2i8 ALBERT JOHANNSEN AND E. A. STEPHENSON 



the separation was complete. The separation of the plagio- 

 clase (Sp. gr. 2.68) from the quartz (Sp. gr. 2.65) was the most 

 difficult. As an indicator, therefore, a small Herkimer County 

 quartz crystal was placed in the solution, which was then diluted 

 until the crystal began to sink, after which a very small amount of 

 concentrated solution was added to make it slowly rise to the top- 

 This adjustment between the specific gravity of the quartz and that 

 of the liquid is here so delicate that a slight change in the density 

 of the solution, such as results from a slight change in the tempera- 

 ture of the room, caused the quartz, which has a very small coeffi- 

 cient of expansion, to settle during the night and rise again during 

 the day. When the equilibrium was established, the plagioclase 

 was removed, washed, dried, and weighed as in the other cases. 

 The quartz was brought down by adding enough water to bring 

 down the indicator. The final material was the orthoclase. 



The greatest objection to the specific-gravity method seems to 

 be the impossibility of preparing powder which consists of homo- 

 geneous material. In many cases material of one kind will cling to 

 grains of another and will increase or decrease their specific gravi- 

 ties. The small amount of mixed material shows up very clearly 

 in the separation, especially when between light and dark minerals, 

 for the mixed heavier grayish material will come down before the 

 pure white mineral, or, on the other hand, the precipitated dark 

 mineral will be covered with a thin layer of material that is grayish. 

 The probabilities are that the two errors compensate. The material 

 should be crushed fine enough to separate each constituent from 

 every other, but not so fine as to produce dust, since some minerals 

 pulverize much more readily than others and a different proportion 

 of each will be lost, destroying the accuracy of the result. A 

 happy medium must therefore be taken between purity and selec- 

 tive sieving, since absolute purity cannot be expected. Micro- 

 scopic examination of a small amount of the material of a certain 

 size will readily make it possible to determine whether or no the 

 rock has been crushed fine enough to produce a good separation of 

 the constituents. For the following experiments various-sized 

 grains were used, as indicated at the top of each column in 

 Table IV. After crushing and sifting, the total samples weighed 



