6 Day, etc. — Determination of Mineral 



2. Method and Apparatus. 



After consideration of the experimental difficulties and the 

 opportunities for error in the few methods available for dens- 

 ity determinations at high temperatures, we adopted the 

 method of Archimedes, namely, weighing the displacement of 

 a liquid of known density. The principal difficulty, as with 

 all investigations at high temperatures, is with materials of 

 construction. Those used finally were graphite, which in a 

 reducing atmosphere retains its form indefinitely, JMarqnardt 

 porcelain, which does not soften appreciably below 1600°, and 

 alundum (A1„0 3 ), which melts at about 2050°, but is very por- 

 ous. Pure magnesia is also available, but lacks mechanical 

 strength and permanence. 



The liquid to be used should have the following properties : 

 (1) high boiling point, (2) low melting point, (3) low volatility 

 between these temperatures, (4) high density, (5) no chemical 

 reaction with graphite, (6) no action on the minerals to be 

 examined. Silver and tin best fulfil these conditions. Tin 

 has the advantage over silver of a very low melting point 

 (232°), and the disadvantage of lower sensitiveness because of 

 its lower density. 



The substance to be investigated, being always lighter than 

 the metal, was contained in an inverted graphite crucible. The 

 measurement consisted in determining the weight necessary to 

 immerse the crucible and material to a given depth on the 

 stem of the crucible. Because of the gases given off by natu- 

 ral rocks and minerals when heated, both the liquid znetal and 

 the containing crucible must be such as to release these gases 

 (or imprisoned air) at once. Imprisoned gas bubbles would 

 destroy the significance of the measurements. The success of 

 the particular form of apparatus used therefore depends upon 

 the permeability of graphite to these gases. As noted above, 

 the use of a graphite crucible requires the maintenance of a 

 reducing atmosphere within the furnace. 



In the form of apparatus first tried, the weight was applied 

 at the top of a long porcelain stem projecting upward from the 

 crucible through the top of the furnace. But the center of 

 gravity of this system was too high, with the result that to 

 prevent tilting guides were required, which introduced too 

 much friction. The weight was therefore applied below (fig. 

 1), by means of a cage surrounding the large crucible contain- 

 ing the metal. 



The fixed level to which the float-crucible was immersed 

 was found by electrical contact. A circuit from a storage cell 

 through a galvanometer entered the bottom of the furnace and 

 made connection with the graphite crucible containing the 



