52 THE ROYAL SOCIETY OF CANADA 



Several indicators from a standard set are placed in a large test- 

 tube, some being lighter and others heavier than the substance whose 

 specific gravity is required, which also is introduced — preferably three 

 or four pieces, if obtainable. The heavy liquid to be used is then 

 poured in, and the tube closed with a cork through which two holes 

 have been bored. A stirrer is introduced through one hole and the 

 tip of a burette is inserted through the other, both fitting loosely. The 

 burette being filled with the diluent, the latter is now run in, with 

 constant stirring of the contents of the test-tube, until the first indica- 

 tor remains exactly suspended, or barely sinks, when the burette 

 reading is taken. Further readings are made as the liquid attains 

 equilibrium with each of the remaining indicators, and also with the 

 fragments of the unknown mineral. The burette readings are then 

 plotted against the known specific gravities of the indicators, and 

 the value for the mineral is found from the point at which its burette 

 reading intersects the resulting curve. 



It is not necessary to use more than three indicators, whose 

 limits of specific gravity embrace that of the mineral fragment to be 

 identified, since the data obtained from them will furnish three points 

 which determine the curve ; but in practice it is better to obtain read- 

 ings for five or six indicators, some lighter and others heavier than 

 the substance, whereby the curve can be more accurately drawn. 



The method can be made as accurate as desired by employing a 

 comparatively large volume of the original heavy liquid, since corres- 

 pondingly large amounts of diluent will then have to be added in 

 order to produce a given change in the density of the liquid; the 

 curve will approach the form of a straight line, and points upon it 

 which indicate only a small difference in specific gravity will be widely 

 separated. 



The advisability of starting with as much as, say, 15 or 20 c.c. of 

 the original liquid renders this method more suitable for use with 

 methylene iodide than with any of the aqueous solutions, such as 

 Thoulet's, which would have to be diluted with water, and subse- 

 quently concentrated again to be of any further use. In the case of 

 methylene iodide, the question of volume is not a very serious matter, 

 since the liquid solidifies at about 5° C, and can by this means be 

 readily freed from the added benzene. 



The test-tube being closed, there is comparatively little loss of 

 benzene by evaporation during the course of a determination; and 

 since any evaporation which does take place proceeds at an essentially 

 constant (or only slightly increasing) rate, the possible error due to 

 this factor is about equally distributed throughout the plotted curve, 

 and does not materially aft'ect the accuracy of the method. 



