"Get the water into an expansion tank at the highest point as quickly as possible and let all the pipes descend from that tank through their whole length until they re-enter the boiler."

This is the substance of a remark once made by an old experienced florist and boiler-maker of Philadelphia, when conversing about greenhouse boilers and hot water heating. Since then I have acted upon the suggestion and put it to a practical test, and am thoroughly convinced that it is the most rational and effectual plan for heating greenhouses.

The circulation of water in pipes is caused by the variation in the specific gravity of water as its temperature varies. It is started and kept in motion by the application of that at the lowest point.

The warmest water always rises to the highest accessible point, tending to create a vacuum in the place whence it emanates, while cooling water is continually settling towards the lowest point and will enter the boiler by the lowest inlet, to replace that which has arisen.

"Hot" and "cool" are only comparative terms and we may say that the only "hot" water there is about a greenhouse heater is in the top of the boiler, because it commences to cool immediately upon leaving it, therefore it should commence to settle or flow down hill from that time. For this reason the water should rise perpendicularly from the top of the boiler into a capacious expansion tank situated several feet above it, and then flow down hill all the way around the houses until it reenters the boiler at its lowest part, having a regular descent throughout the entire length of from six to ten inches in one hundred feet.

There is then no need of expansion tanks in any other part of the arrangement.

Where the position of the boiler or the construction of the buildings will not admit of such an arrangement, take the shortest practicable line to the highest point accessible, and make the expansion tank there. I am not prepared to say what the elevation of the tank should be, but believe that for large boilers it should be at least ten feet.

Heat is the disturbing influence that starts and keeps the water in motion. The rapidity of circulation (upon which depends the heating capacity of the water), is governed, by the amount of pressure upon it and the intensity of the heat in the furnace. Therefore it is necessary to have the expansion tank elevated considerably above the pipes so that there will be a head of water exerting a continual pressure on every part. A displacement of water in any point is then quickly felt throughout the whole system.

A gradual descent of pipe all the way around, with as few bends and turns as possible is the simplest plan that can be adopted, and will give the most perfect and economical results; however it is posbible to make an endless variety of digressions from this rule, and still keep up a circulation which will be more or less retarded by every variation from the direct course.

A pipe may vary from a straight line in any way, provided there is a vent-hole for the escape of air at every point where it changes from ascending to descending grade. It may have a grade of ten inches in ten feet, or drop perpendicularly for several feet, and then fall only a very few inches in the next hundred feet, if necessary, but all such irregularities should be avoided as far as possible.

It is a matter of the first importance that there shall be no elevated points in which air can accumulate, because water will not circulate freely through pipes which contain air chambers.

There is always a considerable quantity of air in water, some of which is expelled by heating. This may be observed by standing a tumbler of water in a warm place. In a short time small bubbles of air will be seen on the inside of the glass which gradually rise to the surface of the water and disappear.

The air which is liberated from the water in a boiler or its connections rises to the highest part and flows on top of the water as long as it can ascend, but it will not flow down hill. If there is an opening at the highest accessible point the air will escape, if not, it remains there occupying space which should be filled with water.

Sometimes heaters which worked very well at one time become deranged by the settling of a pier or other disturbing cause, making a depression in a pipe at one spot, thus forming a barrier to the flow of air which wholly or partially checks the circulation of water.

The remedy for such a defect is to straighten the pipe so that the air can rise to some higher point of escape or else drill it at the summit of each undulation, and insert a stop-cock or other arrangement which can be opened occasionally or constantly for the escape of air.

There is, however, no necessity for vent-hole or other opening throughout the whole system of pipes if the expansion tank is at the highest point, and all the pipes descend from it through their entire circuit. In such an arrangement the air which is liberated in any point can always ascend, and will find its way up to the tank and escape there.