W. P. White — Melting Point Determination. 461 



investigation. This temperature, O of the formula, is the 

 upper end of the melting interval, the point X of figs. 3 and 4. 

 The other characteristics of the curve, the values of the specific 

 heat, etc., are then, at most, questions of minor interest. 

 ^Nevertheless, in such cases the complications due to impurity 

 are. as important and as troublesome as anywhere. For in an 

 impure substance the melting temperature desired is merely 

 one value of a continually changing magnitude. Before it can 

 be measured it must first be located on the oblique curve. 

 But, coming as it does at the top of the melting interval, and 

 therefore at the end of the melting, it falls where the tempera- 

 ture changes are most irregular, disturbed and uncertain of 

 interpretation. The resulting difficulties occupy the rest of 

 this paper. Meanwhile, a glance at fig. 1 shows the difficulty 

 which may arise in locating the single melting point. 



III. Varying rate of heat supply. — In determining the 

 single melting point of an ideally pure substance, the character 

 of the heat supply is a matter of indifference so long as it does 

 not approach zero or infinity. For the exact determination of 

 the melting curve of an impure substance, the heat supply 

 must be well known, as has just been seen. For determining 

 the single melting point of a rather impure substance an inter- 

 mediate condition obtains. The heat supply need not be 

 known and need not be constant so long as its variations are 

 regular. That is, the " break" will show on almost any smooth 

 curve. The determination of a melting point with a varying 

 heat supply, however, often gives rise to a secondary phenom- 

 enon so striking and so apt to be misleading as to deserve 

 mention here. This occurs when the furnace rate is kept 

 nearly constant, as it usually is. As soon as the charge begins 

 to melt, its temperature rise is checked, so that the continued 

 advance of the furnace widens the gap between them ; thus 

 the constant rate of the furnace necessarily involves a very 

 variable rate of heat supply to the charge. The result is to 

 hurry up the latter end of the melting, apparently increasing 

 its obliquity. The same happens at the end of a freezing 

 curve, if such is taken. But the end of a melting curve 

 is the top, of a freezing curve the bottom ; hence the two 

 kinds of curves are distorted out of all resemblance to 

 each other, the melting curve appearing more, the freezing 

 curve less, oblique in the upper part than it really is. The 

 observer who, attempting to secure constant conditions, 

 approaches his melting or freezing determinations at the same 

 rate, and then maintains this rate constant in the furnace, is 

 likely to go through the critical part of his melting curve five 

 to ten times as fast as he realizes. If, however, the initial rate 

 is then adjusted by trial so as to give satisfactory results at 



Am. Jour. Sci.— Fourth Series, Vol. XXVIII, No. 167.— November, 1909. 

 31 



