between Anode and Kathode Spectra. 339 



spectrum of carbon oxide, no doubt from some of that gas having 

 escaped from the glass. This having been pumped out as much 

 as possible, stopcock (6) was opened and well dried hydrogen 

 from the phosphoric anhydride tube allowed to enter the sparking 

 tube, and then the whole system, up to stopcock (a), again ex- 

 hausted. Stopcock (b) was then closed and (a) opened for a short 

 time and then closed again. The system was then left until 

 evening, when the same series of operations were repeated. It 

 was not until the fifth day that the GO' 2 ceased to appear in 

 the sparking tube, so that I could observe the behaviour of 

 the hydrogen alone. Even after such treatment the tubes 

 generally give out more carbon oxide if kept for many days. 

 Much sparking of the tube in a highly exhausted state must 

 be avoided, because the sparking is likely to decompose the glass. 

 If the yellow sodium light is seen at the point where the 

 electrode enters the tube it is a sure sign of decomposition of 

 the glass, and I could never get such tubes to give clean spectra. 



With a fairly clean tube, still attached to the pump and to 

 the other tubes, it is easy to trace the changes in the spectra 

 of hydrogen at different pressures. At about 4 mm. pressure 

 the negative glow is well seen about the kathode. It begins as 

 a close-fitting pink sheath on the kathode, which expands as the 

 pressure is reduced. The spectroscope shews that its light is 

 chiefly that of the rays G, F, and G'. It maps out extremely well 

 the course of the kathode rays. If the kathode be a cylindrical 

 wire of one or two millimetres thickness, terminated by two plane 

 facets inclined to the axis, the gas is lighted up equally all round 

 the cylindrical part of the kathode, while from near the centre 

 of each terminal facet proceeds a cone of luminosity exactly in 

 the course of the kathode rays, see Figs. I, II, and when the 

 density of the gas is so reduced that a sensible fraction of the 

 kathode rays escapes absorption by the gas, the base of the cone, 

 where it meets the glass, makes the tube fluoresce. If the 

 kathode be a six-sided prism instead of a cylinder, and be looked 

 at end on, six conical pencils of luminosity diverge from the plane 

 faces of the prism and are easily traceable right across the 

 Crookes dark space, which is only relatively dark ; see Fig. III. 

 It is well known that the luminosity is deflected by a magnet 

 along with the kathode rays. 



When the pressure of the gas is so far reduced that striations 

 appear, the spectrum of the negative glow consists of the lines 

 G, F, &c. only, without a trace of the second spectrum. In the 

 light of the negative glow photographs shew nine of the ultra- 

 violet lines of hydrogen. Turning the spectroscope on to the head 

 of the first stria, the second spectrum is there easily visible, and 

 so it is at the bright convex edge of succeeding stria?. In the 



