94 



REPORTS OF INVESTIGATIONS AND PROJECTS. 



determinations, but the following are offered tentatively, subject to a redeter- 

 mination of the expansion coefficient of the bulb. The coefficient of expan- 

 sion used for the computation of these values is that of Holborn & Day 

 ("On the gas-thermometer at high temperatures," Amer. Journ. Sci., 4, 57, 

 p. 171, 1900). 



Melting-point Determinations. 



(5) An accurate calorimeter (abstract). Walter P. White. Phys. Rev., 25, p. 137, 

 1907. 



This calorimeter is used in connection with an electric resistance furnace. 

 It is therefore completely inclosed with a double-walled jacket, through which 

 water is kept circulating. As the amount of manipulation necessitates the 

 maximum simplicity in methods of measureiTsent, the temperature is deter- 

 mined by a thermoelement, which can be read on a potentiometer almost at 

 the same time with the temperature of the furnace, the differential tempera- 

 ture of the calorimeter and jacket, and the voltage and current used in 

 calibration. The temperatures can be read concordantly to about 0.001°. 

 The use of large calorimeters and small temperature intervals has been 

 recommended in order to reduce cooling corrections, but at best a small gain 

 in cooling correction, when secured in this way, requires a very large increase 

 in the sensitiveness of the thermometer, while, in so far as the cooling errors 

 come from the difficulty of determining the cooling correction or from the 

 heat generated in stirring, they are actually greater in the large calorimeter. 

 Hence the temperature interval of 20° was chosen. 



Further reasons for a small temperature interval and large calorimeter are 

 found in the closer approximation to Newton's law of cooling and in the 

 reduced effect of the metal parts of the calorimeter. With the present 

 arrangement of jacket and floating cover (to diminish evaporation), how- 

 ever, practically everything within the case is at one of two temperatures — 

 that of the calorimeter water and that of the jacket water — whose difference 

 is actually measured. Uncertain and lagging temperatures of the metal parts 

 are thus practically eliminated. The slight variation from Newton's law is 

 easily determined and applied as a correction, since the difference in tem- 

 perature between the calorimeter and its zvhole environment can be at all 

 times exactly shown. 



Experience has shown that the regularity of the cooling is at least equal to 

 the maximum accuracy obtained in measuring it. The principal errors thus 

 far observed have been clearly traceable to lack of sensitiveness in tempera- 

 ture measurement. The selection of a wide temperature interval, therefore, 

 seems completely justified. 



