342 Day and Sosman — Melting Points of Minerals. 



the old way, by extrapolating the curve of temperature and 

 thermal electromotive force. These results now require to be 

 corrected by the amount of the difference between the old tem- 

 perature scale and the new, which makes it necessary to recal- 

 culate the existing temperature data of this laboratory in terms 

 of the new scale. This paper will present a summary of the 

 values resulting from this recalculation. 



1. 3felti/ig and Inversion Points. 



The " melting point" of a pure substance may be defined as 

 tbe temperature at which the crystalline and liquid substance 

 can remain side by side in equilibrium ; an "inversion point," 

 as a temperature at which two different crystalline forms of 

 the substance can remain side by side in equilibrium. Only 

 the addition or withdrawal of a quantity of heat will cause the 

 disappearance of one of the two forms in contact. Both melt- 

 ing and inversion are therefore characterized by two concur- 

 rent phenomena, tbe appearance or disappearance of a particular 

 crystal structure and the appearance or disappearance of a 

 quantity of heat. 



2. Melting Intervals. 



The above definition applies to pure compounds. If the 

 material is a mixture or a solid solution, it will bave, not a 

 melting point, but a melting interval, with (theoretically) 

 definite temperature limits. 



The so-called melting interval of slow melting compounds is 

 of entirely different character. Tammann,* and more recently 

 Dittleiyf- have criticized certain of the melting-point determi- 

 nations made in this laboratory, more particularly those of 

 anorthite and diopside, as being too high because of inade- 

 quate furnace control and superheating. This is plainly due 

 to oversight in the consideration (1) of the phenomenon of 

 melting in silicates, and (2) of the records of our measurements. 

 If a substance always undercook before crystallizing, the tem- 

 perature of crystallization will vary with the rate of cooling, 

 i. e., will be a random temperature depending upon the condi- 

 tions of experiment and not a physical constant characteristic 

 of the substance. If the same substance melts so slowly that 

 it readily becomes superheated while melting, the temperature 

 of complete liquefaction will vary with the rate of heating and 

 is also a random temperature dependent solely upon the condi- 

 tions of experiment. Pure quartz, albite and orthoclase are 

 such substances, and no doubt others will be found. The 

 temperature-time curve is not a competent method with which 

 to determine the temperature of change of state of such sub- 

 stances. 



*Zts. physikal. Chem., lxviii, 257-269, 1909. 

 fZts. anorg Chem., lxix, 273-304,1911. 



