1886.] Passage of an Electric Discharge through Nitrogen. 335 



formate of potassium ; in this case the nitrogen showed no trace of 

 sulphur dioxide, there was, however, no alteration in its behaviour in 

 the discharge- tube. 



We detected the trace of S0 2 by the change produced by the gas in 

 the colour of paper soaked in a mixture of ferric chloride and ferric 

 cyanide, the amount of it, however, must have been very small, as the 

 gas produced no colouration in a paper soaked in a solution of iodide 

 of potassium and starch, which is a very delicate test for sulphur 

 dioxide. The diminution in the volume of the nitrogen which we 

 observed could not have been due to the trace of S0 2 , as it occurred 

 when the copper oxide had been reduced by CO, and no trace of S0 2 

 was to be detected even by the ferric chloride and ferricyanide 

 solution. 



Determination of the Quantity of Oxygen in the Gas. 



We were unable to detect any change of colour in a small quantity 

 of a solution of pyrogallol and caustic potash when 50 c.c. of our gas 

 was passed through into a eudiometer. It was thought desirable, how- 

 ever, to have a more perfect testing arrangement, and the following is 

 a description of the form ultimately adopted. 



Fig. 4. 



The apparatus consists essentially of four tubes and a bottle whose 

 volume is known ; by means of connexions of indiarubber tubing, 

 taps, and clamps, each tube can be put in communication with the 

 nitrogen supply and with the bottle separately. The tap between C 

 and D being closed, and the tubing temporarily removed, C is filled to 

 about half way up the bulb with carefully boiled solution of caustic 



VOL. XL. 2 A 



