400 PROCEEDINGS OF THE AMERICAN ACADEMY. 



most of the occluded impurity is set free at bright redness, and the first 

 two of them agree quantitatively with the results of Dr. Scott. 



Although it is thus evident that the English and American results are 

 in no way inconsistent, I decided to repeat a part of the work, in order 

 to obtain a more certain estimate of the various temperatures correspond- 

 ing to the varying conditions of the cupric oxide. In my previous paper 

 the higher degrees had been merely guessed, for they were not directly 

 concerned in the point then under consideration. 



The usual modes of determining high temperatures — the calorimeter, 

 the air-thermometer, the bolometer, the meldometer, the platinum-irid- 

 ium thermopile, etc. — are somewhat troublesome to use for a brief 

 series of experiments of the present kind, so that it seemed best to 

 determine the several temperatures approximately by finding for each 

 temperature two substances, whose melting points were respectively 

 above and below the point in question. For instance, an intensity of 

 heat which would melt pure argentic bromide but not pure argentic 

 chloride must be between 427° and 451°, a grade of accuracy more than 

 sufficient for the present purpose. This method of determining the tem- 

 perature has the great advantage of providing an approximate self- 

 registering thermometer, occupying very small space and needing no 

 connection with the outside air. The usefulness of the method obvi- 

 ously depends upon the number of substances easily obtained in a pure 

 state, whose melting points are accurately known. For such knowledge 

 of high melting points we depend mainly upon four researches. — those 

 of Carnelley,* Le Chatelier,t Meyer, Riddle, and Lamb,$ and Ramsay 

 and Eumorfopoulos.§ Unfortunately the results of these four investi- 

 gations do not always agree, their disagreement affording evidence of 

 the great difficulty of measuring accurately high temperatures. At first 

 sight the fact of the existence of such discrepancy might deter one from 

 adontins this standard of reference, but further consideration leads to 

 exactly the opposite conclusion. The melting points of pure salts must 

 be an unchangeable standard, and the uncertainty of 'bur knowledge 

 regarding them must be largely due to experimental difficulties in the 

 thermometric manipulation. These difficulties, if so serious in researches 

 where everything was favorable to accuracy, must be largely augmented 



* Carnelley, Journ. Chem. Soc, XXIX. 489, XXXIII. 273. (1876 and 1878.) 

 t Le Chatelier, Bull. Soc. Chem., XLVII. 300. (1887.) 



I v. Meyer, Riddle, and Lamb, Bericlite der Deutsch. cliem. Gesell., XXVII. 

 3129. (1895.) 



§ Ramsay and Eumorfopoulos, Phil. Ma.?., (5.), XLL 300. (1896) 



