454 W. P. White — Melting Point Determination. 



melting under uniform heat supply, with MN as the interval 

 of constant temperature, the actual result with a substance 

 melting at high temperature more nearly resembles the oblique 

 curves B and C. where the temperature intervals RS may be 

 as much as 60°. 



When the melting curve is oblique, there is a much greater 

 opportunity for both accidental and systematic errors. In our 

 own work these have not been serious, considering the high 

 temperatures concerned, and were for a long time far less 

 uncertain than the extrapolated temperature scale itself,, The 

 greatest discrepancy in a set of determinations of the same 

 point has seldom reached 3°.* 



But other observers had found much larger and sometimes 

 confusing irregularities, and had questioned the value of the 

 thermal (Frankenheim) method of determining melting points ; 

 moreover, the causes for the obliquity of these melting curves 

 were in themselves of interest ; and finally, in investigations on 

 some pyroxenes, a problem was encountered which called for 

 much more accurate comparative measurements than we had 

 been getting. A general investigation was therefore under- 

 taken of the thermal method of determining melting points, 

 with the twofold object of learning more about the properties 

 of matter in the vicinity of the melting temperature and of 

 improving our own technic. As a result, the agreement of 

 our silicate determinations has been increased about five-fold, f 

 along with an actual gain in case of experimental manipulation. 

 An insight into the relations involved has also been obtained 

 which has cleared up several questions once very puzzling, and 

 has pointed the way to a further increase in accuracy whenever 

 this seems necessary. 



A number of substances, organic and inorganic, melting at 

 temperatures from to 1400°, were examined under various 

 conditions. Most of these experiments were devised to test 

 hypotheses and need not now be described in detail. The con- 

 clusions to which they led will perhaps also gain if presented 

 in a different order from that of the actual investigation. 



To fix the ideas, it may be well to recall at the outset the 

 general plan of the experimental arrangement used in our 

 regular work, to which this article more directly applies. The 

 substance to be melted is contained in the crucible (fig. 2). 



* It may make for clearness to emphasize at the outset that the errors con- 

 sidered in this paper, which the investigation here described sought to 

 diminish, are errors of 5° or less, occurring in the regular work of this 

 laboratory upon silicate fusions. With the systematic differences of 100° 

 to 200° sometimes occurring in the literature, this paper has nothing to do. 



f This improvement has already been illustrated in a paper on Diopside 

 and its Eelations to Calcium and Magnesium Metasilicates, this Journal (4), 

 xxvii, 4, 1909. 



