EVAPORATION OF ATOMS 327 



ing no mobility. Since the experiments (see Section XI) have shown no 

 difference in properties of a fihii with a given average value of 6 whether 

 it was obtained by irreversible condensation or evaporation, we must con- 

 clude that surface mobility equalized over the surface. Because of this 

 mobility and the fact that a = i, we may therefore conclude that the values 

 of V defined in terms of apparent surface are identical with the evaporation 

 rates per unit area of true surface. 



III. METHODS FOR DETERMINING 6 



There are four methods available for determining 6. Common to all 

 methods is the measurement of the caesium vapor pressure (controlled by 

 bulb temperature) in terms of n„, by observing the positive ion current to 

 a collector at a negative potential with respect to a hot (>I200°K.) 

 tungsten filament. As will be demonstrated later, every atom striking the 

 filament leaves as a positive ion so that Ip/e, the saturation positive ion 

 current density divided by the electronic charge, gives \ia. 



Becker's method 



Becker observed the time (^„,) required for the electron emission from 

 a filament (initially cleaned by flashing) to reach a maximum value while 

 the filament at temperature Ti was being coated by a constant supply of 

 Cs vapor. At the maximum, 6 was assumed to be unity, that is, the film was 

 taken to be a complete monatomic layer. It was then possible to raise the 

 temperature of the filament to various temperatures (T2), where the 

 equilibrium 6 was less than unity. If the temperature was now lowered to 

 Ti, the electron emission again passed through a maximum as B increased 

 through unity. The time (fs) needed for this completion of the film served 

 to measure the 9 at To since 6 = \ — tz/hn- 



Direct flashing method 



If a tungsten filament having a film for which 6 is less than 0.08 is 

 suddenly flashed in an accelerating field at a high temperature, all the 

 caesium evaporates as ions giving a ballistic kick on a galvanometer. The 

 filament area together with the ballistic calibration of the galvanometer 

 allow calculation of Oa and hence of 6. 



Two filament method 



As shown diagrammatically in Fig. 2, two parallel straight tungsten 

 filaments, A and B, were mounted near the axis of three coaxial cylindrical 

 electrodes Ci, Cq and C2, placed end to end so that by the guard-ring 

 principle the electron or ion emissions from a known length of the central 



