RELIABILITY OF HOLDING TIME MEASUREMENTS 

 is found to be 



E = a^- 



(i - ^^ 



381 



(12) 



ami the standard deviation is 



(t' = •%/« 





(13) 



O 25 



S 20 



0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1 I-2 



DISTANCE FROM NO. 60 SWITCH COUNT TO END OF HOUR IN MINUTES 



Fig. 10 — Switch counts lost on calls which originated after the last switch coutit 



If j is small, a will be correspondingly small making the average switch 

 counts lost, £, small. The distribution of lost switch counts will then be 

 very skew since the case of zero lost will be prominent. For larger ;'s the 

 distribution should assume a unimodal form, gradually becoming less skew. 

 A number of Newark busy hours were studied in this fashion, and the results 

 are compared with theory on Fig. 10. The agreement is seen to be fair 

 and a modal line estimated by eye falls substantially below the expected 

 mean corroborating the decided skewness just predicted. The wide fluctu- 

 ations in lost switch counts, each one of which if incorrectly estimated 

 results in a considerable error (in the present example about .2 second) 



