EVAPORATION OF ATOMS 333 



equilibrium values of 6 by both the direct flashing and 2-filanient methods 

 in the range of B from 0.05 to 0.08 were in good agreement. 



Becker's method was tried out extensively and the limitations already 

 given as to testing temperature and Cs pressure found. 



IV. DESCRIPTION OF THE EXPERIMENTAL TUBE 



To obtain accurate results by the foregoing methods many precautions 

 are necessary in the construction of the apparatus and in its use. 



The caesium pressure must be exactly controlled. The ordinary con- 

 struction with metal cylindrical electrodes inside a glass bulb does not allow 

 this. These cylinders become heavily coated in the presence of caesium 

 vapor and since their temperature is indefinite and afifected by radiation 

 from the filaments, the rate of supply of Cs is non-uniform. The use of 

 metal electrodes deposited directly on the walls of the tube avoids this, by 

 giving more direct thermal contact with the bath liquid. 



The guard ring principle must be employed to allow measurement of 

 currents from a known length of filament and to prevent measurement 

 of currents from the ends or any portion of the filament which is not at 

 the temperature of the central part of the filament. This is far more im- 

 portant in studies of caesium on tungsten than for clean tungsten. For 

 example, if the central part of the filament is hot enough to be practically 

 free from caesium, the end portions may be cooled sufficiently by the leads 

 to have high values of 6 from which the electron emission is thousands of 

 times greater than from the cleaner central part. 



The filaments should be so long that the cooling effect of the leads does 

 not cause a non-uniform temperature over the central part even at low 

 filament temperature. 



(i) Preparation of the tube 



Fig. 6a shows the glass mantle. The two annular folds {E) were made 

 by pressing together while hot two enlarged sections of the tube. These 

 folds divide the tube into three sections, Ci, Co and Co, spaced ■^^ mm apart, 

 which later are to serve as electrodes. The connections to each electrode 

 consist of tungsten seals from which 5 mil platinum wires are led to and 

 are partially imbedded in the glass wall. 



In Fig. 6a, P is the 10 mil platinum filament in place ready for the 

 coating of the walls by evaporation. The tube at this stage was evacuated 

 and baked at 400° C. for 2 hours, after which the oven temperature was 

 lowered to about 300° C. The filament was now heated to a temperature 

 where the evaporation of platinum in a period of about two hours was 

 sufficient to coat the glass walls with a film whose resistance as measured 



