ent could
easily occur. The pipe found best adapted to the purpose was the
so-called standard one-eighth inch brass pipe with an actual internal
diameter of 7 millimeters. The opening of a valve allowed cold water to
flow through this pipe and the considerable mass of water passing
through produced a very noticeable cooling effect. In the attempt to
minimize the cooling effect of the mass of water remaining in the pipe,
provision was made to allow water to drain out of this pipe a few
moments after the valve was closed by a system of check-valves. In
building the new apparatus, use was made of the compressed-air service
in the laboratory to remove the large mass of cold water in the pipe. As
soon as the water-valve was closed and the air-cock opened, the
compressed air blew all of the water out of the tube.

[Illustration: FIG. 13.--Detail of drop-eight feed-valve and arrangement
of outside cooling circuit. The water enters at A, and the flow is
regulated by the needle-valve at left-hand side. Rate of flow can be
seen at end of exit tube just above the union. The water flows out at C
and compressed air is admitted at B, regulated by the pet-cock.]

The best results have been obtained, however, with an entirely new
principle, namely, a few drops of water are continually allowed to pass
into the pipe, together with a steady stream of compressed air. This
cold water is forcibly blown through the pipe, thus cooling to an amount
regulated by the amount of water admitted. Furthermore, the relatively
dry air evaporates some of the water, thereby producing a somewhat
greater cooling effect. By adjusting the flow of water through the pipe
a continuous cooling effect of mild degree may be obtained. While
formerly the air in the space next the zinc wall was either cooled or
heated alternately by opening the water-valve or by passing a current
through the heating coil, at present it is found much more advantageous
to allow a slow flow of air and water through the pipes continuously,
thus having the air-space normally somewhat cooler than is desired. The
effect of this cooling, therefore, is then counterbalanced by passing an
electric current of varying strength through the heating wire. By this
manipulation it is unnecessary that the observer manipulate more than
one instrument, namely, the rheostat, while formerly he had to
manipulate valves, compressed-air cocks, and rheostat. The arrangement
for providing for the amount of comp