon dioxide), these lines appear already at 

 * Zoc. cit. p. 178. 



