552 REPORT — 1897. 



with those obtained by others. At these great dilutions the surface tension is a 

 linear function of the concentration in every case studied. 



2. On a New Method of Deterjiiining the Sjjecijic Heat of a Liquid in 

 terms of the International Electrical Units. By H. L. Callendar, 

 M.A., F.B.S., Professor of Physics, and H. T. Barnes, M.A.Sc, 

 Denwnstrator of Physics, of McGill University, Montreal. 



In view of the probable adoption of the Joule or Watt-second as the absolute 

 unit of heat, it becomes of special interest at the present time to make direct 

 determinations of the natural thermal units in terms of the electrical standards 

 now universally adopted. 



In recent years the specific heat of water has been very carefully determined in 

 this manner by Griffiths, and also by Schuster and Gannon. These observers 

 employed the usual calorimetric method, in which a mass of water is heated 

 through a carefully observed range of temperature by means of a measured 

 quantity of electrical energy. Although their methods differed widely in points 

 of detail, their results agreed to within one part in a thousand with each other. 

 But, as Schuster points out, the result so obtained by the electrical method for the 

 specific heat of water differs by one part in 400 from the result obtained by direct 

 mechanical measurements of Joule, Kowland, and Miculescu. 



Whatever the cause of this discrepancy, it seemed desirable to repeat the 

 electrical comparison by an entirely difl'erent method, to avoid any possible source 

 of constant error which may have remained unsuspected in the calorimetric method 

 as usually practised. 



The method which we have adopted consists in passing an electric current 

 through a fine tube, through which a steady current of liquid is flowing. The 

 electrical measurements required are the current and the difference of potential 

 between the ends of the tube. The thermal measurements are the steady difference 

 of temperature and the quantity of liquid flowing in a given time. 



The electrical measurements are all made on one potentiometer, preferably a 

 Thomson- Varley si. de-box, and present no difficulty, as it is easy to keep the 

 current steady to one part in a thousand for an hour or more, and there is no 

 change in the resistance of the circuit. 



The difference of temperature between the inflowing and outflowing liquid, 

 which is also very nearly constant throughout the duration of the experiment, is 

 measured by means of a differential platinum thermometer. The instruments used 

 for this purpose, consisting of a compensated slide-wire resistance box and pair of 

 thermometers, are the same as were exhibited by Prof. Callendar at the con- 

 versazione of the Royal Society in 1893, on which occasion the instruments were 

 used for demonstrating the lowering of the melting-point of ice under one 

 atmosphere of pressure. Readings can be taken to the ten-thousandth part of a 

 degree on a rise of temperature of ten degrees. 



The current of liquid is kept steady by means of an automatic electromagnetic 

 device, and the quantity flowing in a given time, the interval being also 

 automatically recorded on an electric chronograph, is determined by weighing. 



It will be observed that in this method, as compared with that usually employed, 

 since the temperature distribution is exceedingly steady, it is not necessary to 

 determine the thermal capacity of the calorimetric tube with any degi'ee of accuracy. 

 The rate of external loss of heat is also much more steady and more easily 

 determined, and there is no question of lag of the thermometers. 



The external loss of heat, which is generally the largest and the most uncert'iin 

 correction in all calorimetric experiments, can, in the present instance, be made 

 extremely small and regular by the expedient of enclosing the calorimetric tube, &c., 

 in a glass jacket, which is exhausted as perfectly as possible and then hermetically 

 sealed, so that the vacuum cannot sufler further change. The loss can also be 

 measured and eliminated in a very simple manner. If observations are taken with 



