456 



BELL SYSTEM TECHNICAL JOURNAL 



expected when one considers that in an experimental filament the 

 patches are not all of the same size and neighboring patches do not 

 have the same differences in work function. One example is illustrated 

 in Fig. 19 which shows (kT/e) log i/iao vs. yj F ior a thoriated tungsten 



0.12 



0.10 



0.06 



0.04 



0.02 



-0.02 



20 40 60 80 100 120 140 160 180 200 



Y?" (VVOLTS PER CM ) 



Fig. 19 — Comparison of experimental and calculated curves. 



filament, 70 per cent of whose surface was covered with thorium. 

 For this curve iao is merely an arbitrary constant, chosen so as to give 

 the best fit with the computed curves. By comparing the experi- 

 mental curve with calculated curves in Figs. 16 and 17, we decided to 

 try h = 3.0 X 10~^ cm. and /x = 0.50 volt.* The average curve for 

 these values is shown in the figure. It appeared that h was too small 

 and M was too large and a new curve was computed and plotted 

 assuming 6 = 4.5 X 10~* cm. and y. = 0.47 volt. The agreement is 

 better. Probably a slightly better fit could be obtained with 

 b = 4.0 X 10~^ cm. and m = 0.48 volt. For other thoriated tungsten 

 filaments we have found values of b from 1 X 10~^ to 1 X lO^'* and n 

 values from 0.25 to 0.48 volt. 



* In choosing these values some allowance had to be made for the fact that the 

 experimental curve was for 7" = 1380° K., while the calculated curves were for T 

 = 1160° K. 



