6 Professor Dixon and Mr. Harker on 



to the two free ends of the tubes were steel flanges with 

 ground faces. Their interior was Hned with glass. 



The gas passed directly from the electrolytic cell through 

 the drying tubes into the long explosion tube — driving out 

 the air before it : at the further end it passed through an 

 "analyser," which served to determine the composition of 

 the gas issuing from the tube. This consisted of a large 

 bulb holding 80 c.c, provided at the bottom with a three- 

 branch tap leading to a funnel containing dilute KI solution 

 and to a small bottle containing water. At the upper end 

 was a tail-tap leading to a pipette and to a lime and coke 

 tower. The pipette, which was constructed of narrow tubing^ 

 about 6 mm. diameter, was provided with a millimetre scale 

 and calibrated by mercury. The tube between the three- 

 way tap and the pipette was very fine, less than i mm. bore. 

 The dry gas was passed from the explosion tube through 

 water into the bulb, which had been previously dried ; then 

 the excess passed into the tower. After collecting the 

 sample, the taps were all closed, the gas-pipette being 

 completely full of water. 



The bulb was first exposed for some time to diffused 

 daylight and then to sunlight or to some magnesium flashes. 

 The tap below was then opened to the KI reservoir, and 

 the liquid allowed to rush up. A small residue generally 

 remained, which was measured in the pipette. This small 

 residue was found to be hydrogen ; it was never more than 

 I per cent of the whole, and in most of the experiments 

 varied between "2 and '4 per cent. 



To ensure complete combination between the hydrogen 

 and chlorine in the bulb by the action of light, it was neces- 

 sary to moisten the gases. The gases as used in the dry 

 experiments may be exposed for days to sunlight without 

 complete combination taking place. The potassium iodide 

 serves to show that there is no chlorine left. 



The filling of the explosion tube took generally two to 



