White — Melting Point Methods at High Temperatures. 483 



read as galvanometer deflections.* At the end of a determina- 

 tion the adjustment of the potentiometer storage battery is 

 verified by means of the standard cell, and the effect of elec- 

 trolytic leakage, if any, in the furnace is tested by breaking, or 

 better, suddenly reversing, the heating current. 



Table I gives a typical notebook page for a melting-point 

 determination. All but the italicized portions belong to the 

 original record, made during the observations. Here the 

 furnace was first regulated for constant rate by the control 

 element alone, and then when the temperature difference cor- 

 responding to that rate had been found, it alone was used for 

 the rest of the observations. 



II. Melting points of viscous substances. — If the melting 

 shows much hysteresis (preceding paper, page 456), the thermal 

 method fails entirely. This difficulty, however, contains its 

 own remedy, for the viscous substance can at any time be 

 removed from the furnace near the expected melting point and 

 chilledf so as to fix the exact stage of melting then reached. 

 The portions already melted solidify to glass which is perma- 

 nent at room temperature. (With extremely viscous sub- 

 stances, even the chilling is not necessary, as recrystallization 

 takes place very slowly at any temperature.) This gives a sort 

 of "successive approximation" method of determining the 

 melting point. It excels the old methods of watching the sub- 

 stances in the furnace for signs of melting, in two respects ; (1) 

 it is better thermally, since the furnace can be larger and more 

 completely closed, hence temperatures are more easily kept 

 both constant and uniform ; (2) it is better optically, since the 

 substance is examined at room temperature. 



III. Quasi-calorimetric determinations. — It has been pointed 

 out in the previous paper that a melting-point determina- 

 tion involves some knowledge of heat supply as well as of 

 temperature. It is thus essentially a calorimetric determina- 

 tion and may easily be so conducted as to bear a calorimetric 

 interpretation, of greater or less accuracy. This may be a 

 determination of the entire melting curve, discussed in the 

 previous paper, or simply of the latent heat of a single change 

 of state. Hiittner and TammannJ have given examples of the 

 latter, using a falling temperature, i.e., freezing instead of 

 melting curves. They take the heat loss as simply propor- 

 tional to the time, and propose the method as a rapid but only 

 an approximate one. Plato§ finds, by using a somewhat differ- 

 ent procedure, that the heat loss may be taken proportional to 



* Compare Potentiometer Installation, etc., Phys. Rev., xxv, 339, 1907. 

 f A. L. Day and E. T. Allen, this Journal (4), xix, 120, 1905. 

 \ Zeitschr. f anorg. Chem., xliii, 215, 1905. 

 gZeitschr. f. phys. Chem., lv, 721, 1906. 



