l,
so placed in the copper can that the water has a maximum circulation
about the heater. The whole device is thoroughly insulated with
hair-felt. By connecting the electric reheater with the rheostat on the
observer's table, control of the quantity of electricity passing through
the coil is readily obtained, and hence it is possible to regulate the
temperature of the ingoing water to within a few hundredths of a degree.

The control of the amount of heat brought away from the chamber is made
either by (1) increasing the rate of flow or (2) by varying the
temperature of the ingoing water. Usually only the second method is
necessary. In the older form of apparatus a third method was possible,
namely, by varying the area of the absorbing surface of the cooling
system inside of the chamber. This last method of regulation, which was
used almost exclusively in earlier experiments, called for an elaborate
system of shields which could be raised or lowered at will by the
operator outside, thus involving an opening through the chamber which
was somewhat difficult to make air-tight and also considerably
complicating the mechanism inside the chamber. The more recent method of
control by regulating the temperature of the ingoing water by the
electric reheater has been much refined and has given excellent service.

_Insulation of water-pipes through the wall._--To insulate the
water-pipes as they pass through the metal walls of the calorimeter and
to prevent any cooling effect not measured by the thermometers presented
great difficulties. The device employed in the Middletown chamber was
relatively simple, but very inaccessible and a source of more or less
trouble, namely, a large-sized glass tube embedded in a large round
wooden plug with the annular space between the glass and wood filled
with wax. An attempt was made in the new calorimeters to secure air
insulation by using a large-sized glass tube, some 15 millimeters
internal diameter, and passing it through a large rubber stopper,
fitting into a brass ferule soldered between the zinc and copper walls.
(See N, fig. 25.) So far as insulation was concerned, this arrangement
was very satisfactory, but unfortunately the glass tubes break readily
and difficulty was constantly experienced. An attempt was next made to
substitute hard-rubber tubing for the glass tube, but this did not prove
to be an efficient insulator. More recently we have used with perfect
success a special form o