An advance was made when a chain, running along the entire length of the train, was arranged so as to pull on subsidiary chains branching off under each carriage and operating levers connected with brake blocks pressing on every pair of wheels. The guard strained the main chain by means of a wheel gear in his van. This system was, however, radically defective, since, if any one branch chain was shorter than the rest, it alone would get the strain. Furthermore, it is obvious that the snapping of the main chain would render the whole arrangement powerless. Accordingly, brakes operated by steam were tried. Under every carriage was placed a cylinder, in connection with a main steam-pipe running under the train. When the engineer wished to apply the brakes, he turned high-pressure steam into the train pipe, and the steam, passing into the brake cylinders, drove out in each a piston operating the brake gear. Unfortunately, the steam, during its passage along the pipe, was condensed, and in cold weather failed to reach the rear carriages. Water formed in the pipes, and this was liable to freeze. If the train parted accidentally, the apparatus of course broke down. Hydraulic brakes have been tried; but these are open to several objections; and railway engineers now make use of air-pressure as the most suitable form of power. Whatever air system be adopted, experience has shown that three features are essential:--(1.) The brakes must be kept "off" artificially. (2.) In case of the train parting accidentally, the brakes must be applied automatically, and quickly bring all the vehicles of the train to a standstill. (3.) It must be possible to apply the brakes with greater or less force, according to the needs of the case. At the present day one or other of two systems is used on practically all automatically-braked cars and coaches. These are known as--(1) The _vacuum automatic_, using the pressure of the atmosphere on a piston from the other side of which air has been mechanically exhausted; and (2) the _Westinghouse automatic_, using compressed air. The action of these brakes will now be explained as simply as possible. THE VACUUM AUTOMATIC BRAKE. Under each carriage is a vacuum chamber (Fig. 85) riding on trunnions, E E, so that it may swing a little when the brakes are applied. Inside the chamber is a cylinder, the piston of which is rendered air-tight by a rubber ring rolling between it and the cylinder walls. The piston rod works through an air-tight stuffing-box in the bottom of the casing, and when it rises operates the brake rods. It is obvious that if air is exhausted from both sides of the piston at once, the piston will sink by reason of its own weight and that of its attachments. If air is now admitted below the piston, the latter will be pushed upwards with a maximum pressure of 15 lbs. to the square inch. The ball-valve ensures that while air can be sucked from _both_ sides of the piston, it can be admitted to the lower side only.