t prolific cause of explosions is this same
strain from unequal expansions.

[Illustration: Fig. 4]

[Illustration: 386 Horse-power Installation of Babcock & Wilcox Boilers
at B. F. Keith's Theatre, Boston, Mass.]

Having thus briefly looked at the advantages of circulation of water in
steam boilers, let us see what are the best means of securing it under
the most efficient conditions We have seen in our kettle that one
essential point was that the currents should be kept from interfering
with each other. If we could look into an ordinary return tubular boiler
when steaming, we should see a curious commotion of currents rushing
hither and thither, and shifting continually as one or the other
contending force gained a momentary mastery. The principal upward
currents would be found at the two ends, one over the fire and the other
over the first foot or so of the tubes. Between these, the downward
currents struggle against the rising currents of steam and water. At a
sudden demand for steam, or on the lifting of the safety valve, the
pressure being slightly reduced, the water jumps up in jets at every
portion of the surface, being lifted by the sudden generation of steam
throughout the body of water. You have seen the effect of this sudden
generation of steam in the well-known experiment with a Florence flask,
to which a cold application is made while boiling water under pressure
is within. You have also witnessed the geyser-like action when water is
boiled in a test tube held vertically over a lamp (Fig. 3).

[Illustration: Fig. 5]

If now we take a U-tube depending from a vessel of water (Fig. 4) and
apply the lamp to one leg a circulation is at once set up within it, and
no such spasmodic action can be produced. Thus U-tube is the
representative of the true method of circulation within a water-tube
boiler properly constructed. We can, for the purpose of securing more
heating surface, extend the heated leg into a long incline (Fig. 5),
when we have the well-known inclined-tube generator. Now, by adding
other tubes, we may further increase the heating surface (Fig. 6), while
it will still be the U-tube in effect and action. In such a construction
the circulation is a function of the difference in density of the two
columns. Its velocity is measured by the well-known Torricellian
formula, V = (2gh)^{.5}, or, approximately V = 8(h)^{.5}, h being measured
in terms of the lighter fluid. This velocity will increase until