And now let us turn our attention to the sending machine. Upon its cylinder there is an arrangement which automatically reverses the current flowing to the main wire once in every revolution. Normally the current flows to the wire as described in the last chapter, carrying by means of its variations the details of the picture for reproduction by the receiving machine at the other end. But for an instant once in every revolution that current is interrupted and a current sent in the opposite direction instead. This the sending machine does of itself, quite automatically. And now the reader knows of all the apparatus; it remains only to see how the different parts work in combination. Standing by the sending machine we first of all turn on the current, which goes coursing along the wire to the distant station. Then we set the motor to work and the cylinder begins to rotate. Before it has completed a single revolution the "reverser" is operated, and just for a moment the reverse current goes to the wire. On arrival at the other end that lifts the catch and the receiving cylinder starts. That first partial revolution of the sending cylinder counts for nothing. Real business begins when the reverser first acts, and that is the moment when the receiving cylinder also begins to move. Similarly, when the sending cylinder stops it sends no more reversed currents, and so the receiving cylinder is caught by the catch and not released. So starting and stopping are quite automatic. The same arrangement enables a continual readjustment of the relative speed of the two cylinders to take place. With all the best devices, the tuning-forks and the rest, it is still impossible to attain perfect unanimity, but the variation in a single revolution cannot be enough to matter; it is only when the error in one revolution goes on multiplying itself that serious difference might arise, and that is prevented in the following beautifully simple way. The motor which drives the receiving drum is so regulated that it travels _slightly faster_ than does the other. Thus the receiving cylinder completes every revolution slightly in advance of the other, and consequently it is stopped and held by the catch every time. The catch retains it, of course, until the reverse current arrives and releases it. Thus not only does the sending cylinder start the other when the operations first commence, but it does so every revolution. Every revolution, therefore, the two cylinders start together. So the two cylinders are set, according to the frequency meter, at as nearly as possible exactly the correct speeds, and the action of the reverser, the reverse current and the catch, ensures quite automatically that at the commencement of every revolution there shall be perfect agreement between the two. No accumulation of errors can possibly occur, and the problem, though apparently so difficult, if not insuperable, at first sight, is surmounted. CHAPTER XV