Spodumene and Us Alterations. 323 



1. Norwich, Mass., by C. Doelter* (Min. u. petrog. Mittheil., 

 1878, New Ser., I, 517). 



2. Brazil, by C. Doelter (toe. cit.). 



3. Brazil, by F. Pisani (Oompt. rend., 84, 1509: 1877). 



The mineral from Brazil is transparent, yellowish -green, re- 

 sembling Chrysoberyl, but differing in inferior hardness (7), 

 and has a Sp. Gr.=3.16. 



FeO MnO MgO CaO Li 2 Na 2 K 2 



.39 .21 .73 7.04 1.10 .12 



1.15 



1.05 .12 — 





SiO 2 



A1 2 3 



1. 



Norwich 63.79 



27.03 



2. 



Brazil 63.34 



27.66 



3. 



Brazil 63.80 



27.93 



73 



7.04 



1.10 



69 



7.09 



.98 



46 



6.75 



.89 



By these analyses, Na : Li : : 1 : 15. 



In my analyses above given, the alkalies were separated by 

 J. L. Sm : th's method (ignition with calcium carbonate). The 

 residue of the three alkaline chlorides was carefully purified from 

 adhering traces of magnesia, by repeated precipitation with 

 barium hydrate. The excess of that reagent was removed from the 

 filtrate as carbonate, the solution of the chlorides evaporated to 

 dryness, and the Li CI separated by digestion in the mixture of 

 ether and alcohol. The lithia was then determined as sulphate, 

 and afterwards, for precaution, as lithium-phosphate, Li 3 PO 4 , 

 by repeated precipitation and thorough washing in the usual 

 way. However, the two methods did not produce accordant 

 results, the latter yielding amounts of lithium-phosphate sup- 

 posed to be pure, which in the two analyses were equivalent to 

 the following per centages : 



I II 



Lithia, 7.43 7.62 



The cause of these discrepancies has just been explained by 

 Eammelsberg (Monatsber. der Berl. Akad., 613-631, 1878). 



* In his determination of the formula of the mineral from Norwich, with the same result as 

 my own given below, Doelter previously eliminates as impurities o< er three per cent, out of 

 the percentage stated above, and yet finds an inaccuracy which he attributes to incipient 

 decomposition. This deducted amount he assigns to Orthoclase and Hedenbergite, which 

 have never been found as associates of Spodumene and are therefore of improbable occur- 

 rence as impurities. The detection of Killinite in masses, of which analyses are given beyond, 

 explains the given tinge which Doelter naturally refers to Hedenbergite, on the ground of the 

 isiinoiphous relationship of Augite and Spodumene. 



