16 



MR. 0. W. RICHARDSON ON THE IONISATION 



the upper limit. In doing so, great difficulty was experienced owing to the purely 

 irregular effects mentioned on p. 6. These effects were characterised by sudden 

 changes in the rate of leak of a purely temporary kind, and were quite different from 

 the hysteretic effects previously described. Very often in the midst of a reading, 

 when the electrometer spot was moving quietly across the scale at the usual rate, it 

 would suddenly give a kick and dart right off without any warning or apparent 

 cause. These kicks were much more marked at high than at low pressures. At 

 pressures of about 1 millim. it was rare for a kick to increase the leak in 15 seconds 

 by 25 per cent., whereas values of the leak equal to six times the minimum have been 

 recorded at atmospheric pressure. The following numbers, obtained at a pressure of 

 366 millims. and a temperature of 809 C., furnish an excellent example of this kind 

 of thing. 



The above represents the greatest value of the kick recorded. Generally speaking, 

 the value of the leak only oscillated up to twice its minimum value. Further 

 experiments on the source of these irregularities will be described later ( 10). As no 

 change in the arrangements has been found to eliminate them, they were avoided by 

 always taking the minimum value of the leak, and readings were always taken until 

 two consecutive minima agreed with one another. For instance, the true value of 

 the current from the above observations was taken to be 32. This method was found 

 to give consistent results. 



The curves shown in fig. 6 represent the result of an experiment with the 

 resistance of the wire maintained equal to that which it would have at a pressure of 

 1 millim. and a temperature of 816 C., the potential on the filament being +40 volts. 

 The unit of current is 6 x 10~ 13 ampere. All the observations except two were taken 

 with the pressure decreasing. The lower curve represents the same observations as 

 the upper, but the pressure scale is magnified ten-fold. The two observations marked 

 thus (*) are on the scale of the upper curve. They were made with the pressure 

 increasing after the other series were finished, and were taken very rapidly, so that 

 equilibrium was not fully established. They both fall well below the rest, owing to 

 tin- time-lag effect described alx>ve. The other points fall very nearly on the curves 

 drawn. 



