970 THE BELL SYSTEM TECHNICAL JOURNAL, JULY 195G 



as circuit order changes were made. Many small troubles were found and 

 cleared. As the result of such rigorous circuit order work and the use of 

 statistical analyses, the distribution grades at the end of the trial were 

 reduced to about 1 db and the biases were brought close to zero. 



The maintenance activities were conducted by the regular test forces 

 during normall}' available maintenance time. Although the initial work 

 in\'oh'ed in cleaning up the trunks necessitated some slippage in the 

 periodic maintenance tests, troubles requiring realignment were eventu- 

 ally reduced to the point where it became possible to carry on the periodic 

 testing work concurrently with the more rigorous circuit order work. 



Complex Layouts 



During the field trial of the automatic transmission test and control 

 circuit discussed in a companion paper, there was an opportunity for 

 studying transmission data taken on intertoll trunks of greater length 

 and complexity of layout. These trunks were composed of two or more 

 carrier links and connected Washington, D.C. to several outlying points; 

 namely, Atlanta, Georgia; Boston, Massachusetts; Hempstead, New 

 York; New York, New York; Oakland, California; and Richmond, Vir- 

 ginia. A total of 231 trunks were in the groups. When the trial began, 

 without preliminary rigorous circuit order work on the trunks involved, 

 the distribution grade was 2.26 db and the bias was -|-0.35 db. Mainte- 

 nance investigation was initiated only when trunks were foimd to have 

 departed more than a prescribed amount from their specified net losses. 

 Initially this value was 4 db and later it was reduced to 3.5 db. 



As many of these wide de\'iations were investigated and corrected as 

 available manpower permitted. The layouts were so complex, however, 

 that it was found impracticable to give prompt attention to all of them; 

 and in many cases it was impossible to check carrier systems that were 

 suspected of being the source of some of the deviations. At the end of 

 the trial the distribution grade had been reduced from the original 2.26 

 db to a range of about 1.8 to 2 db. The bias had not been changed sig- 

 nificantly from the original -1-0.35 db. 



These results indicated very little improA'ement from the limited re- 

 adjustments found practicable diu'ing the tests. Analysis of the test re- 

 sults has shown that transmission maintenance methods must be im- 

 proved in some respects. An example of this was a case where the data 

 indicated several trunks to be affected by excessive variation from some 

 common cauKC. This was traced to a group pilot being out of limits. If 

 routine maintenance methods had indicated this difficulty earlier, the 

 amount of time in which service could have been affected by thesc^ trunks 

 would have been reduced. This is important because of the difficulty of 

 finding evidence of common trouble sources, with complex layouts. 



