332 



PHENOMENA, ATOMS, AND MOLECULES 



right-hand ordinates are the values of 6. Each point shown corresponds 

 to a separate run in which Cs was allowed to accumulate for the time t 

 after cleaning the filament by flashing. 



The straight line in the figure has been drawn with the slope \ia as given 

 by the steady ion current. The points lie on this line within the experimental 

 error of < i percent for 6 below about 0.07. The deviation above 9 = 0.07 

 is due to evaporation of Cs partly as atoms during the flash to obtain the 

 ballistic kick. These curves remained unchanged if the coating temperature 

 was varied from 300° to 800°. At still higher temperatures, deviations set 

 in due to the reevaporation of arriving Cs atoms. These will be described in 

 Section XI. 



In the same way the 2-filament method was tested by measuring ac- 

 cumulated amounts of caesium corresponding to 6 from r—' 0.05 to nearly 

 i.o. In this case the ballistic kick Q should equal SAV^Jf^, where / is the 

 fraction of the atoms intercepted by filament B. As found from the distance 

 (2.85 mm) between filaments A and B, measured by a microscope with 

 eyepiece scale, / was 0.0071. On plotting the observed values of 04, cal- 

 culated from the ballistic kicks, (a^ = Q/SaCJ), it is seen, Fig. 5, that 

 these values have accurately the calculated slope, \ia, but are displaced 



Fig. 5. Calibration of two filament method. 



slightly with respect to the calculated line. Thus the kicks extrapolated to 

 zero time were not zero. This displacement was shown to be independent 

 of 6 and caesium pressure. A brief investigation did not reveal the cause of 

 the displacement ; it is probably connected with the pressure lags mentioned 

 in the next section. This small empirical correction (—0.0086) was applied 

 to all values of 6 obtained by the 2-filament method. 



In addition to these preliminary comparisons, later measurements of 



