RICHARDS AND CHURCHILL. — TRANSITION TEMPERATURES, 279 



These gentlemen studied in a very hasty fashion the transition tem- 

 perature of Glauber's salt in the presence of an excess of common salt. 

 Although theoretically sound, such a system labors under a serious prac- 

 tical disadvantage ; for the addition of heat to it means the dissolving 

 of common salt as well as the melting of crystallized sodic sulphate and 

 the depositing of anhydrous material. The first of these processes is 

 obviously less speedy than the others, and must surely occupy appreci- 

 able time even if the solid is finely powdered. Such a " lag " inevitably 

 affects the temperature ; our own experience with this mixture as well 

 as with other similar ones su[)ports this inference, and is anything but 

 reassuring. Indeed, we found that Glauber's salt itself did not give 

 absolutely accurate results if it was allowed to " freeze " instead of to 

 " melt," for a similar reason. 



We agree with Messrs. Meyerhoffer and Saunders as to the great 

 desirability of uniting upon some normal temperature for the graduating 

 of flasks, etc., but we cannot conclude with them that 18° is the best 

 temperature. In America the steam-heated winters and sun-heated 

 summers raise the average temperature of our laboratories at least to 

 20°, and indeed this temperature is more comfortable than 18° unless 

 one is performing active manual labor. Hence at Harvard we have set- 

 tled upon 20° as the normal room temperature. Sodic chromate (19.G°) 

 clearly gives us very nearly the standard of reference which we desire. 

 In determining the specific gravities of liquids, a temperature above that 

 of the room is preferable to one beloio, — for the expansion of the liquid 

 during the drying of the exterior of the pycnometer is otherwise apt to 

 be troublesome, hence 18° is not suitable for this purpose. The authors 

 before mentioned suggest the use of a bath of mixed salts as a means of 

 keeping the temperature constant during determinations of electrolytic 

 conductivity ; but it should be pointed out that in such work the neigh- 

 borhood of a very large amount of a good electrolyte is necessarily risky, 

 except in the best of hands. This is especially the case when the sub- 

 stance effloresces to form a fine powde"", easily wafted around by currents 

 of air. In short, while for some work with closed vessels demanding the 

 greatest accuracy such a bath may be invaluable, the Ostwald thermostat 

 is the safest and most convenient appliance for preserving a constant 

 temperature in the laboratory. The baths of " melting " crystals will 

 find their greatest use in the standardizing of thermometers at fixed 

 points ; and these thermometers will continue to serve as the most 

 handy means of attaining and registering any desired temperature. It 

 is obvious that if a thermometer is standardized under exactly the 



