THE INTERCONNECTION OF TELEPHONE SYSTEMS 



555 



when the above typical total loads are substituted in the generalized 

 graded formula. 



TABLE IV 



The Effect of Subgroup Load Variations on the Grade of Servtce^36 Trunks 

 Total Load Submitted = A = 18, a; = 4, y = 16, g = 5. 



TABLE V 



The Effect of Subgroup Load Variations on the Grade of Service — 52 Trunks 



Total Load Submitted = A = 33.80, x = 8, y = 12, g = 5. 



In the last two columns of Tables IV and V are indicated the 

 measures of subgroup load variation and the expected grades of 

 service, respectively, for various theoretical unbalances studied on 

 these two symmetrical graded multiples. Figs. 20 and 21 indicate 

 the rapidity with which the overall probability of loss on the general- 

 ized "Full Gain" formula basis may be expected to rise with increases 

 in the load unbalances in the subgroups. Reading off on the abscissa 

 of Fig. 21 a measure of subgroup unbalance of .45, for example, 

 indicates that for a total load of ^ = 33.80 being submitted to 52 

 trunks arranged in a grade of five subgroups and an access of 20, the 

 correct probability of loss is not the "Full Gain" efficiency value of 

 .00619 but rather the more conservative figure of about .0090. 



