be magnified again. With correct adjustments this experiment will
prove more surprising than the first. Put on a third and still larger
tube in the same way, and still larger surprise will meet a still
larger volume and force of blast, like a stiff breeze set in motion by
the puny effort of a single expiration. Of course, the prime impulse
must bear a certain proportion to the result; and the inductive or
tractional friction of the initial blast, of flame or breath, will be
used up at length unless re-enforced. In ventilating practice, there
_is_ such re-enforcement, from an excess of gravity in the cooler
atmosphere outside the flue in which the flame is operating with its
heat as well as its ascensional traction; so that there has been found
no limit to the extensions and fresh inductions that may be added to
the first or trunk flue, with increase rather than diminution of power
at every point. But the terms on which such extensions must be made
have been referred to in our illustration, and must be accurately
ascertained and observed. They constitute what is, in effect, the
third factor in the phenomenon of a roaring draught, and also,
therefore, ineffective ventilation. That is, the entering or induced
current of air must always find its channel of progress and exit
certain correct degrees larger than the opening by which it entered.
Every one knows that a stove or chimney wide open admits of but little
suction in connection with even the blaze of paper or shavings.

The mobility of air seems almost preternatural, when the proper
conditions for setting a current in motion are supplied. But without a
current established, it is surprising in turn to find how obstinately
and elusively immovable it can be. It is like tossing a feather; or
trying to drive a swarm of flies; dodging and evading every impulse
applied. But, given a flue, to define and conduct a stream; an upright
flue, to take advantage of the slighter gravity of the warmed air