l'l;ol)UCKD 1!V HOT I'LATINUM IN DIFFERENT OASES 



19 



200 



400 

 Pressure : 



Fig. 7. 



600 



800 



Treated in this way the numbers plotted in fig. 6 yield those exhibited in fig. 7. 

 As In-fore, the temperature is 816 C., the voltage +40, and the unit of current 

 6 x 10~ 13 ampere, the pressure being ex- 

 pressed in millimetres. 



It will be seen that despite the tempera- 

 ture correction which has been made the 

 leak varies very little with the pressure at 

 high pressures and is probably asymptotic 

 to a line somewhere about y = 56. 



The preceding observations show that 

 the positive ionisation produced by a hot 

 platinum wire in oxygen at temperatures 

 below 900 C. is approximately proportional 

 to the square root of the pressure at low 

 pressures. As the pressure is raised, the 

 rate of increase of leak with pressure gradually diminishes so that the leak tends to 

 approach a steady value asymptotically at high pressures. A similar result was 

 found to hold at higher temperatures, with the difference that the rate of increase 

 was greater at low pressures. At 1180 C. the leak was nearly proportional to the 

 pressure below 2 millims., and the rate of increase at higher pressures fell off as 

 before. The numbers supporting this conclusion will be found below in Part IV. 



Before leaving this part of the subject it is necessary to consider another source of 

 error to which the experiments were liable. This was due to the walls of the tulie 

 getting heated. During the course of the experiments the author tried the effect of 

 heating the tube in which the measurements were being made by means of a Buusen 

 burner placed outside it. This was found to produce an enormous increase in the 

 leak. In one instance, where the pressure before heating the tube was 0'0005 millim., 

 heating for a few minutes with a Bunsen burner increased the leak from 2'2 x 10~' 

 amj>ere to 5 x 10~* ampere, i.e., in a ratio of 1 to 20,000. At the same time the gas 

 evolved from the walls only sent up the pressure to O'OOl millim. This curious and 

 interesting effect, which is being further investigated, does not appear to depend on 

 the stute of cleanliness of the tube, as it showed itself with apparently undiniinished 

 vigour after the tube had been taken down and boiled out three times with pure 

 nitric acid and sul>8equently rinsed out seven times with boiling distilled water. The 

 effect was also obtained in air at atmospheric pressure and in a vacuum produced by 

 ]ii[iiid air and charcoal, where it was impossible for the wire to come in contact with 

 mercury vapour or vapours given out by phosphorus pentoxide. 



Whatever the cause of this effect may turn out to be, it is clear that it might have 

 completely vitiated the present measurements of the leak in oxygen and other gases. 

 In fact, tlu> following oonstderationa will show that a small trace of the above effect 

 might easily simulate the effects investigated. When the gas pressure is very low, 



D 2 



