414 Dr. C. Shearer. Heat Production and Oxidation 



sensitivity of the galvanometer was such that, at 3 - 5 metres distance, every 

 millimetre of the galvanometer scale represented 0-00139° C. The leads from 

 the galvanometer and thermocouples were brought to a specially constructed 

 dial box furnished with two keys, by which three or four thermocouples could 

 be thrown into circuit with the galvanometer, and also small resistances 

 introduced in any of these circuits as desired. All leads and terminals, 

 including those of the galvanometer, were made of copper throughout, thus 

 avoiding any possible thermo-electric effects. 



The vacuum flasks were the ordinary narrow-necked silvered Dewar flasks, 

 made as " refills " for commercial thermos bottles. They were used in two- 

 sizes, having a capacity of 400 c.c. and 800 c.c. respectively. The larger size 

 have a coefficient of heat loss half that of the smaller, and are therefore more 

 accurate to work with where sufficient experimental material can be obtained. 

 The selection of the flasks was carried out in the following manner : — Some 

 40 to 50 flasks were obtained for rough testing. These were all filled with 

 the same quantity of water at 60° C. They were then closed with plugs of 

 cotton wool, and put aside in a corner of the room free from draughts, and. 

 allowed to warm up for an hour. Their temperature was then taken with a 

 Beckmann thermometer, after which they were allowed to stand for 

 24 hours, when their temperature was again taken with the Beckmann 

 thermometer. It was usual to find four or five flasks out of the lot that 

 had very similar rates of temperature fall, and these were selected for 

 further calibration. Their coefficients of heat loss were then carefully 

 worked out, under conditions as similar as possible to those obtaining in 

 experiments by the use of the formula T — T /A — T e— kt, a mean of ten or 

 twelve determinations being taken. The final calibration was carried out 

 under actual conditions of an experiment, with air bubbling through the 

 flask contents, and air tubes and thermocouple junctions in position, and the 

 flask itself sunk down in the water of the thermostat. One flask was given 

 a slightly higher temperature than the other, which was the exact tem- 

 perature of the bath water ; as the temperature of the flask under calibration 

 fell slowly to that of the control flask, a series of readings were taken with 

 the thermocouple and galvanometer ; these, on being plotted out, gave a 

 curve from which the value of k could be directly taken. The value given 

 by this last method was the one actually employed for experiments. As a 

 matter of fact, both methods gave very similar values for k in nearly all 

 instances. Prof. Hill was kind enough to place at my service two flasks of 

 400 c.c. capacity, which were remarkable for having almost the same value 

 of k. They could be used differentially by placing the same amount of 

 fluid in each. They have been extensively used in the present experiments.. 



