88 Dr. Norman Campbell on 



for those potentials. The other columns give the values or 



v - for the various potentials, where L is the measured 



ne l ' p 



current through the condenser at the pressure concerned, i 

 that with the same potential at zero pressure. If (1) were 

 applicable they would thus represent the part of the current 

 due to the gas and would be equal to 



1-2^0+ ~r°f(V)dY. 



The last column (marked calc.) gives the values of this 

 expression on the assumption that the 8 rays from a layer of 

 gas are emitted with the same speeds as those from the 

 surface of a metal, so that ;'(V)=/(V). The integral is 

 calculated by quadrature from the numbers in the first column. 

 The probable error of the numbers in all the columns but 

 the iirst is about O03. It is so large, firstly because the 

 currents to be measured are small and the numbers recorded 

 are only differences between measured currents ; secondly, 

 because the value of N is somewhat uncertain. As several 

 observers have found, N depends in some measure on the gas 

 film on the surface of the electrodes and may alter when the 

 pressure is changed. 



5. Let us first consider the results for hydrogen. 



The measurements when V is 20 volts or more are clearly 

 affected by ionization by collision and hence have no bearing 

 on the present discussion. For the rest attention is directed 

 to the following features : first, the numbers increase regu- 

 larly with l/X ; second, the numbers in each column do not 

 increase regularly with V but have a minimum at about 

 3 volts ; third, the numbers for small values of l/X and V 

 are very much less than those calculated from (1) ; they are 

 often less than those corresponding to the second term in 

 that equation which represents the current carried by the 

 positive ions only ; fourth, in the first columns there are 

 negative numbers, that is to say, the presence of a little gas 

 in the condenser actually diminishes the current flowing 

 through it with small voltages *. 



It is clear that none of these features can be explained on 

 the basis of (1) ; but they may be explained, at any rate 

 partially, on the basis of the discussion of the effect of a 

 small amount of hydrogen. We saw that the main effect of 



* The decrease is very small and on the margin of experimental 

 error, but the experiment was repeated several times aud always gave a 

 decrease. 



