was made to transverse the furnace several times before being delivered
to the drum above. The tendency here, as in all similar boilers, was to
form steam in the middle of the coil and blow the water from each end,
leaving the tubes practically dry until the steam found an outlet and
the water returned. This boiler had, in addition to a defective
circulation, a decidedly geyser-like action and produced wet steam.

[Illustration: No. 12]

All of the types mentioned, with the exception of Nos. 5 and 6, had
between their several parts a large number of bolted joints which were
subjected to the action of the fire. When these boilers were placed in
operation it was demonstrated that as soon as any scale formed on the
heating surfaces, leaks were caused due to unequal expansion.

No. 12. With this boiler, an attempt was made to remove the joints from
the fire and to increase the heating surface in a given space. Water
tubes were expanded into both sides of wrought-iron boxes, openings
being made for the admission of water and the exit of steam. Fire tubes
were placed inside the water tubes to increase the heating surface. This
design was abandoned because of the rapid stopping up of the tubes by
scale and the impossibility of cleaning them.

[Illustration: No. 13]

No. 13. Vertical straight line headers of cast iron, each containing two
rows of tubes, were bolted to a connection leading to the steam and
water drum above.

[Illustration: No. 14]

No. 14. A wrought-iron box was substituted for the double cast-iron
headers. In this design, stays were necessary and were found, as always,
to be an element to be avoided wherever possible. The boiler was an
improvement on No. 6, however. A slanting bridge wall was introduced
underneath the drum to throw a larger portion of its heating surface
into the combustion chamber under the bank of tubes.

This bridge wall was found to be difficult to keep in repair and was of
no particular benefit.

[Illustration: No. 15]

No. 15. Each row of tubes was expanded at each end into a continuous
header, cast of car wheel metal. The headers had a sinuous form so that
they would lie close together and admit of a staggered position of the
tubes when assembled. While other designs of header form were tried
later, experience with Nos. 14 and 15 showed that the style here adopted
was the best for all purposes and it has not been changed materially
since. The drum in this design was supported