460 W. P. White — Melting Point Determination. 



temperatures contains the possibility of tracing exactly the 

 progress of a melting from beginning to end. 



2. Variation of the obliquity with temperature. — In accord- 



'O20 2 

 ance with the familiar formula, A = — - — , the melting point 



depression, A, is proportional to the square of the absolute 

 temperature and therefore (see equation (3) for instance) the 

 melting curve obliquity must vary in the same way. Hence 

 the large effect produced at high temperatures by amounts of 

 impurity which would be quite negligible lower down. Thus, 

 to take a roughly approximate illustration, *1 per cent of an 

 impurity of molecular weight 80, say CaF 2 , dissociated to 

 double the number of molecules, would by the formula lower 

 the normal melting point (1392°) of diopside 1'3°, and would 

 cause an apparent doubling of the specific heat 20° below that, 

 while if the equation (1) held strictly, the beginning of the 

 melting would be easily perceptible 50° lower still. A similar 

 impurity would lower the melting point of ice only -05°, and 

 the doubling of the apparent specific heat would occur within 

 3° of the melting temperature. 



These facts have a direct bearing on the question of the 

 value of melting point determinations in natural minerals. 

 While the impurities of natural minerals in rare cases run within 

 a few parts per thousand, they are usually to be reckoned 

 in per cents. But one per cent of impurity may be expected to 

 lower the melting point from 3° to 10° and extend the dis- 

 tinctly perceptible melting interval over some 100°. And 

 silicates with 3 per cent of impurity usually show in most 

 decided fashion the characteristic behavior of two-component 

 systems. 



3. Experimental determination of the quantity of heat 

 supplied to the charge. — A quantitative knowledge of the heat 

 supply is evidently essential to a correct melting curve. The 

 use of time as a measure of it, which forms the basis of nearly 

 all the common melting point methods, is recognized as merely 

 a rough approximation,* and would not have answered at all in 

 most of the cases here treated. Hence a more accurate method 

 was employed. It rests on the use of the control element, C, 

 fig. 2, which allows a determination of the temperature differ- 

 ence of furnace and charge, on which, and not on the 

 temperature of either alone, the heat flow directly depends. 

 This method is more fully treated in the second paper, p. 485. 



4. Single melting points. — In most determinations on melt- 

 ing substances, the melting temperature is the sole object of 



* G. K. Burgess, Methods of Obtaining Cooling Curves, Bull. Bur. Stand- 

 ards, v, 223, 1908 ; W. Eosenhain, Observations on Eecalescence Curves. 

 Proc. Phys. Soc. London, xxi, 183, 1908. 



