CONTINUOUS ELECTRIC CALORIMETRY. 
I 01) 
of the tubs and the difference of temperature between the inside and outside of the 
glass. In heating the apparatus during exhaustion, there is some risk of breaking the 
flow r -tube, as it heats and cools very slowly in a good vacuum. The difference of 
temperature between the thick flow-tube and the jacket in some cases caused the tube 
to bend so as nearly to touch the sides of the jacket. This effect would be reduced 
by making the tube thinner. 
It does not appear that very much could be gained by taking excessive precautions 
to improve the vacuum. With a less perfect vacuum there would probably be less 
variation of the heat-loss, due to the evolution of minute traces of gas during exposure 
to a high temperature. The greater certainty of the correction in that case might 
compensate for its larger magnitude. The first three calorimeters, one for mercury 
and two for water, were exhausted in the laboratory on a five-fall Sprengel pump 
which I had set up some time previously for experiments on radiation and on X-ray 
tubes. This pump gave a very perfect vacuum, but the tubes were merely heated by 
hand with a bunsen burner during the process. They showed, however, approximately 
the same rate of heat-loss as the best of the calorimeters which were subsequently 
exhausted by Messrs. Eimer and Amend in an asbestos oven. The vacuum in one of 
the calorimeters exhausted in the laboratory was also tested by means of a powerful 
oscillating discharge from a battery of Leyden jars passed through a coil surrounding 
the tube. This failed to induce a ring discharge inside the vacuum-jacket, although 
the same test w T ould produce a brilliant discharge in the majority of commercial 
vacuum vessels for liquid air. 
Probably the most effective method of reducing the heat-loss would be to silver the 
inside of the vacuum-jacket. I have tried this method in my experiments on the 
specific heat of steam by the electrical method with a vacuum-jacket calorimeter, and 
have found it very advantageous, owing to the low radiative power of the silver. The 
vacuum-jacket in this case requires very careful evacuating on account of the difficulty 
of drying the silver film. 
With these improvements, the heat-loss could probably be reduced to about one- 
quarter of its value in the existing apparatus, and the uncertainty of the correction 
would be greatly diminished, though perhaps not quite in the same proportion. In 
any case the results obtained with a calorimeter 1 laving a very different value of the 
heat-loss could not fail to be a valuable confirmation of the previous work. 
(29.) Effect of Variation of Viscosity. 
The rise of temperature of the water due to friction in its passage through the 
tube, can be easily estimated from the observed difference of head at the inflow 
and outflow. 1 made this observation for each of the conductors mentioned on p. 117, 
with a flow of half a gramme per second in a tube very slightly less than 2 millims. 
diameter and 50 centims. long. The difference of head was found to vary from 
