404 BABCOCK. 



the heating current is cut off before these temperature differences 

 are equahzed, which makes the equahzed temperature at the time 

 the switch is opened, difficult to determine. 



Some time was de\oted to an experimental study of the law of 

 cooling of this calorimeter for temperature differences of 4° or 5° 

 between the calorimeter and the bath, and these experiments showed, 

 what is well known, that Newton's Law of Cooling is not exactly 

 obeyed. The actual law was complicated and no simple expression 

 was derived for it. The error made by the assumption of Newton's 

 Law, is however, much smaller than the error made by the uncer- 

 tainty in the final equalized temperature. Its use was therefore con- 

 sidered to be legitimate under these conditions. The assumption of 

 linearity of the initial and final lines is theoretically unjustifiable, but 

 practically, no error is made by using them as such. A very slightly 

 higher value of Ro and therefore a lower value of f„ would result from 

 taking account of the curvature of the final line. 



The values of k in Newton's Law of Cooling, varied from one meas- 

 urement to another, which is accounted for by the fact that the 

 environment of the calorimeter was not absolutely reproducible. In 

 a vacuum-jacketed calorimeter, a major part of the heat transfer to 

 the surroundings takes place through the cover; and in this apparatus, 

 the stirring shaft and leads served to conduct a considerable c^uantity 

 of heat from the cover directly to the room, which in most measure- 

 ments was considerably below the temperature of the calorimeter, 

 and furthermore, was not controllable as was the oil bath. The most 

 accurate method, therefore, of dealing with the heat transfer, seemed 

 to be : to compute k for each measurement from the differences of slope 

 and temperature of the initial and final lines. 



The constancy of the stirring energy input is hardly open to ques- 

 tion, because during all the measurements, readings were taken at one 

 minute inter\'als of the voltage drop across the armature of the stirring 

 motor, and in no case did this voltage change enough to give a meas- 

 ural)le change of slope in the temperature curve. This effect was 

 separately investigated by changing the motor voltage and observing 

 the effect on the temperature-time curve. A change of 10% in the 

 stirring motor ^'oltage was found to be necessary before the slope 

 of the temperature curve was changed by a measurable amount. 

 During the measurements the voltage variations were very much less 

 than this. 



As the measurements were being made and computed, it was realized 

 that the accuracv attainable bv this method was less than had been 



