CONTINUOUS ELECTRIC CALORIMETRY. 
107 
the smaller flows. As a consequence of this irregularity of the small flows. I do not 
think these results can be fairly treated by the somewhat artificial method adopted 
by Dr. Barnes. The extrapolation of the heat-loss per degree for zero rise of 
temperature is too uncertain in the case of the small flow. The natural method of 
treatment would be to take the difference between the sum of the electrical watts 
SEC, and the sum of the heat-watts SJQ (16, and divide by the sum of the 
temperature differences Scl9, to find D for each flow. Combining the two flows in the 
usual manner, without any arbitrary assumptions, we thus obtain practically the 
same result as that deduced by Dr. Barnes. 
(27.) Design of the Water Colorimeter. 
The design of the water calorimeter presented certain points of difficulty which 
were not settled without some preliminary experiments on the conditions of flow in 
fine tubes. The greater part of these experiments were carried out in the summer of 
1896, in the Thermodynamical Laboratory of the Engineering Building of McGill 
College, with the kind permission of Professor Nicolson, who erected for this purpose 
a large supply-tank of water on an upper floor in a room at a very constant 
temperature. The results of some of these experiments were mentioned in a paper on 
the ‘ Law of Condensation of Steam,’ which was communicated to the Institution of 
Civil Engineers in September, 1896, and read in the following year. 
Since water is practically a non-conductor of electricity, it was necessary either to 
make the fine flow-tube of platinum instead of glass, or else to thread a conducting 
wire or strip of platinum through it. A fine bore metallic flow-tube would have 
presented some advantages in point of smallness of radiation loss, but in the end 1 
decided to use a glass flow-tube and a central conductor, chiefly on account of the 
importance of securing the greatest possible constancy and perfection in the vacuum- 
jacket. This could be most easily and certainly attained by making the vacuum-jacket 
entirely of glass. 
The glass-work of the water calorimeter differed from that of the mercury calori¬ 
meter (fig. 6) only in having a straight flow-tube 50 centims. long and 2 millims. 
bore, instead of a spiral flow-tube 100 centims. long and 1 millim. bore. It was 
necessary to make the flow-tube straight on account of the difficulty of threading the 
conductor through it as well as the connecting wires to which it was attached. This 
operation would have been facilitated by using a larger tube, but, apart from this 
necessity, it was desirable to have the flow-tube as fine as possible to secure uniformity 
of temperature of cross-section and other advantages. From one point of view, it was 
desirable to make it as short as possible, in order to minimise the heat-loss, which 
depended chiefly on the surface of the flow-tube; but on the other hand it was 
necessary to have sufficient length to eliminate leakage of the current through the 
water, and to secure a suitable resistance for the conductor, and sufficient surface to 
p 2 
