252 PROCEEDINGS OF THE AMERICAN ACADEMY. 



surface temperature of the calorimeter. Secondly, that the tempera- 

 ture distribution throughout the entire contents of the calorimeter 

 is uniform both when r^ and 1\ are taken, so that these are really 

 rates of exchange with surroundings, and not resultants in which the 

 desired rate is more or less modified by redistribution within the 

 calorimeter. Thirdly, that the heat capacity of calorimeter and con- 

 tents is the same both when rj and 1\ are taken. 



Proper location of thermometer, efficient stirring, and the preven- 

 tion of direct contact with the calorimeter of any hot or cold object 

 within it, readily insure the fulfilment of the first with sufficient 

 closeness. 



Wherever the tem25erature of the calorimeter is varying, uniform 

 distribution of temperature throughout its contents is obviously 

 impossible if these are heterogeneous, especially where there is a 

 large metallic mass surrounded by the water, as the bomb in the 

 Berthelot or Mahler combustion apparatus. Such objects will neces- 

 sarily lag more or less behind the water as the temperature of the 

 latter falls or rises by external exchange. The effect of this is to 

 falsify the apparent rate r^ or rg, and still further to vitiate the 

 results through the usual assumption that all parts are at the same 

 temperature. A form of cooling correction, taking this into account 

 and applying to a certain class of cases, has been described by Pro- 

 fessor Rowland, but the only method which could entirely eliminate 

 the error would be to arrange the calorimetric process so that the 

 rate of exchange should be sensil)ly zero or constant during the 

 entire measurement. This means that the jacket temperature must 

 be so controlled as to be always at the temperature of the calorimeter, 

 or at a constant difference from it. "While this at first sight appears 

 wholly impracticable, and is so for most cases, I am disposed to think 

 that it might be feasible and helpful in certain investigations, such as 

 the stiady of the specific heat of water, and the mechanical equivalent 

 of heat, using electrical methods of heating, combined with thermo- 

 electric means of detecting and controlling the temperature difference 

 between the calorimeter and jacket. 



The " Modified Method " described above deals with this error in 

 another way. It makes the final rate r^ as nearly as possible zero, 

 and hence secures the greatest possible constancy of temperature 

 during the portion C D Eoi the curve when the rate is about r,. The 

 advantages gained are as follows. The apparent rate r„ is sensibly 

 the true rate, since it will in general be much less than the rate of 

 redistribution of heat amongst the contents of the calorimeter. This 



