470 W. P. White — Melting Point Determination. 



no part of the very small charge can then be at or near the 

 melting point itself. And therefore some heat absorption due 

 to melting must occur at least 20° below the melting point. 

 The obliquity is also, within the experimental error, such as 

 would be expected from the known impurity, and can be 

 increased in silicates, and produced in substances originally 

 pure, by the. addition of impurity. 



Fig. 4. 









c 



/ 















B 







\ 



A. 













A 



lO 



\ \ 



\ ___ 





'X 



^X 



o 



1— 1 



Z \ 



y^W 









o 



- / 











02 

 CD 

 CD 



3d 



ft 













Heat 



Fig. 4. Actual melting curves 

 (melting point 801°). C, 



Heat 



y curves. A, Diopside (melting point 1392°). 

 ', Na 2 S0 4 (melting point 885°) (see text pp. 



,. B, NaCl 

 pp. 469-471). 



The slope of the curve at the top from ¥ to X is inconsist- 

 ent with equations (1) to (4) and is not easily accounted for by 

 any other known impurity effect. An attempt was made to 

 determine its cause. Variations in rate of heating, in the 

 depth to which the thermoelement was immersed, and in the 

 size of the wire, did not materially alter this part of the curve, 

 thus showing that neither distortion (effect IV) nor conduction 

 down the thermoelement (VI) was responsible. 



An easily crystallizing substance was next tried. Sodium 

 chloride (freezing point 801° ±1°)* gave the curve B in fig. 4, 

 which shows the same peculiarity. Variation of rate and of 



* This is a new determination. 



