320 Penfield and Sperry — Mineralogical Notes. 



present no special features which have not been noted in pre- 

 vious papers. The occurrence of alkalies and water, even in 

 our most transparent and purest beryls, is well established. 

 The water must still remain problematical. It has previously 

 b>een determined as loss on ignition, but we have also made di- 

 rect determinations by heating the mineral in a Gooch tubulated 

 crucible, collecting the water in a chloride of calcium tube, 

 with special care, and obtained results agreeing with the loss on 

 ignition. If this water is to be included in the formula, we 

 would suggest, as was done in a previous paper,* using the 

 usually accepted formula of beryl, Be 3 Al 2 Si 6 18 , plus one-half 

 a molecule of water. This expresses nearly the quantity which 

 is present, while the excess, over 0*50 H 2 in the above ratios, 

 may be accidental or perhaps even basic water, 3H 2 being 

 equivalent to A1 2 3 and H 2 to BeO. That all of the water 

 is basic seems to us improbable, as it is present in too large 

 proportion, and if added in any way to the A1 2 3 and BeO, it 

 would distort the simple ratio of Si0 2 : A1 2 3 : BeO to too 

 great an extent. The excess of H 2 over and above one-half 

 a molecule in the three analyses is I, 0*63 ; II, 0*70 ; III, 042 per 

 cent. Such quantities of accidental water, as we may call it at 

 present, are not uncommon in the analyses of our ordinarily 

 accepted anhydrous minerals. 



2. Phenacite. 

 Having proved that alkalies are almost always present in 

 beryl, replacing the BeO, we decided to analyze a phenacite, 

 with especial reference to the detection and estimation of any 

 alkalies that might be present. The material, which was fur- 

 nished to us by Mr. Geo. F. Kunz, of ~New York, was the flat 

 rhombohedral variety from Topaz Butte, near Florissant, 

 Pike's Peak region, Colorado. All of the material was clear 

 and colorless and had a specific gravity between 2*966 and 

 2*957, determined by means of the heavy solution. The analy- 

 ses are 





Sperry. 



Penfield. 



Mean. 



Calculated for 

 Be 2 Si0 4 . 



Si0 2 



54-46 



54-42 



54-44 



54-47 



BeO 



45-57 



45-60 



45-58 



45-53 



Na 2 



0-21 





0-21 





Li 2 



trace 









IgD. 



0-26 





•26 





100-49 



The atomic weights Be— 9*08 and 0=15*96 were used in cal- 

 culating the above values, while for Be =9*4 and = 16*, the 

 values are Si0 2 = 54*15 and 4585. Special care was taken in 

 the determination of the alkalies, the solutions coming in con- 



Tais Journal, III, xxxii, 110. 



