When the question of laying an Atlantic cable began to be debated in the middle of the nineteenth century, Professor Thomson undertook the discussion of the theory of signalling through such a cable. It was not generally understood by practical telegraphists that the conditions of working would be very different from those to which they were accustomed on land lines, and that the instruments employed on such lines would be useless for a cable. Such a cable consists of a copper conductor separated from the sea-water by a coating of gutta-percha; it forms an elongated Leyden jar of very great capacity, which, when a battery is connected to one end of the conducting core, is gradually charged up, first at that end, and later and later at greater distances from it, and then is gradually discharged again when the battery is withdrawn and the end of the conductor connected to earth. Here, again, an application of Fourier's analysis solved the problem, which, with certain modifications, and on the supposition that the working is slow, is essentially the same problem as the diffusion of heat along a conducting bar, or the diffusion of a salt solution along a column of water. The signals are retarded (and this was one of the results of the investigation) in such a manner "that the time required to reach a stated fraction of the maximum strength of current at the remote end," when a given potential difference is applied at the other, or home end, is proportional to the product of the capacity and resistance of the cable, each taken per unit of the length, and also proportional to the square of the length of cable. In other words, the retardation is proportional to the product of the resistance of the copper conductor and the total capacity of the cable. This gave a practical rule of great importance for guidance in the manufacture of submarine cables. The conductor should have the highest conductivity obtainable, and should therefore be of pure copper; the insulating covering should, while forming a nearly absolutely non-conducting sheath, have as low a specific inductive capacity as possible. The first of these conditions ran counter to some views that had been put forward, to the effect that it was only necessary to have the internal conductor highly conducting on its surface; and some controversy on the subject ensued. The inverse square law, as it was called, was vehemently called in question, from a mistaken interpretation of some experiments that were made to test it. For if the potential at the home end be regularly altered, according to the simple harmonic law, so that the number of periods of oscillation in a second is n, the changes of potential are propagated with velocity 2√(πn⧸cr), where c and r are the capacity and resistance of the cable, each taken per unit length. In this case, for a long cable, there is a velocity of propagation independent of the length; and this fact seems to have misled the experimenters.