3 66 PHENOMENA, ATOMS, AND MOLECULES 



to 0.98. We shall see that the deviations of the observed points from these 

 straight lines as 6 approaches a limiting value, are due to evaporation and do 

 not indicate values of Ua less than unity. 



In the next section we shall discuss the theoretical significance of the 

 experimental fact that Ug = i up to nearly — i. 



Value of Op 



Moon ■'•^ has shown that when a beam of Cs ions (without Cs atoms) 

 is directed against a tungsten filament heated to high temperatures, no 

 net current flows to the filament if there is a field near the filament which 

 draws away ions. Without this field, or at a lower filament temperature, a 

 current is observed which presumably measures the number of ions which 

 strike the filament. Moon concludes that all of the ions which condense on 

 the very hot filament leave it again as ions (none as atoms) . This is a proof 

 that Va/Vp is very small, but does not necessarily prove that a^ = i, although 

 it makes this probable. In view of the strong attractive forces between the 

 ions and the tungsten surface (image force) it is, however, almost certain 

 that there cannot be any appreciable reflection of low velocity ions, and we 

 may safely conclude that a^ = i. 



Transient effects in atom evaporation 



We use the term transient effects to describe the phenomena involving 

 changes in as distinguished from steady states in which 6 stays constant. 

 The accumulation periods which we have discussed correspond to transient 

 states in which v^ and Vp are negligible compared to [Xq. Let us now consider 

 the theory of the changes in 6 which occur when v^ and \ia are comparable 

 in magnitude, and when a retarding field for ions makes Vp = \ip. 



The data of Figs. 23 and 24 show that if a clean filament is held at 

 constant temperature in Cs vapor of a definite pressure, 6 increases at first 

 at the steady rate \ia/(^Ai but thereafter the rate decreases until finally 6 

 approaches a steady limiting value which we shall call 60^. Eq. (29), which 

 applies to this case, may be written in the following form, since 0^=1 



OAide/dt = ^ - V. (38) 



In this discussion we shall omit the subscripts of [Xq and v^ except where 

 necessary to prevent confusion. 



Values of Boo 



When 6 = 6cc we should have v = j.i. Thus, since by Eqs. (6) to (10) 

 and the data of Table I, v is given as a function of 6n and T, we can 

 calculate On from \i and T. For each of the temperatures used in the ex- 



^^P. B. Moon, Proc. Catnb. Phil. Soc, 2;, 570 (1931). 



