508 



Mr. H. A. Wilson on the Variation of the 



series of observations, partly owing to changes in the gas 

 pressure which occur sometimes when the discharge is passed 

 for any length of time, and partly owing to the difficulty of 

 obtaining a sensibly steady discharge except within certain 

 limits of current value. 



Another source of trouble is the spluttering of the negative 

 electrode which occurs even with aluminium electrodes, and 

 forms a conducting coating on the walls of the discharge-tube 

 which may cause the character of the discharge to change as 

 the electrodes are moved into a different portion of the tube. 

 The only remedy for this is to open the tube and remove the 

 deposit as soon as it begins to have an appreciable effect. The 

 backs of the electrodes were covered with disks of mica to 

 prevent spluttering there. 



The gases tried were air, nitrogen, and hydrogen. All 

 were dried with P 2 5 , and the nitrogen was prepared from 

 NaN0 2 and NH 4 C1, and passed through FeS0 4 solution and 

 H 2 S0 4 . The hydrogen was prepared by electrolysis of dilute 

 H 2 S0 4 and passed through KMn0 4 solution. Oxygen which 

 was used in some later experiments was prepared by heating 

 pure KMn0 4 . The light from the discharge was in each case 

 examined with a pocket-spectroscope, and when no impurity 

 could be detected in this way the gas in the tube was con- 

 sidered sufficiently pure. 



Fig. 3. — Discharge in Nitrogen. 

 Pressure TOG mm. Current 0676 milliampere. 





— 









i 











~T 











1 















1 



8o 

 70 

 60 



















































1 

















,■ — - 









1 



I 



u 50 



<o 40 

 ^ 30 



^o 



10 



^ 

















\ 



L_ 



























\ 





1 



i 

 j 















T 























\ 



V 



! 



' ! 1 

















\ 



/ 







1 I 











--. 



A, 



s 



'GAT/i 



=^V, 



+ 





POS/r/KS CO/.UMA 



r 







eGion 



V .-'"'. 



10 II 12 



Some of the results obtained are shown in figs. 3, 5, 6, 7, 8. 

 The distribution of light along the discharge is roughly indi- 

 cated by the dotted curves. 



