i. 



Experiments with Electron Currents in Different Gases. 427 



furnace. Fig. 1 exhibits the relative size and position of 

 the metal parts which are of electrical importance. The 

 mercury was supplied from the bottom of the tube L D, 

 which could be heated by a current through a metal spiral 

 outside it. D is a piece of tube 7 cm. long perpendicular to 

 the plane of the rest of the section, and is outside the furnace 

 already referred to whose boundaries are indicated by the 

 dotted Hues. In this way a little mercury could be distilled 

 from Dinto the bulb AB 0, and the pressure of the mercury 

 vapour could then be regulated by means of the surrounding 

 electric furnace. The currents across the gap AB — >Q> were 

 measured by a mic roam meter, and the photo-electric currents 

 to E by means of a quadrant electrometer. Beyond K the 

 quartz tube was ground to fit a glass cone, the joint being 

 made tight with a minimum quantity of involatile grease, 

 and precautions were taken to prevent it from becoming 

 warm under any circumstances. Beyond the joint were 

 (1) a side tube leading to a Geissler tube to test the spectrum 

 of the gas present, (2) a U-tube which could be immersed 

 in liquid air, (8) a mercury cut-off, (4) a bulb containing 

 coconut-charcoal in a liquid-air bath, (5) a tube with glass 

 tap leading to Gaede pump, McLeod gauge, &c. So far as 

 we have been able to ascertain, the experimental tube was 

 absolutely air-tight. Before beginning the observations the 

 quartz tube was heated in the furnace, and maintained at 

 a red heat for some time by means of a blowpipe; the 

 filament AB glowed with the charcoal in liquid air, and the 

 pump was running, so as to get all the occluded gases out of 

 the apparatus. This procedure was found satisfactory; but 

 it appeared that continued heating of the filament AB 

 liberated traces of hydrogen for some time after the expe- 

 riments were begun. After this evolution of gas had ceased 

 it was found that on heating the tube LD lines due to 

 hydrogen and carbon monoxide could be detected. The 

 experiments to be described refer to conditions in which 

 these contaminants were not present, at any rate in sufficient 

 amount for spectroscopic detection, unless the contrary is 

 specifically stated. For the construction of the quartz appa- 

 ratus we are indebted to the skill of Mr. Reynolds of the 

 Silica Syndicate, Ltd. 



The relation between current and potential difference across 

 the gap AB — >C depends on the pressure and nature of the 

 gas and on the temperature o£ the filament, which controls 

 the maximum possible electron emission. It may possibly 

 depend on the temperature of the gas as well. The effect of 

 gradually raising the applied potential difference is shown 



