CONTINUOUS ELECTRIC CALORIMETRY. 
103 
water was sucked in through the central tube E, 1 inch in diameter, and ejected 
through the annular space surrounding the bulb of the regulator by the tube F. 
The whole was mounted on a large gas burner, and shielded with an asbestos screen. 
The object of placing the regulator bulb inside the heater, and in close proximity to 
the pump, was to secure quickness of action in response to any change in the gas- 
pressure. The violent stirring, and the comparatively small capacity of the heater in 
proportion to the bulb favoured this result. It was most important that the regulator 
should be very sensitive, and that there should be no forced oscillations of temperature, 
because the jacket-temperature determined that of the inflowing water-supply to the 
calorimeter, and did not merely affect the external heat-loss. Any temperature 
oscillation would j3roduce a serious effect on the results, especially on the smaller 
flows, in which the total mass of water passing, about 250 grammes in 15 minutes, 
was not very large compared with the effective thermal capacity of the calorimeter, 
which was about 50 grammes. 
When the regulator was made sufficiently sensitive to cut off the gas for a very 
small change of temperature, I found it necessary at the higher points of the range, 
where a large supply of gas was required, to adjust the by-pass so that, if the 
regulator were cut off, the temperature would very nearly reach the required point. 
Forced oscillations could only be avoided if the regulator controlled a very small 
fraction of the heat supply, acting merely as a fine adjustment on the temperature of 
the system. Under these conditions, however, when the total gas-supply was large, 
any small accidental change in the gas-pressure might exceed the limits of control of 
the regulator. A rapid change in the quality of the gas produced similar effects. It 
was therefore absolutely necessary at the higher temperatures to keep the gas- 
pressure very constant. The best forms of gas-governor were tried, but did not 
prove sufficiently delicate. I therefore fitted up a large copper gas-holder, delicately 
suspended and counterpoised by means of a steel tape passing over a wheel with ball 
bearings so as to move with very little friction. This arrangement proved to be 
capable of regulating the pressure to within a tenth of a millimetre of water. 
The large capacity of the gas-holder tended also to minimize the effect of sudden 
small variations of quality of the gas, such as might be produced by air in the 
pipes, &c. 
The action of this constant-pressure gas-supply was so perfect that for many 
purposes no other temperature-regulator was required, and as a matter of fact none 
was used in many of the preliminary experiments. Since variations of gas-pressure 
were practically eliminated, it was found to be unnecessary to have the regulator bulb 
inside the heater, and the fine adjustment-regulator was placed in the tank C, 
(Barnes, figs. 14 and 15, pp. 211, 213). This regulator was employed to operate an 
electric heating arrangement, as described by Dr. Barnes in Section 4, in which slow 
period oscillations were prevented by the device of the reciprocating contact suggested 
by Gouy (‘Journ. de Phys.,’ 1897, p. 479). I found it necessary to introduce a few 
