592 N. L. Bowen — Melting Phenomena of the 



mined temperatures. None of the determined points lies far- 

 ther from the calculated curves than the possible limits of 

 experimental error warrant. The greatest deviation occurs 

 where the experimental errors are greatest. 



Table V is given in order to show that this agreement is 

 clearly not fortuitous and that no values of the latent heats 

 except those very close to the calculated values will give curves 

 that pass near enough to the experimentally determined points 

 to permit the interpretation that the deviation of the points is 

 due to experimental error. 



The table gives the temperatures of some points on liquidus 

 and solidus curves calculated by the use of values of the latent 

 heats differing 10 per cent from the mean values 12740 and 

 29000 calories. The latent heats are allowed to vary from the 

 mean values both separately and jointly and, in the latter case, 

 both in the same direction and in opposite directions. In each 

 case the differences between the calculated and observed 

 temperatures are well beyond the limits of error of the tempera- 

 ture measurements. The observed temperatures, therefore, 

 fix closely the values of L An and L Ah . 



Table V. 



Case I. 



Assuming a value of L Ab 10 per cent lower than above mean value 

 and a value of L An equal to above mean value 





Point on 



Solidus. 





Composition 



Calculated 



Observed 



Difference 



Ab,An B 



1475° 



1465° 



+ 10° 



Ab>n 2 



1390° 



1372° 



+ 18° 



Ab 1 An 1 



1305° 



1287° 



+ 18° 





Point on Liquidus. 





Ab 1 An 2 



1493° 



1490° 



+ 3° 



Ab a An x 



1454° 



1450° 



+ 4° 



Ab 2 An T 



1400° 



1394° 



+ 6° 





Case 



77. 





Assuming a value of L Ab equal to above mean value and a value 

 of L An 10 pe,r cent less than mean value 



Point on Solidus. 



Composition Calculated Observed Difference 



Ab.Am 1458° 1465° 





Ab,An„ 1363° 1372° -9 C 



