478 THE BELL SYSTEM TECHNICAL JOURNAL, MARCH 1956 



Table X — Load Overflowing Crossbar Tandem 



Average. 

 Variance , 



Observed 



6.47 

 33.48 



and those calculated (in the above manner) are given in Table IX. 

 Fig. 33 shows the distribution of XBT offered loads, observed and calcu- 

 lated. The agreement is very satisfactory. The random traffic (Poisson) 

 distribution, is of course, considerably too narrow. 



In a manner exactly similar to previous cases, the Ecjuivalent Random 

 load method was applied to the XBT group to obtain estimated param- 

 eters of the traffic overflowing. Comparison of observation and theory 

 at this point is given in Table X. 



Fig. 34 shows the corresponding observed and calculated distributions 



0.15 



0.10 



f(n) 



0.05 



)RANDOM TRAFFIC 



THEORY OBSD 



AVG 6.55 6.47 



VAR 23.80 33.48 



^'NEGATIVE BINOMIAL 



5 10 15 20 25 30 35 



n=NUMBER OF SIMULTANEOUS CALLS 



P^n 



_^^RANDOM TRAFFIC 

 --NEGATIVE BINOMIAL 



THROWDOWN 



5 10 15 20 25 30 35 



n = NUMBER OF SIMULTANEOUS CALLS 



Fig. 34 — Distribution of calls from crossbar tandem trunks; negative binomial 

 theory versus throwdown observations. 



! 



