plastered surface on the dome-shaped bottom provided the necessary
imperviousness, it would seem that plastered walls would have proved
satisfactory.

Apparently, the sheet-metal tank is intended to exclude the possibility
of exterior leakage, but it occurs to the writer that it will fail to be
efficient in this particular, because, under pressure, the water will
force itself under the steel tank and the dome thrust rings and out to
the exterior of the tower just below the tank, thus showing that
insurance against leakage is actually provided by the plastered interior
surfaces and not by the sheet-metal tank, and, for this reason, ordinary
deformed rod reinforcement, in the writer's opinion, would have proved
cheaper and better, and more in line with other parts of the
reinforcement.

Mr. Kempkey states:

"Before filling, the inside of the tank was given a plaster coat,
consisting of 1 part cement to 1-3/4 parts of fine sand. This
proved to be insufficient to prevent leakage, the water seeping
through the dome and appearing on the outside of the structure
along the line of the bottom of the rings. Three more coats were
then applied over the entire tank, and two additional ones over the
dome and about 8 ft. up on the sides, and, except for one or two
small spots which show just a sign of moisture, the tank is
perfectly tight."

This substantiates the writer's contention that water-tightness was
actually obtained by a liberal use of cement plaster, which would also
have been true had the reinforcement been rods.