362 Dr. J. W. Capstick. 



it could always in the case cf the compound gases be made to sing by 

 a sufficient increase in the current, but a reduction of the current 

 density by the use of electrodes half an inch in diameter made it no 

 easier to obtain a constant current. 



The method by which it was hoped that errors due to the decom- 

 position of the gas would be got rid of was to start the current, and 

 when by accident as it proved rather than by design, the telephone 

 was silent, to open the mercury trap and allow a stream of pure gas 

 to play on the electrodes, whilst the reading of the electrometer was 

 taken. It was not often that this could be carried out, for when 

 after repeated attempts the telephone was silenced, the usual result 

 of admitting fresh gas was either to make the current intermittent 

 or to stop it altogether. 



The proceeding was carried out successfully only twice. In the 

 first experiment the current was steady as soon as the circuit was 

 closed. The negative glow covered half the kathode and the 

 kathode fall was 471 volts, but quickly rose to 484 as decomposition 

 proceeded. A little vapour was then admitted, and the fall of 

 potential at once sank to 467 but soon rose again to 482 when the 

 stream of gas ceased. More vapour was then admitted whereupon 

 the kathode fall sank to 469, and in half a minute rose again 

 to 478. 



The second experiment was less satisfactory as only a single reading 

 was obtained. After the current had been running for a few minutes 

 the telephone became silent, the negative glow consisting of a bright 

 tip covering about a tenth of the kathode and the kathode fall being 

 484. On admitting more vapour it sank to 469, but before it had 

 risen more than a few volts the telephone begun singing again, and 

 could not be made to stop. 



Thus we have four observations of the kathode fall in the 

 undecomposed gas, namely : 471, 467, 469, and 469, the mean being 469. 



In both experiments the pressure of the vapour was about 2 mm. 



Ammonia. The gas was prepared by the action of soda on 

 ammonium sulphate that had been treated with nitric acid in the 

 ordinary way to remove organic substances. The ammonia was 

 dried with lime and absorbed in calcium chloride contained in a bulb 

 sealed to the apparatus. 



Here, as with water vapour, the numerous attempts to secure a 

 constant current met with little success. Only two readings of the 

 kathode fall were obtained, and these were not very concordant. In 

 the first experiment the current had been running for half an hour 

 when the interim ttence ceased, and the kathode fall in the partially 

 decomposed gas was 510 volts. When a stream of pure ammonia was 

 allowed to play on the kathode, the reading of the electrometer rose 

 to 595, where it remained steady until, after a few seconds, the dis- 



I 



