12 



MR. O. W. RICHARDSON ON THK IONJSATIOX 



In a previous paper, which has already been quoted,* it was shown that the leak 

 from a hot platinum wire in air at low pressures fell off as the pressure was reduced 

 from about O'l millim. to in such a way as to indicate that the leak consisted of two 

 parts, one proportional to the pressure, and one independent of it. A similar 

 experiment was now made with pure oxygen, except that observations were started 

 at a higher pressure. The results of this experiment are given in the next table. 

 The temperature was 816 C., and the potential on the filament +80 volts. 



The pressure was varied by making one stroke of the pump and then making an 

 observation, after waiting for everything to become steady, at the reduced pressure 

 thus obtained. The order in which the observations were taken was thus that of 

 diminishing pressure, so that any secular change which might be going on would go 



entirely in the one direction and might 

 vitiate the results. In order to test whether 

 an effect of this sort was coming in, fresh air 

 was let into the apparatus the next day and 

 observations again taken over a similar range. 

 The results are exhibited along with the 

 previous ones in fig. 3. The points marked 

 thus x refer to the observations in Table IX., 

 whilst those marked thus O refer to those 

 taken on the following day. The experi- 

 mental conditions were the same in both 

 cases. 



The lower curve may be compared with 

 the numbers for air up to O'll millim. in the 

 paper referred to above.* It will be seen that 

 at these low pressures the part of the leak 



which depends on the gas is roughly proportional to the pressure, agreeing with 

 what was previously found for air. The part of the leak which is independent of the 

 pressure, which is clearly marked in fig. 3, is smaller relatively to the rest than in the 

 previous experiments. It will not be noticed in the later curves, as it became too 

 small to be noticeable after a few days' more heating. 



Although the results given in fig. 3 show that hysteretic effects had been largely 



* 'Camb. Phil, Proc.,' vol. 13, p, 58, 1905. 



