Determinimj the Specific He<it of a Liquid. 99 



crossed the specified division, the flask was dropped into the 

 calorimeter and the rise of temperature noted on the large 

 thermometer. 



Theoretically this is a very simple process and should give 

 very good results, but the difficulties were found to be practically 

 insurmountable. The first difficulty met with was the accurate 

 calibration of the galvanometer scale in terms of degrees centi- 

 grade. Thermometer bulb and junction were put into as close a 

 contact as possible, and the temperatures of the vessels in which 

 they were immersed were carefully regulated, and yet no 

 constancy of results could be obtained. The resulting conviction, 

 that thermo-electric junctions are not reliable to an accuracy of 

 one per cent, when used to measure differences of temperature, I 

 find to be shared by a number of those who, like myself, have 

 attempted to use them in this way ; but the inaccuracies of this 

 method, as revealed by comparing the results of a number of 

 experiments, were far greater than could be accounted for merely 

 by the variations of the thermal junctions. The explanation 

 seems to lie in a fact which I frequently observed ; theoretic- 

 ally the thermometer in the calorimeter should show its 

 maximum indication at the instant that the spot of light on 

 the galvanometer-scale returns to its zero position, for this is 

 supposed to indicate that the two junctions are at the same 

 temperature, and that state once reached there should be no 

 further rise. In practice this was never the case ; at times the 

 thermometer reached its maximum before and at others after the 

 galvanometer had returned to zero. This was probably due to 

 one or both of two causes, viz., either a lagging or retardation in 

 the action of the junctions or insufficient agitation and mixing 

 among themselves of the contents of both the calorimeter and 

 the small silver flask. In either case the temperature indicated 

 by the galvanometer is not the mean temperature of the liquid 

 surrounding the junction, and if this is the case when both 

 junctions are inside a well-protected calorimeter in which the 

 temperature-differences are vei'y small, the uncertainty as to the 

 temperature of the contents of the small silver flask at the 

 moment of immersion must be very considerable, as at that 

 instant the outside of the flask is necessarily exposed to the air. 

 Keeping the flask in agitation only confirms these doubts, as it 



