KEUAHIIATY OF HOLDING TIME }rEAS['KE.\fK.\TS 



377 



^'m = w;/ 



I - e , 



(5) 



1 - e"'7 



(6) 



In Fig. 7 is siiown a comparison between this theon' and the actual 

 numbers of 1-minute switch counts contributed bv these carry-over calls 



) 2 4 6 6 10 12 14 16 IS 20 22 2-: 



m= NUMBER OF CALLS CARRIED FORWARD INTO HOUR 



Fig. 7 — Gains Dver true switch counts at the start of the observation {)eriocl 



in 30 hourly periods of observation on the Newark interotitice trunk group 

 previously described. Since w, the average number carried forward is 

 about 15, it was thought that the sum-distribution of so many variables 

 would rather closely approach normality. On this assumption the ±50'^'(, 

 and ±99.73'^ (^ "control" lines have been drawn on Fig. 7. One point falls 

 on the latter limit lines, while 16 of the 30 fall within the SO'^'f lines, thus 

 providing a gratifying corroboration of the theory. 



