o than by other modes,
and the changes of temperature of the outer air do not materially
affect it. But the case is different with regard to the air of the
house, which is frequently reduced below the freezing point, in severe
weather. If the bottom heat is of the required temperature, any attempt
to counteract the coldness of the air of the house by increasing the
fire, would produce an injurious excess of bottom heat. It is evident
that while the required supply of heat for the bottom is uniform, and
that for the top exceedingly irregular, both objects cannot be properly
secured except by a separate supply of heat for each. For these reasons
we would employ a hot water pipe or pipes, passing around the house, on
the same level with the tanks, supplied with a valve to regulate the
heat at pleasure, or a brick smoke flue constructed in the usual manner.

Tanks are usually divided in the centre, thus forming channels for the
flow and return circulation side by side, equalizing the temperature
throughout their whole length. This form is sometimes departed from by
carrying the tank around the house, and connecting each end with the
boiler, but in this case, except in small houses, a uniform temperature
cannot be maintained, as the water will have lost several degrees of
heat before it has accomplished its circuit. Another arrangement is to
connect the remote end of the tank by an iron pipe for the return
circulation, passing under the tank the whole distance to the boiler.
This is not as perfect and effective an arrangement of pipes and tanks
as that before referred to, as in this case we do not have the heat from
the pipe under control.

A writer in a late number of the "Gardeners' Monthly," gives the
following description of tanks erected by him to obviate excessive
moisture and radiate a portion of their heat into the atmosphere of the
house.

"In the winter of 1863-4, I finished two span-roof houses, each 60 feet
in length, with water tanks three feet in width, running entirely around
on both sides of each house, and heated by a single furnace. The tanks
were made with wooden bottoms and sides, and covered with slate
carefully cemented. My design was to heat the houses entirely by the
tanks, by far the larger portion of the heat being given off from the
slate covering, and as a bottom heat for plants. As I understand the
various writers upon this subject, this is the approved plan. But I have
found considerable diff