THEORIES FOR TOLL TRAFFIC ENGINEERING IN THE U. S. A. 427 



0.10 0.2 0.5 1.0 2 



AVERAGE "submitted" LOAD IN ERLANIGS 



Fig. 3 — Distributions of simultaneous calls on three-trunk toll groups at 

 .\lbany and Buffalo. 



I nit ted" loads a' to produce the observed carried loads is required. On 

 Fig. 4 are shown the corresponding comparisons of theory and obser- 

 vations for the proportions of time all paths are busy ("NC Existing") 

 for 2-, 4-, 5-, 7-, and 9-circuit groups. Good agreement has also been ob- 

 served for circuit groups up to 20 trunks. This has been found to be a 

 stable relationship, in spite of the considerable variation in the actual 

 practices in ringdown operation on the resubmission of delayed calls. 

 Since the estimation of traffic loads and the subsequent administration 

 of ringdown toll trunks has been performed principally by means of 

 Group Busy Timers (which cumulate the duration of NC time), the 

 Erlang relationship just described has been of great importance. 



