EVAPORATION OF ATOMS 329 



The correct potential for A is most easily found by flashing A when nearly 

 fully coated, with B cold. If A has been made too negative with respect 

 to Cq, a ballistic kick due to electrons is observed ; if not sufficiently negative 

 a kick due to ions is noted. With the correct adjustment (critical only 

 to about ±2 volts) there is no detectable galvanometer deflection when A 

 is flashed even if the initial values of 6 on A are varied from o to above i. 

 This fact that the potential of A can be varied as much as 4 volts without 

 detectable galvanometer deflections when B is cold proves that the zero 

 value of current is not due to a balance between electron and ion currents 

 from A, but that each of these currents is zero. This is undoubtedly caused 

 by space charge limitation of the currents during the excessively short 

 time that they can flow while the filament is being flashed. We therefore 

 believe that no source of error is introduced into these measurements by 

 currents from A. Although the sudden evaporation of atoms was usually 

 accomplished by heating the filament by a condenser discharge, increasing 

 the filament current to a value giving a temperature of about 1800° by 

 closing a switch in a battery circuit also gives sufficiently rapid evaporation. 



The accumulation method 



If Ha is small, changes in the properties of the film with time can be 

 measured as increases slowly from zero. In this method v„ must remain 

 equal to zero since 04 and 6 are found from the product \ia X t. This is 

 accomplished either by working in the lower ranges of $ or at low filament 

 temperatures. 



Comparison of the methods for determining 6 



(a) Becker's method may be used to determine ^'s from about 0.0 1 

 to the optimum value ^opt which gives the maximum in the electron 

 emission ; it will be shown later that ^opt = 0.67. The method requires not 

 only that ^,,pt shall be attained and that the electron emission at the optimum 

 be measurable, but also that reevaporation of atoms is negligible until the 

 maximum is passed. Thus with the sensitive galvanometer used in the 

 present experiments (7 X 10"" amp./mm) the allowable !.i„ was limited 

 to the range between approximately 6 X 10^^ and 2 X 10^^, using a testing 

 temperature (Ti) of ^ 550° K. At lower testing temperatures the electron 

 emission cannot be measured. A higher testing temperature raises the 

 lower limit for |i„. Even though ^opt may be passed through, unless the test- 

 ing temperature is low enough, atom evaporation may begin near ^„|,t 

 causing an increase in the time (^3) required to reach ^opt and a consequent 

 error in the determination of 6. At higher values of \i„ the coating time 

 becomes too short in comparison with the times required to bring the 

 filament from a high temperature to a steady state at the lower testing 



