I'ltoltUCKD BY HOT IM.ATINTM IN DIFFERENT GASES. 



51 



the vessel, sin<v tins j>ntracted increasing of the pressure in a vacuum was uot 

 observed after a wire had been heated in other gases. Even if the above large 

 amount of gas has to be got rid of the rate of escape seems very slow ; in fact, the 

 numbers show thai the law for the rate of diffusion of hydrogen through hot platinum 

 obtained by RICHARDSON, NICOL and PARNEIL* at pressures greater than 1 millim. 

 does not hold at low pressures. It is probable that at these pressures it is necessary 

 to take external dissociation into account (vide loc. cit.). 



Whilst the preceding measurements of the rate of evolution of hydrogen were 

 being recorded, readings of the current with 13 volts on the filament were taken 

 simultaneously. This value of the jx>tential was used in order to ensure saturation 

 (see fig. 1 o) and at the same time to avoid the occurrence of ionisation by collisions 

 as far as possible. The readings were commenced at a pressure of 0'3 millim., and 

 the current was found to decrease by about 40 per cent, of its value on reducing the 

 pressure to 0'002 millim. This additional part of the current, which is nearly 

 proportional to the pressure, may be accounted for by sujjposing it to l>e due to 

 ionisation by collisions. The wire was then heated for 16 hours at a low pressure, 

 during which time gas was given off at the rates indicated by the numbers in the last 

 table. The values of the leak (1 = 10~* ampere) and the times, reckoned from the 

 instant at which the apparatus was first pumped down to 0'002 millim. pressure, at 

 which they were recorded are given in the next table : 



At a first glance these numbers indicate a continual falling off in the value of the 

 leak at constant temperature as the gas escaj>es from the wire. The criterion for 

 constant temperature was the resistance of the wire, and it was found that, owing to 

 the spluttering of metal which takes place at high temperatures, the resistance 

 of the wire at C. had increased considerably during the course of the above 

 experiment. When this was allowed for it was found that the average temperature 

 <-t' the wire at the close of the above experiment was 1280C. instead of 1370C M 

 its value when the experiment started. The leak at 1280 C. should have been 

 alwmt three divisions instead of twelve, so that pumping out the gas had 

 apparently increased the leak. This paradoxical result is probably caused by the fact 

 that the above method of reckoning over-corrects for the effect of loss by spluttering. 

 A calculation from the nuintars in the last table but one shows that the wire had lost 



* ' Phil. Mag.,' vol. 8, p. 1. 



11 _' 



