MOTION OF GASEOUS IONS IN STRONG ELECTRIC FIELDS 213 



averages (52), (54) and (57) for Ca , cl and cl . These formulas ought 

 to be verified in the Monte Carlo calculation. This is indeed approxi- 

 mately true. A sampling covering 9492 out of the 10,000 collisions gives 



by Monte Carlo by deduction 



(w^) 1.912 2 



(wl) 0.801 ^=0.889 



(wl) 5.165 ^=6.222 



The agreement is essentially there but deviations are noticeable. In 

 judging these we have to realize that fluctuations are quite large in this 

 problem. For instance if the calculation is broken down into ten runs of 

 approximately 1,000 events each one finds the following time averages 

 for the partial runs: 



{w,} {w%) (w|) 



predicted 2.000 0.889 6.222 



Among these runs there are some having averages higher than the pre- 

 dicted values, but the data clearly show that the Monte Carlo averages 

 are generally lower. In the search for reasons it was first felt that perhaps 

 the desired mean value for r is not actually reached, perhaps through 

 systematic errors introduced by the operator when rejecting certain 

 runs. This seems indeed to be the case. The mean free time obtained 

 from the 9492 runs mentioned above is 



r = 0.4269 



which is slightly low. Indeed it is observed that the runs with high r 

 were particularly troublesome in the calculation and were preferably 

 rejected by the operator. It seems doubtful however that this error 



