8 PROCEEDINGS OP THE AMERICAN ACADEMY. 



2. THE CONSTANCY OF RESULTS OBTAINED BY ONE OF THE 



NEW METHODS. 



By Theodore W. Richards and George S. Forbes. 



The experiments whose description follows in this section show how, 

 in cases of comparative calorimetry, measurements can be conducted by 

 heating the environment with constancy greater than that attainable with 

 the same apparatus by the use of a cooling correction. It will also be 

 shown that furthur study and improvement of the method may render 

 possible the direct determination of the true adiabatic rise of temperature 

 in any calorimetric process. 



In the course of a certain investigation, to be described later in de- 

 tail, it became desirable to compare the heat capacities of mercury and of 

 liquid amalgams of zinc and cadmium. It is not necessary to describe 

 the apparatus in great detail here. A sheet-iron calorimeter containing 

 eleven hundred grams of toluol was supported in the center of a tinned 

 iron vessel, which was immersed in a large bath of water contained in a 

 galvauized-iron barrel. An annular cover protected the air space around 

 the calorimeter from disturbance. Twenty-three hundred grams of mer- 

 cury were delivered from a warm reservoir of very constant temperature 

 and distributed into a series of shallow pans ; all of these were insu- 

 lated from the walls of the calorimeter by the toluol, which was kept in 

 rapid circulation by a complicated and efficient stirrer. The warming 

 effect due to the agitation of the liquid was considerable, but very con- 

 stant. The maximum temperature was always attained within four min- 

 utes, and was read by a Beckmann thermometer graduated to hundredths; 

 an electric buzzer prevented errors due to the friction of the thread. The 

 thermometer was not calibrated, but the same part of its scale was used 

 in every experiment. After the apparatus was in running order, and the 

 details of its manipulation mastered, a series of determinations of the 

 heat capacity of mercury, in certain arbitrary units, was made. The in- 

 dications of the thermometer were carefully noted, and interpreted in the 

 usual way, applying the usual Regnault-Pfaundler method for correcting 

 the error from radiation.* 



Four successive determinations of the heat capacity of mercury in these 

 arbitrary units were 608, 613, 614, 607, average 611 ± 1. 



The probability is that this mean value represents approximately 

 the mean of an indefinite number of results by this process, but never- 



* Pfaundler, Pogg. Ann., 129, 113 (18G6). 



