THEORIES FOR TOLL TRAFFIC ENGINEERING IN THE U. S. A. 463 



\\ith an offered load a, v which has arisen in any manner of overflow from 

 earlier high usage groups, as illustrated in Fig. 24. 



This is found to be the case, as will be illustrated in several studies de- 

 scribed in the balance of this section. In the determination of the charac- 

 teristics of the overflow traffic a', v' in the cases of non-full-access groups, 

 such as Figs. 24(b) and 24(c), the equivalent straight group is visualized 

 [Fig. 24(a)], and the Eciuivalent Random load A and trunks S are found.* 

 I Using A, and *S + C, to enter the a and v curves of Figs. 12 and 13, a 

 , and v' are readily determined. To facilitate the reading of .1 and S, Fig. 

 25 1 and Fig. 26 f (which latter enlarges the lower left corner of Fig. 25) 

 have been drawn. Since, in general, a and v will not have come from a 

 simple straight group, as in Fig. 24(a), it is not to be expected that *S, 



OVERFLOW THEORY OBSD 



AVERAGE 5.76 5.98 



VARIANCE 12.37 14.89 



= = _ = OST N0.1 



t t t t t t t 



OFFER 10.66 3.24 2.44 11.46 9.81 9.59 1.42 ERLANGS 



OVERFLOW THEORY OBSD 



AVERAGE 2.83 2.87 



VARIANCE 3.35 3.34 



OST N0.14 



t t t t 1 1 t 



OFFER 2.52 1.08 0.94 0.94 0.59 1.13 0.85 ERLANGS 



Fig. 20 — Comparison of joint-overflow parameters; theory versus throwdown. 



* A somewhat similar method, commonly identified with the British Post 

 Office, which uses one parameter, has been employed for solving symmetrical 

 graded multiples (Ref. 9). 



t Figs. 25 and 26 will be found in the envelope on the inside back cover. 



