The present paper presents an extensive theoretical investigation of the impedance of the “sea return” of various types of submarine cables. In the case of the cables used for submarine telegraphy the impedance of the sea return has been practically negligible because of the low frequencies involved. For these low frequencies the cross-section of the return path is very large and its resistance low, even though the specific resistance of sea water is of the order of ten million times that of copper. As the frequency of the cable current is raised, however, the return currents crowd in nearer the cable and the resistance of the return path is increased. For frequencies in and above the telephone range, the return currents are forced into the steel armor wires around the cable and into the water just outside of the insulation. The small cross-section of the water involved and the loss in the armor wires cause the resistance of the return path to become a very large part of the total resistance of the circuit. The present investigation led to the conclusion that the resistance of the return path could be greatly diminished by winding a low resistance conductor in the form of a copper tape immediately around the gutta percha insulation applied to the core of the cable. The concentric, cylindrical conductor thus formed lies within the armor wires but is not insulated from them and the sea water. Estimates of the sea return which would have been obtained in the Key West-Havana cable if no copper tape had been provided give values of 4, 6.5, and 8 ohms per nautical mile at 1,000, 3,000 and 5,000 cycles. The resistance actually obtained with the copper tape does not exceed 1.7 ohms at 5,000 cycles. The greater values would have increased the attenuation by approximately 30% at 1,000 cycles and by 50% at the two higher frequencies. The present cable permits of the operation of a carrier telegraph channel at 3,800 cycles, this lying above the range of telephone frequencies. The paper gives a comparison of the theoretical conclusions with experimental data and the agreement is so satisfactory as to indicate that the theory is a reliable guide in the design of such a cable. — Editor.