542 



BELL SYSTEM TECHNICAL JOURNAL 



The data were summarized by size of group and by circuit usage. Sepa- 

 rate counts were maintained for groups with and without alternate routes 

 and for person and station traffic. The following tabulation shows the 

 type of data available for each point, i.e., each size of group at each level of 

 usage. 



One Trunk — with Alternate Route — 61-70% Use 



Average Speed— All Calls 3 .86 Mins. 



Average Speed — NC not Encountered 2 .04 Mins. 



Average Delay Due to NC (Trunk Speed) 1 .82 Mins. 



The results were plotted for each level of usage by steps of 10% as shown 

 in P'ig. 1, using a 3-point moving average to smooth out the deviations and 

 to establish a more definite trend. Each of these curves was then redrawn 

 in relation to the others and combined results are shown on Fig^ 2. 



The delay intervals indicated in Fig. 2 represent the total delay which 

 resulted from the fact that there was no circuit available when the operator 

 was first ready to make use of one. It includes not only the time spent in 

 waiting for a circuit to become idle but also the time required for the operator 

 herself to return to that call if she had engaged in some other work in the 

 meantime. If the operator is not free to utilize the circuit as soon as it 

 becomes available some other operator may use it for another, later call. 

 The subsequent call is then delayed less than the average, or not at all, 

 but the original call is delayed longer than the average. While the delays 

 experienced by individual calls may vary considerably from the average, 

 the data have been treated in terms of averages for engineering purposes. 



Mathematical Formulae 

 The summarized data were referred to the different mathematical ex- 

 pressions frequently applied to trunking problems, such as the. Poisson, 

 Erlang "B" and Erlang "C" formulae. It was found that the observed 

 average NC delays were considerably shorter than the theoretical average 

 delays in those formulae which make allowance for variable holding times, 

 such as would be encountered in local trunking where the average trunk use 

 is short and the deviations from average on a percentage basis are apt to be 

 appreciable. 



