THEORIES FOR TOLL TRAFFIC ENGINEERING IN THE U. S. A. 437 



would be expected to exceed slightly the values for completely random 

 re-trials. These particular curves are based on a re-trial interval of 1.67 

 times the average circuit-use time. Such moderation on the part of the 

 customer is probably attainable through instructional literature and 

 other means if the customer believes the "NC" or "busy" to be caused 

 by the called party's actually using his telephone (the usual case in local 

 practice). It would be considerably more difficult, however, to dissuade 

 the customer from re-trying at a more rapid rate if the circuit NC's 

 should generally approach or exceed actual called-party busies, a con- 

 dition of which he would sooner or later become aware. His attempts 

 might then be more nearly described by the (4) curves on Fig. 7 cor- 

 responding to an average exponential return of only 0.2 holding time — or 

 e\en higher. Such a result would not only displease the user, but also 

 result in the requirement of increased switching control equipment to 

 handle many more wasted attempts. 



If subscribers are to be given satisfactory direct dialing access to the 

 iiitertoll trunk network, it appears then that the probability of finding 

 XC even in the busy hours must be kept to a low figure. The following 

 engineering objective has tentatively been selected: The calls offered to 

 the ^'final" group of trunks in an alternate route system should receive no 

 more than 3 per cent NC(P.03) during the network busy season busy hour. 

 (If there are no alternate routes, the direct group is the "final" route.) 



Since in the nationwide plan there will be a final route between each 

 of some 2,600 toll centers and its next higher center, and the majority 

 of calls offered to high usage trunks will be carried without trying 

 their final route (or routes), the over-all point-to-point service, while 

 not easy to estimate, will apparently be quite satisfactory for cus- 

 tomer dialing. 



5. ECONOMICS OF TOLL ALTERNATE ROUTING 



In a general study of the economics of a nationwide toll switching plan, 

 made some years ago by engineers of the American Telephone and Tele- 

 graph Company, it was concluded that a toll line plant sufficient to give 

 ihe then average level of service (about T-40) with ordinary single-route 

 procedures could, if operated on a multi-alternate route basis, give the 

 desired P.03 service on final routes with little, if any, increase in toll line 

 investment.* On the other hand to attain a similar P.03 grade of service 

 by liberalizing a typical intertoll group of 10 trunks working presently 



* This, of course, does not reflect the added costs of the No. 4 switching equip- 

 I nient. 



