INTERFERENCE ON SUBMARINE TELEGRAPH CABLES 411 



components lying within a certain frequency range, the limits of which 

 depend upon the speed of signalling. This is due to the fact that the 

 characteristics of an ordinary submarine cable are such that the low 

 frequency components of a signal are transmitted with much less di- 

 diminution of amplitude than are the higher frequency components. 

 Consequently "^ it is found necessary, in order to render the signal 

 intelligible, to employ a correcting network at the receiving terminal, 

 one function of which is to attenuate the arriving low frequency com- 

 ponents so that they finally are in the proper proportion to the higher 

 frequency components. Also it is found that frequencies which are 

 higher than about one and one-half times the signal frequency are not 

 required in order to obtain intelligible signals, so that the receiving 

 network can be designed to remove disturbances of the higher fre- 

 quencies. The receiving apparatus therefore acts as a band filter 

 towards the interference arriving at the terminal and emphasizes the 

 part played by the components of interference of frequencies in the 

 neighborhood of the signal frequency. On this account it is possible, 

 in the majority of cases, to obtain the significant portion of the sus- 

 ceptibility-frequency curve by limiting the integration in (5) to the 

 portion of the cable submerged to a depth of approximately 1000 feet 

 or less, since, as has been previously indicated, only disturbances of 

 extremely low frequencies are picked up on the deep water portion 

 of the cable. 



Given the problem of predetermining the interference at the ter- 

 minal of a projected cable, the following procedure can be employed : 



1. Over a period of time sufficiently long, a series of records of in- 

 terference is taken on a cable terminating in the same general neigh- 

 borhood as the proposed cable. Oscillographic records of the type 

 shown in Fig. 1 are very desirable for this purpose. 



2. From these records, and from the computed susceptibility-fre- 

 quency curves of the existing and projected cables the interference on 

 the latter can be predicted. 



The method just described v/as applied to predetermine the inter- 

 ference at the terminals of the Nev/ York-Azores permalloy loaded 

 cable. At the Azores terminal the cable reaches deep water within a 

 few miles of the terminal, and the results indicated that the magnitude 

 of interference to be expected would be sufficiently small to permit of 

 signalling at the speed at which it was desired to operate. At the New 

 York terminal, however, the ocean for a distance of about 100 nautical 

 miles is comparatively shallow, and cables in this vicinity are exposed 

 to rather severe disturbances. This is partly due to unusually strong 



^ See Milnor "Submarine Cable Telegraphy," Trans. A. I. E. E., Vol. 41, p. 20, 1922. 



