238 Research Staff of the G. E. C, London, on the 



our assumption, is ij>/i n , or the number of ions formed by each 

 thermionic electron. At the lower pressures the points on 

 this second curve lie well on the curve, but at the higher 

 they are irregular*. 



In one respect the curve for a. agrees well with expectation- 

 At pressures so high that every ion formed collides with a 

 neutral molecule, we should expect a to be independent of 

 the pressure, assuming that the reaction takes place at every 

 collision. At lower pressures, on the other hand, some ions- 

 may get to the grid without collision, and a. should decrease* 

 And that is what is found. The limit at which a becomes 

 dependent on the pressure is such that the m.f.p. of a 

 molecule is about O'l of the distance between anode and 

 grid, so that something less than O'l of the ions should 

 escape collision. It is clear, then, that the reaction must take 

 place at a large proportion of the collisions ; and, if it takes 

 place nearly every time, there is surely no reason to doubt 

 that it takes place every time. 



But there are discrepancies. The most serious is that the 

 maximum of a appears to be, not 2 as was anticipated, but ] . 

 A less serious disagreement is that a falls off too rapidly at 

 low pressure and, if the curve were prolonged, would reach 

 zero at a finite value of p. '• 



9. A simple explanation of the first discrepancy may be 

 offered. The maximum value of a is not reached, because 

 the action is reversible, and its products are not removed 

 with sufficient rapidity from the vessel. It is well known 

 that hot tungsten reduces C0 2 to CO, being oxidized in the 

 process. The reduction results in the production of one 

 molecule of CO from one of C0 2 , whereas the main reaction 

 under consideration produces one molecule of C0 2 from two 

 of CO. Accordingly, if the discharge is reducing the pres- 

 sure of CO at the rate Q (proportional to i p ) 3 and if p lf p 2 - 

 are the partial pressures of CO and C0 2 in the gas, we have 



-^ = Q-/^ 2 (3) 



-jP«=iQ+flp,+l)pto • - • • • W 



where /3p 2 is the rate at which CO is being produced from 

 C0 2 by the hot tungsten, and Dp 2 the rate at which C0 2 is 

 diffusing to the cooled trap, both these rates being assumed 

 proportional to p 2 . If (/3 + D) is sufficiently great, the re- 

 action (4) will have established equilibrium before p x changes 



* The irregularity is almost certainly due to the accumulation of CO2. 

 in the gas (see par. 9) ; i f ji n is higher for this gas. 



