W. P. White — Melting Point Determination. 465 



or two lower down, due to the rapid increase in heat supply 

 before the melted layer has touched the thermoelement at all. 



In brief, then, the inevitable heat distribution in an unstirred 

 charge has this effect on the center, where the thermoelement 

 is located. It retards the normal temperature rise at the 

 beginning of the melting and partly makes up the loss by 

 accelerating it at the end, thus increasing the obliquity of the 

 curve just where such increase is most undesirable. This 

 effect increases somewhat less rapidly than the rate of heating 

 and than the square of the diameter. The difference of tem- 

 perature between center and surface remains nearly constant 

 for small crucibles, but diminishes during melting for large 

 diameters and rates. 



3. Magnitude of the distortion. — Diopside, a well-crystalliz- 

 ing silicate, melting at 1392°, showed a time lag of 24±6 sec- 

 onds for a crucible of l cm radius before melting began. The 

 charge was in shape a short cylinder. The latent heat was 

 enough to raise the solid over 300°, but was so distributed 

 that the maximum value of the virtual specific heat was only 

 about fifty times the true specific heat. 



If, now, this substance is heated in a crucible of l cm radius 

 with a constant heat supply, giving a temperature rise before 

 melting begins of 8° per minute, the melting will last (roughly) 

 thirty-five minutes. The time lag (surface to center) would 

 by formula (6) reach twenty minutes for a charge all at the 

 maximum specific heat, but in this case, as experiment shows, 

 will be not much over half that. Even so, however, the period 

 of accelerated heat supply to the center will last ten minutes, 

 so that the last third (or more) of its melting will be distorted. 

 This third covers an interval of 2 to 3°. The resulting uncer- 

 tainty in the melting point would be less than that, and, there- 

 fore, not very serious, but with a crucible of 4 cm diameter the 

 uncertainty would be much increased. 



If the crucible has l cm diameter (-5 cm radius), the maximum 

 lag by formula (6) is only five minutes, or one-seventh the 

 melting interval, and the whole accelerated interval is not over 

 half a degree. If this smaller crucible, however, contains a 

 substance of greater purity, say sodium chloride, melting 

 through a 1° interval, the virtual specific heat might reach 400 

 times the true, and the accelerated interval would again cover 

 a third or more of the melting. 



The distorting effect produced on the melting curve by the 

 normal gradient through the charge is, then, practically 

 important only in large charges, and though serious in them, 

 can be rendered negligible by a reduction of dimensions which 

 is easily attainable in practice. The distortion is also rela- 

 tively greatest in the purest substances, where the total 

 obliquity is least detrimental. 



