EVAPORATION OF ATOMS 363 



The effect wave-length of the radiation producing these photo- 

 currents (a kind of Crova wave-length) can be determined by equating the 

 coefficient of i/T in Eq. (35) to 2.303 X C2A; it is found to be 5070A, 

 which is reasonable for caesium photoelectric cells. 



We may estimate the magnitude of 6 under the conditions of the ex- 

 periments of Table IV by the limiting form (as 6 approaches zero) of 

 Eq. {22), 



logio (/^/./^J- 26.625 -10,294/r. (36) 



Here Vp is the ion evaporation rate without accelerating field. Since in these 

 experiments Vp, which was equal to [la, was increased 7-fold by the field, we 

 may put Vp = 4.9 X 10^^ in Eq. (36). The values of 6,,, calculated in this 

 way, are given in the last column of Table IV. For such low values of 

 6, Ve cannot differ appreciably from that for a pure tungsten surface, so that 

 the assumptions made in deriving Eq. (34) are justified. 



The experiments thus indicate that within the experimental error of 

 about 0.2 per cent the saturation ion current is independent of temperature 

 and of field strength for ranges of temperature from 1200 to 1500° and 

 for fields from 3000 to 20,000 volts cm"^. 



We must conclude from both theory and experiment that the ion 

 saturation current method provides an extremely accurate measurement of 

 ctol^a ; where a,, is the value of Oa for very small values of 9. 



Methods for the experimental determination of ag, and Op 



If, now, we had some independent means of determining the vapor 

 pressure of caesium, from which we can calculate [la, we could determine 

 tto from our knowledge of ao\ia- However, as none of the available vapor 

 pressure methods appears to be comparable in accuracy or sensitivity with 

 that of the measurement of v^, we need to investigate other ways of find- 

 ing tta. 



Experimental data, such as that of Table IV, which were also obtained 

 for a wide range of other values of \Xa, prove that ag is strictly independent 

 of temperature in the range from about 1000 to 1500°, and is independent 

 of E, the accelerating field. This suggests strongly that Op is unity since 

 any smaller value would probably vary with temperature. There are, how- 

 ever, other methods open to us for measuring a„. 



(i) Direct flashing method. Caesium is allowed to accumulate on the 

 filament at a temperature Ti at which v^ is negligibly small, at the rate aa|i«. 

 After a time t, a a. is measured by the D.F. method (see Section III and 



Fig. 4). 04 = ctavl^a^ where Oav is an average value of ao over the range in 

 from o up to the final value at time t. 



