390 E. T. Allen, F. K Wright and J. E. Clement— 



sealed on without delay and the crucible connected with a 

 generator furnishing a slow stream of well-dried hydrochloric 

 acid gas.* In our experiments the crucible was heated over a 

 burner to about 1000°. It was not possible to measure the 

 temperature inside the crucible during the progress of the 

 work, but in the empty crucible a bare thermoelement, touch- 

 ing the bottom, read 1050° just before the charge was inserted. 

 This was therefore the approximate temperature in the hottest 

 part of the crucible. In the first experiment the outlet was 

 left unprotected against moisture. After four days the silicate 

 was found to be converted entirely into fosterite. 



In all the later experiments with magnesium chloride, the 

 outlet tube of the crucible was well guarded by a II-tube con- 

 taining sulphuric acid and a straight tube containing calcium 

 chloride to prevent the passage of moisture into the crucible. 

 At temperatures between '1000°-1100° (in the hottest part of 

 the crucible) we invariably obtained some fosterite and peri- 

 clase as by-products, but in the main the substance crystal- 

 lized as the monoclinic pyroxene. This method gives by far 

 the best results of any we have discovered. 



Optical Constants of Fosterite. — Inasmuch as crystals which 

 appeared to be fosterite were often obtained in the course 

 of our work, — crystals which by optical tests alone it was 

 found difficult to distinguish from minute crystals of ensta- 

 tite, it seemed advisable to determine the crystallographic 

 constants of these from magnesium chloride. The crystals were 

 colorless, transparent, and about O2x0-lX0*l mm in size, short 

 prismatic in habit, doubly terminated, and similar to natural 

 fosterite in appearance. The cleavage, perfect after 001 and 

 010, was obtained by actual fracture of a crystal under the 



Table I. 





Letter. 



Symbol. 



Miller. 



Measured. 



| Goldsclimidt.f 



No. 



<P 



P 







P 



1 



b 







001 





0° 00' 





0° 00' 



2 



a 



0oo 



010 



0° 00' 



90 00 



0° 00' 



90 00 



3 



n 



GO 



110 



64 26 



90 00 



65 01 



90 00 



4 



d 



10 



101 



90 00 



51 37 



90 00 



51 32 



5 



k 



02 



021 



00 



49 44 



00 



49 39 



6 



e 



11 



111 



65 51 



54 19 



65 01 



54 15 



*The platinum is often slightly attacked during the fusion. Whether 

 this comes from the decomposition of the chloride by oxygen : 



MgCl 2 + O + MgSiOs = Mg 2 Si0 4 + 2C1 (?) 

 has not been worked out. When the magnesium coloride is mixed with 

 sodium chloride, the platinum is very strongly attacked. 



f V. Goldschmidt, Winkeltabellen, Berlin, p. 251-252, 1897. 



