Discharge in Rarefied Gases. 373 



A piece of metallic sodium was introduced into a cylindrical 

 vessel 4 centim. in width and more than 20 centim. long, having 

 the electrodes at its ends; and the tube was then quickly filled 

 with dry nitrogen. After the tube had been so far exhausted 

 that its positive light filled the section of the tube, the sodium 

 was brought upon a part of the wall of the cylinder, in a hori- 

 zontal position, played upon by the positive light. The sodium 

 was next warmed until no more hydrogen was evolved, the tube 

 was refilled with fresh nitrogen, and exhausted again to the 

 same density as before. The sodium is then heated strongly 

 until it begins to volatilize, and the discharge, which was 

 reddish purple before, assumes a golden-yellow colour in its 

 neighbourhood. If the heating be carefully managed, it is seen 

 that the sodium vapour diffuses itself very slowly; so that the 

 discharge in the upper part of the tube still shows the red 

 colour due to the nitrogen, whilst it is of a golden yellow in 

 the lower part of the tube. If the tube be brought in a hori- 

 zontal and equatorial position near to a strong magnet, whose 

 poles are so placed that the positive light is drawn upwards, 

 the discharge, which at first filled the whole width of the tube, 

 is concentrated into a thread of greater or less tenuity against 

 the upper surface of the tube. But this thread possesses the 

 pure purple colour of the nitrogen discharge without any trace 

 of the sodium-yellow. The sodium vapour is consequently not 

 displaced by the magnet, together with the current, as we are 

 accustomed to see with movable carriers of electricity; the 

 current seems to obey the magnet with- 

 out affecting the gas-molecules. The 

 result is exactly the same in experi- 

 ments ! made with the Holtz machine 

 instead of the induction-coil. 



I have further examined whether it 

 is possible to recognize a transport of 

 gas- molecules by means of the magnet, 

 in the local increase of density which 

 must result from the assumed transport 

 in a closed space traversed by the dis- 

 charge. Two discharge-tubes, A and B, 

 fig. 2, were joined together in the man- 

 ner shown by means of a tube in which 

 a stopcock was inserted. A second stop- 

 cock at the end of a short tube shuts B 

 off from the pump during the time oc- 

 cupied by an experiment. The cylin- I 

 drical portions of A were sufficiently long to show stratified 

 positive light, at least in the cylinder containing the anode, 



