494 THE BELL SYSTEM TECHNICAL JOURNAL, MARCH 1956 



A choice of all weights g equal to unity will often be satisfactory for the 

 present purpose. The desired high usage group size for the first routed 

 traffic is then found from standard £'i-tables showing trunks x required, 

 as a function of offered traffic a and proportion overflow b. 



Since the different parcels of traffic have varying proportions h of their ' 

 loads overflowing to the final route, by equation (34) the parcel with 

 the largest proportion will determine the permitted value of R2 . Thus ' 



R2 = P/&max (36) 



where P is the specified poorest overall service (say 0.03) for any parcel. 

 It may be noted that on occasion some one parcel, perhaps a small one, 

 may provide an outstandingly large bmax value, which will tend to give 

 a considerably better than required service to all the major traffic 

 parcels. Some compromise with a literal application of a fixed poorest 

 service criterion may be indicated in such cases. 



An alternative and somewhat simpler procedure here is to use an 

 average value b in (36) instead of ^max , with a compensating modifica- , 

 tion of F, so that substantially the same R2 is obtained as before. The 

 allowance in P will be influenced by the choice of weights g in (35). It 

 will commonly be found in practice that overflow proportions to final 

 groups for large parcels of traffic are lower than for small parcels. Choos- 

 ing all weights, as unity, as opposed to weighting by traffic volumes for : 

 example, tends to insert a small element of service protection for those , 

 traffic parcels (often the smaller ones) with the higher prportionate high . 

 usage group overflows. 



Having determined R2 , a ready means is needed for estimating the 

 required number of final route trunks. Curves for this purpose are pro- 

 vided on Figs. 46 to 49, within whose range, R2 = 0.01 to 0.10, it will 

 usually be sufficiently accurate to interpolate for trunk engineering 

 purposes. These F2-curves exactly parallel the i?i-curves for use when 

 first routed traffic is offered directly to the final group without benefit 

 of individual high usage trunks. If R2 is well, outside the charted range 

 a run-through of the ER calculations may be required. 



8.3 Area in Which Significant Savings in Final Route Trunks are Realized 

 by Allowing for the Preferred Service Given a First Routed Traffic Parcel 



Considerable effort has been expended by alternate route research 

 workers in various countries to discover and evaluate those areas where 

 first routed (random) traffic ofl'ered to a final route enjoj^s a substantial 

 service advantage over competing parcels of traffic which have over- 



, 



