484 BET J. SYSTEM TECHNICAL JOURNAL 



question now suggests itself: Suppose that a great number Q of elec- 

 trons, all having kinetic energy V, falls upon a thin stratum of gas con- 

 taining (IN atoms per unit area: how many atoms will they ionize, how 

 many ions will he produced? Designating this number by Qf{V)dN: 

 what is /(F)? 



This question is much easier to formulate in words than to answer 

 by experiment. Suppose for instance that one should try to answer 

 it by means of the scheme of apparatus sketched in Fig. 1. In going 

 from d to G2, coming to a stop between G2 and G3, and returning again, 

 the electrons pass successively through all values of kinetic energy 

 from their highest down to zero and back to their highest again ; and 

 ions are produced by electrons of all values of kinetic energy, from their 

 highest down to the ionizing-energy. The ions collected by the col- 

 lector at C represent a sort of integral of Qf{ V)dN taken between Vo as 

 minimum and the energy possessed by the electrons at d as maximum. 

 To determine Qf{V)dN it is necessary to measure the total ionization 

 at several values of V and then construct a sort of differential curve. 

 To determine Q it is necessary to know how many electrons come from 

 the filament into the ionizing-region, and in addition how many extra 

 ones are introduced through primary electrons knocking them out of 

 the gauze of Gi. 



Another scheme consists essentially in making G2 into a solid wall 

 and using it to collect the electrons, so that after passing from Gi to Go 

 they vanish from the scene. This would be excellent if the region be- 

 tween Gi and G2 could be left equipotential; but it is necessary to in- 

 trude a negatively-charged electrode in order to collect the positive ions, 

 and apparently whenever this electrode is sufficiently large and suffi- 

 ciently negative to capture the ions it is also sufficiently large and 

 sufficiently negative to distort the field between Gi and G2 quite seri- 

 ously; so that the electrons are at first slowed down and later speeded 

 up again as they pass from d to Go, and the ions received by the col- 

 lector are as before a sort of integral of Qf{V)dN. 



In spite of these difficulties the various experiments performed with 

 extremely rarefied gases yield fairly concordant results. ^^ The function 

 /(F) mounts steadily, from zero at the ionizing-energy Vo, to a broad 

 and flattish peak culminating somewhere between 100 and 400 volts 

 (depending on the gas), and thereafter declines slowly as F increases. 

 Thus, although an electron striking an atom (or molecule) can detach 

 the loosest electron if it has just the requisite energy, its chance of 

 doing so is improved if its energy is greater than the just-sufficient 



'« K. T. Compton and C. C. van Voorhis, Phys. Rev. (2) 26, pp. 436-453 (1925) 

 and literature there cited; also W. P. Jesse, ibid. pp. 208-220. 



