Unfortunately, hot-water systems are not always arranged as they should be, and both pipes and radiators often contain "traps" for air-bubbles, unprovided with means of escape for the air to the expansion-tank, so that, in place of such means, air-valves must be used, either opened by hand, or automatically operated; by means of a little float inside them, to prevent the circulation from being stopped, and the radiators or pipes from becoming cold.

Where a hot-water radiator will not heat, there is reason to suspect a stoppage of the flow from one of these two causes; and the course of the pipes should be examined, to ascertain whether a dip downward, or a bend upward, is responsible for the trouble. If the former is discovered, the pipes should be straightened; or, if that is impracticable, a rising loop may be made above the dip, remembering to provide an escape for air-bubbles gathering at the top of the loop; or, in some cases, a pipe may be carried down from the bottom of the dip to a return-pipe. If, on the contrary, an upward bend is found, without provision for the escape of air, such escape should be provided, either by drilling the pipe, and putting in an automatic air-valve, or by connecting the top of the bend with a pipe communicating with the expansion-tank. If neither of these is found practicable, relief may sometimes be obtained by carrying either the flow or return pipe from the ailing radiator, or both, directly to the boiler, instead of taking them as branches from a main vertical pipe. In some small houses every radiator has its own separate flow and return pipe from the boiler, the upper ends of the loops thus formed being connected with the expansion-tank. Although this arrangement involves a greater length of pipe than branching from a main supply and return, the pipes used are smaller, and, therefore, much cheaper, and a good deal of complication is avoided, while the circulation in the radiators is always good.

Circulation in a steam system is as liable to interruption as that in a hot-water system, but in a different way. The pressure under which the steam is driven through the pipes prevents stoppage by the collection of water in downward dips, or of air in moderate upward bends; but radiators, in which the steam enters at the bottom, and, after a devious course through the radiator, escapes again at the bottom, are very apt to fill with air, so that steam cannot be forced into them, and they are usually fitted with air-valves, automatic or otherwise, to allow the air to escape. Many steam radiators will not heat at all until the air-valve is opened; and, as a sudden rush of steam and water usually follows the expulsion of the air, such valves, unless automatic, must be carefully watched, when opened, or the room may be flooded with water, or filled with steam, to the ruin of the furniture.

In a large system of steam heating every radiator has two pipes, one for supplying it with steam, and the other for draining off the water which forms in it from the condensation of the steam, and returning it to the boiler; the return-pipe being usually a size smaller than the steam-pipe. Most house systems are, however, for the sake of economy, arranged for what the steam-fitters call a "one-pipe job," the steam entering the radiator through the same pipe which carries the condensed water away. Where the pipes in a one-pipe system are large enough, they are not likely to give trouble; but, where they are of insufficient size, the steam rising through them to the radiators has not room enough to pass amicably by the condensed water flowing in the opposite direction, on its way down to the boiler, and pushes it back in part, at the same time that it squeezes past it in what space may be left. The water thus held back by the current of steam accumulates, until there is enough of it to fill the bore of the pipe completely, cutting off the passage of steam beside it. The steam then, being unable to pass by, pushes the plug of condensed water violently before it, until it is dashed against the inside of a radiator, or a turn in the pipe, producing the "hammering" noise so common in steam-heating systems, and, with high-pressure steam, sometimes breaking the radiators or fittings.

Single-pipe system.

Even with a two-pipe system, where the steam enters the radiators by a different pipe from that which carries off the condensed water, hammering often occurs in the steam-pipes, from the collision of the steam with water condensed in the steam-pipes themselves. It is obvious that, wherever the steam is exposed to cooling, water will be condensed from it, whether the process takes place in a supply-pipe or in a radiator; and, if the supply-pipes are long, particularly if they have also branches, the amount of water condensed in them, and flowing down, in opposition to the current of steam, may easily be sufficient, if pressed back by the steam, to fill the bore of the pipe. The remedy for hammering in steam-pipes, whatever may be its source, is simple in theory, but is sometimes expensive. It is obvious that, if the steam and the condensed water can be made always to move in the same direction in the pipes, there will be no collisions, and no hammering; and all steam-heating systems should be arranged as far as possible on that principle. In Figure 11, for example, it is evident that the steam-pipe S, supplying the radiators above, will, even if the radiators have separate returns, and much more if they do not, be at all times partially filled with water condensed from the steam passing through it, and flowing back to the boiler; and on very cold days, when condensation is rapid, or early in the morning, when the steam is first turned into cold pipes, there will be a large amount of water running back, in a direction contrary to that of the steam current, and collisions and hammering are at such times inevitable. By arranging the same pipe as shown in Figure 12, and adding the "relief " or "drip" pipe shown at D, the condensed water in the horizontal portion of the pipe will flow in the same direction as the steam toward the point P, where it is immediately drained off through the pipe D into the main return-pipe R. A little water will still condense in the vertical steam-pipe between P and the radiator, and will flow down in opposition to the current of steam, but it is too small in amount to give trouble, and is drained off as soon as it reaches the point P, by the drip-pipe D.