W. P. White — Melting Point Determination. 467 



approach it from the side and not from the end. This calls 

 for a thermoelement of fine wire, and for a charge narrow but 

 not too short. Failure to center the thermoelement is evidently 

 a minor fault. A thermojunction inclosed in a porcelain tube is 

 from the present point of view undesirable, since its temper- 

 ature is dependent on a considerable portion of the charge. A 

 resistance thermometer is evidently still worse. In fact, the 

 difference in sharpness is very marked between melting points 

 measured with an inclosed and a bare thermoelement. But a 

 bare element of platinum wire *6 mm in diameter, immersed 8 mm 

 in a cylindrical charge of 9Xl4 mm , has been found to give 

 (from 800° up) points quite as sharp as can be obtained with 

 wire much smaller. Since finer wire than this and narrower 

 and higher charges can be easily used, there seems to be no 

 difficulty in avoiding the effect of irregular temperature dis- 

 tribution for substances in which a bare element is admissible. 

 A few experiments indicate that at lower temperatures charges 

 narrower in proportion to their height may perhaps be needed, 

 but at lower temperatures they can also easily be employed. 



VI. Conduction of heat along the thermoelement. — The ther- 

 mal conductivity of most substances is much less than that of the 

 metal thermoelement, so that the junction often receives heat 

 along the wires. This causes a gradual additional rise of 

 temperature toward the end of melting, thus increasing 

 the obliquity, and it also tends to make the final break come 

 high. The effect is little altered by change of rate, since this 

 acts in the same way upon the heat, whether coming through 

 the charge or down the wire. It increases with the diameter 

 of the charge, since this increases the time of melting for the 

 same external gradient. It is diminished by thinness of the 

 thermoelement, and by depth of immersion in the charge. 

 The avoidance of this cause of obliquity, then, demands the 

 same conditions as for the preceding, namely narrow and high 

 charges and fine thermoelement wires. And the experience 

 just now cited to show that the one cause of obliquity can be 

 easily avoided is equally conclusive with respect to the other. 

 On the other hand, an enclosed element, immersed 3 cm in a cru- 

 cible of sodium chloride, gave a melting point 7° too high. 

 This error is much greater than that found with some other sub- 

 stances melting at higher temperatures,* but it shows that 

 melting point results obtained with inclosed thermoelements 

 must be interpreted with care. A sufficient estimate of the 

 error resulting from heat conduction down the thermoelement 

 can in general be easily reached by varying the depth of 



*See, e. g., Isomorphism and Thermal Properties of the Feldspars, pp. 

 23-25 ; Diopside and its Relations to Calcium and Magnesium Metasilicates, 

 pp. 3-4, loc. cit. 



