TESTS OF ACCURACY OF, HEAT-MEASURING APPARATUS. 171 



part in 1,000. A sufficient number of cells of storage battery are 

 employed to give a strength of current through the coil of about i 

 ampere with a voltage of about 1 20. 



ELECTRICAL UNIT USED. 



In the fall of 1904, in a discussion of the accuracy of the bomb calo- 

 rimeter l used in connection with these experiments, it was pointed out 

 by Dr. I,. J. Henderson, of Harvard University, that the heat of combus- 

 tion of standard materials such as naphthalene, benzoic acid, and cane 

 sugar were noticeably different when determined by the bomb calorimeter 

 used at Wesleyan University and when determined by the bomb calo- 

 rimeter used by Fischer and Wrede." The calorimeter used by these 

 writers was standardized by Jaeger and von Steinwehr 3 by an electrical 

 method in which the factor 0.2394 was used to convert watt-seconds to 

 calories. 



This matter was referred to Dr. E. B. Rosa, formerly professor of 

 physics in Wesleyan University and at present physicist of the National 

 Bureau of Standards. The following statements are essentially those 

 furnished us by Dr. Rosa. 



The values found for the mechanical equivalent of heat by the elec- 

 trical method differ appreciably from those obtained by the mechanical 

 method. There is reason for believing, however, that the values of the 

 international volt and ampere are about o. i per cent too large. This 

 is a subject the Bureau of Standards and others are now investigating, 

 but absolute measurements for determining independently the volt 

 and ampere are difficult to make and the question is not yet settled. 

 Assuming this error in the electrical units, the values of J deter- 

 mined electrically agree very well with Rowland's value determined 

 mechanically, and this is the best value yet obtained by the mechanical 

 method. 



The most probable value for J (assuming the correction of o. i per 

 cent in the electrical units) is 



J= 4.181 x io 7 ergs, at 20. 



Allowing for the variation in the specific heat of water, the heat 

 required to raise the temperature of a gram of water i at 10 would 

 require 4.181 x io 7 x 1.0030 = 4.1935 X io 7 ergs. 



1 For a description of the form of bomb calorimeter here used see Jour. Am. Chem. 

 Soc. (1903), 25, p. 659. 



'- 1 Sitzungsber. K. Akad. Wiss. (1904), pp. 687-715. 



3 Verhandlungen ber. deut. phys. Gesell. (1903), 5, 2, pp. 50-59. 



