The planning of hot-water pipes is much more difficult than that of steam-pipes, but the general principle is that the pipes must rise all the way from the boiler to the farther end, where there must be an expansion-box or supply-cistern to allow the water to rise and fall and dissolved air to escape. From this the pipes must fall more or less, throughout the distance, back to the boiler, entering it at the bottom. If at any point the pipe has to fall, leaving an upward bend, a tap must be provided for the escape of air, but such upward bends are a fertile source of difficulty and failure of action. With long runs of either steam or water pipes, arrangements must be made to allow of expansion and contraction, which will amount to 1-2 in. per 100 ft., according to the temperature employed. If one end of the system can be left free, all that is needed is to support the pipes on rollers (pieces of old pipe may be used); if not, stuffing-boxes must be provided. The air heated by boilers, and other sources of waste heat, may often be utilised for heating purposes, but generally requires to be driven by a fan, unless the drying-room can be arranged directly above the source of heat. If air has to be conveyed, the air-ways must be of ample size, and if the ascending force of heated air be relied on, passages less than 2 ft. sq. are seldom of much use. This ascending force is generally much overrated where the differences of temperature are so small as those employed in a drying-room. In a boiler chimney, where the temperature of the escaping gases is 552° F. (289° C.), the specific gravity of the air is about half that outside, and a chimney of 50 ft. in height gives a draught equal to the pressure of a column of about 1/3 in. of water, and the hot gases theoretically have a velocity of about 80 ft. per second; whereas the same chimney with a difference of temperature of 30° F. would have a draught equal to 1/300 in. of water only, and a velocity of 8 ft. per second. The following table will enable the reader to calculate approximately loss in friction in air-passages and the pressure required to pass a given volume of air. The pressure needed increases in proportion to the length of the pipe and the square of the velocity of the current of air to be passed. Thus if we double the length of the pipe we must double the pressure to pass the same quantity; and in order to double the quantity of air through a given pipe, the pressure must be quadrupled. Table IV. Head, or Difference of Pressure at the two ends of a Circular Pipe 1 yd. long in inches of water required to pass 1000 cub. ft. of air per minute.