Mode of Conduction in Gases. 817 



favour, or, shall we say, holds quite dogmatically, the opinion 

 that electric convection in a partial vacuum is effected entirely 

 by negatively charged carriers/'' 



This statement is incorrect, as I never held such an opinion, 

 and no such explanation of electric convection is to be found 

 in any of the papers I have written on that subject. 



How inconsistent the above statement is with the theory 

 of sparking which I have published, may be seen if I mention 

 the simplest case of the application of the theory of ionization 

 by collisions to the explanation of the sparking potential for 

 parallel plates in a gas, at pressures above the critical pressure. 

 All the leading features of the phenomena can be explained 

 on the supposition that equal numbers of positive and negative 

 ions are generated by the collisions of the ions with the 

 molecules of the gas. A preliminary account of this theory 

 was given in the ' Electrician,' April 3rd, 1903, and it was 

 more fully described in a paper published in the Philo- 

 sophical Magazine, November 1903, and in subsequent com- 

 munications. 



The underlying principle, which is not of dogmatic origin 

 but is founded on experimental evidence, is as follows. 

 When negative ions are set free from a metal plate by the 

 action of ultra-violet light, and travel through a distance d 

 to a parallel plate under a constant force X in a gas at 

 constant pressure p, the number of ions n that arrive at the 

 positive electrode is given by the formula n~n Q e ad , where 

 ?*o is the number of ions set free from the negative electrode, 

 the remainder n (€ ad — 1) being the number generated in the 

 gas. The quantity a represents the number of positive or 

 negative ions generated per centimetre by the collisions of a 

 negative ion with molecules of the gas, and can easily be 

 determined for any given force and pressure. The formula 

 for n holds for short distances ; but when d becomes large 

 the action of the positive ions begins to have an appreciable 

 effect on the currents. If j3 is the number of positive or 

 negative ions produced per centimetre by a positive ion in 

 traversing a centimetre of the gas, then the total number 

 of negative ions arriving at the positive electrode is 



n = n n - ^~, — 5x3-, the two formulae giving the same values 



u a — j36 {a ~P' d 



of n for the shorter distances as {3 is small compared with a. 

 The effects of the positive ions may thus be obtained from 

 the currents obtained with the larger values of d when the 

 conductivity is maintained with potentials less than the 

 sparking potential, by the multiplication of the ions set free 

 by the light from the negative electrode. 



