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



485 



Table XV — Comparison of E.R. Theory and Throwdowns on 



Disparity of Loss Between High Usage Overflow and 



Random Offer to a Final Group 



(8 trunks in each high usage group; 9 final trunks serving 2.0 erlangs 



high usage overflow and 2.0 erlangs first routed traffic.) 



Limited data are available showing the disparity of Ri and Ro in 

 actual operation in a range of load and trunk values well beyond those 

 for which Ri values have been calculated. Special peg count and over- 

 flow registers were installed for a time on the final route during the 1955 

 Newark alternate route tests. These gave separate readings for the calls 

 from high usage groups, and for the first routed Newark to Pittsburgh 

 calls. Comparative losses for 17 hours of operation over a wide range of 

 loadings are shown on Fig. 39. The numbers at each pair of points give 

 the per cent of final route offered traffic which was first routed (random). 

 In general, approximately equal amounts of the two types of traffic were 

 offered. 



In 6 of the hours almost identical loss ratios were observed, in 7 hours 

 the overflow-from-high-usage calls showed higher losses, and in 4 hours 

 lower losses, than the corresponding first routed calls. The non-random 

 calls clearly enjoyed practically as good service as the random calls. This 

 result is not in disagreement with what one might expect from theory. 

 To compare directly with the Newark-Pittsburgh case we should need 

 curves on Fig. 38 expanded to correspond to A', V values of (50, 85) 

 to (120, 200). Examining the mid-range case of C = 65, A' = 70, V = 

 120, we find A = 123, >S = 54. Here A is approximately 2C; extrapolat- 

 ing the A = 2C curves of Fig. 38 to these much higher values of A and C 

 suggests that R2/R1 w^ould be but little different from unity. 



