When the soundings are complete the cable-ship takes up her task. The cable is coiled in tanks on board, and is kept constantly under water to prevent injury to the gutta-percha insulation by overheating. As each section is placed in the tank, the ends of it are led to a test-box, and labelled so that they can be easily recognized. Insulated wires run from the test-box to instruments in the testing-room, so that the electrical condition of the whole cable is constantly under observation. During the whole time the cable is being laid its insulation is tested continuously, and at intervals of five minutes signals are sent from the shore end to the ship, so that a fault is instantly detected. The cable in its tank is eased out by a number of men, and mechanics are posted at the cable drums and brakes, while constant streams of water cool the cable and the bearings and surfaces of the brakes. The tension, as shown by the dynamometer, is at all times under careful observation. When it becomes necessary to wind back the cable on account of some fault, cuts are made at intervals of a quarter or half a mile, tests being made at each cutting until the fault is localized in-board. As soon as the cable out-board is found “O.K.,” the ends are spliced up and the paying-out begins again. If the cable breaks from any cause, a mark-buoy is lowered instantly on the spot, and the cable is grappled for. This may take a day or two in good weather, but a delay of weeks may be caused by bad weather, which makes grappling impossible. The practical working of a submarine cable differs in many respects from that of a land telegraph line. The currents used in submarine telegraphy are extremely small, contrary to the popular impression. An insulated cable acts like a Leyden jar, in the sense that it accumulates electricity and does not quickly part with it, as does a bare overhead wire. In the case of a very long cable, such as one across the Atlantic, a current continues to flow from it for some time after the battery is disconnected. A second signal cannot be sent until the electricity is dissipated and the cable clear, and if a powerful current were employed the time occupied in this clearing would be considerable, so that the speed of signalling would be slow. Another objection to a powerful current is that if any flaw exists in the insulation of the cable, such a current is apt to increase the flaw, and finally cause the breakdown of the line.