402 PROCEEDINGS OF THE AMERICAN ACADEMY. 



The most serious discrepancies are to be noticed in the cases of potas- 

 sic and sodic chlorides, where the respective averages have been chosen 

 as the most probable values. These averages are almost identical with 

 the figures obtained by Ramsay and Eumorfopoulos. On the other 

 hand, Joly's meldometer, used by these investigators, seems to have been 

 much less satisfactory at lower temperatures ; for their melting points 

 of plumbic chloride and litliic chloride are widely different from Car- 

 nelley's. Repeated qualitative experiments have convinced me that 

 plumbic chloride is less easy to melt than argentic chloride, and lithic 

 chloride less easy to melt than argentic iodide ; therefore Carnelley's 

 calorimetric results have been adopted here as the more accurate. 

 Strontic nitrate, and strontic and baric chloride all begin to decompose 

 in the air at or near their melting points ; undoubtedly this fact is 

 responsible for the very low results obtained by Ramsay and Eumorfo- 

 poulos in these cases. Obviously such salts cannot be relied upon 

 as trustworthy thermometers, unless one takes the precautions observed 

 by Meyer, Riddle, and Lamb in their work with the air thermometer. 

 One is reminded of the interesting work of Hill upon changeable 

 organic melting points.* The other substances in the list are sufficiently 

 stable, and are easily obtained in a pure state. In the light of Le 

 Chatelier's complete confirmation of Carnelley's results by means of the 

 pyrometric thermopile, it is odd that the lamented Victor Meyer ac- 

 corded Carnelley so little credit as he has done. 



It is needless to state that pains was taken to eliminate impurities 

 from these substances by recrystallization or other suitable means, and 

 to dry the salts thoroughly, in the work which follows. If the tempera- 

 ture was below 800°, the dry coarsely powdered substances were con- 

 tained in small melting-tubes of the hardest Jena glass, sealed while 

 hot at both ends, and sometimes provided with a small platinum sinker 

 to show whether or not the salt had liquefied. If, on the other hand, 

 the height of the temperature excluded glass, the substances were con- 

 tained in small cups beaten into stout strips of platinum foil. The small 

 amount of impurity taken from the glass tubes in which some of the 

 substances were melted could hardly have had much effect on the melt- 

 ing points until after a considerable mass of the substances had been 

 melted ; bitt then their mission had been fulfilled. 



The table above might well be amplified to cover a wider range, as 

 well as smaller differences of temperature, but such an amplification 



* H. B. Hill, These Proceedings, XXIII. 219. 



