598 BELL SYSTEM TECHNICAL JOURNAL 



In addition, at the mid splice of each loading section, a test spHce 

 was made to equaHze the capacity deviations. For these spHces 

 special Hnen boarding strips were used, each with 40 holes designated 

 by a capacity ranging from about 15 per cent below to about 15 per 

 cent above the expected average capacity of 1500 feet of cable. Small 

 inexpensive capacity meters were used, and each pair was placed in 

 the hole in the boarding strip corresponding to its capacity. The 

 pairs were then spliced high to low capacity. This method did not 

 require special testers, and substantially reduced splicing manhours. 



Upon the completion of the splicing in each section, the section 

 capacity of each pair was measured and recorded, and from this was 

 determined the root mean square of the capacity deviations from the 

 average capacity. These deviations combined with the deviation of 

 the loading section average capacitances, loading coil spacings, and 

 loading coil inductances, gave an irregularity function of 2 per cent 

 which is almost identical with that for 19-gauge B88-50 cable. From 

 this irregularity ^ function a 63 per cent return loss frequency curve 

 was obtained, which is shown as curve M ' in Fig. 8. 



When 15 miles of the cable had been completed south from Barrie, a 

 100-pair cross-connecting box was temporarily spliced in so that data 

 could be obtained as a further check on the design estimates. 



For crosstalk tests each pair was terminated at the box in a 1700 ohm 

 resistance, and measurements were made of all pair combinations 

 (approximately 5000). For these measurements a 15A oscillator and 

 2A Noise Measuring Sets were used, thereby very materially reducing 

 the manhours required as compared with the labour that would have 

 been required had crosstalk measuring sets been used. Analysis of 

 these tests indicated 99 per cent of combinations better than 76.0 db, 

 an average of 86.6 db and 99.5 per cent meeting the required 74.5 db 

 of the preliminary studies. 



For attenuation measurements, the pairs were looped back at the 

 cross-connecting box. In order to obtain a value of the attenuation 

 at a known temperature, a complete set of measurements was made 

 at about 6 o'clock in the morning after the resistance of one of the pairs 

 had been found to have ceased dropping due to temperature change 

 and the outside temperature at the Barrie office had been very nearly 

 constant for about one-half hour. During the time the attenuation 

 measurements were being taken, air temperatures were measured at 

 four places along the 15-mile length of cable. From these tests the 

 average 1000 cycle attenuation at 62° F. was found to be 0.26 db per 



* See "Irregularities in Loaded Telephone Circuits," George Crisson, Bell System 

 Technical Journal, October 1925. 



