CONQUEST OF DISTANCE BY WIRE TELEPHONY 361 



lengths of commercial cable made up, the unbalances were worse than 

 those in the 1908 experimental lengths. From then on, engineers from the 

 Engineering Departments of the Western Electric Company and the 

 American Company coo])erated closely with the factory engineers through- 

 out the entire manufacturing period in working out important fundamental 

 improvements. One of the greatest difficulties was in obtaining sufficiently 

 symmetrical twisting of the individual pairs, prior to forming the quads. 

 Machine limitations and the need for conductors of identical size and duc- 

 tility, insulated with exactly the same thickness of paper, were factors in 

 this problem. Other difficulties too many and too involved for present dis- 

 cussion were also encountered. 



At the beginning of manufacture it was appreciated that it would probably 

 be impossible with known techniques to obtain quadded cable that would be 

 completely satisfactory from the crosstalk standpoint, especially phantom- 

 to-side crosstalk, without resorting to capacitance unbalance adjustments 

 during installation of the cable. Some preliminary consideration was given 

 to the use of balancing condensers. Further studies of possible methods of 

 field balancing led to the development of a technique for measuring the 

 capacitance unbalances in adjacent lengths of cable and selectively splicing 

 the conductors of the quads of one length to those of another length in such 

 a manner that the like-type unbalances would tend to annul one another. 

 Suitable field test sets were developed for determining the magnitudes of 

 the cable capacitance unbalances and their relative phase relations. The 

 planned splices made in accordance with the proposed technique became 

 known as capacitance unbalance test splices. 



In splicing the Boston-Neponset cable, a total of 7 capacitance unbalance 

 test splices were made in each full loading section at intermediate points 

 approximately | of a full loading section apart. The first set of test splices 

 was made at the |, |, f , and | section points. Then "semi-final" tests were 

 made at the \ and f points. The test splice made at mid-section was most 

 important because it was the final test splice. Splices required at points 

 between the test splices were made on a random basis. The test splices 

 were made primarily for the reduction of phantom-to-side unbalance. 

 When individual quads had objectionably high side-to-side unbalances, 

 reductions in the residual unbalances could usually be obtained by planned 

 splicing to other quads having high side-to-side unbalances. 



The test splicing procedure was very effective in reducing the pile-up 

 of objectionable unbalances. In general, the maximum residual unbalances 

 per loading section were kept below a predetermined tolerable value. The 

 average residual phantom-to-side capacitance unbalance per full loading 

 section turned out to be of the same order as the average unbalance in the 

 individual (approx.) 500-foot lengths when they left the factory. Statisti- 



