Hillebrand, Turner and Clarice — Roscoelite. 453 



may fairly be assumed that his water was weighed directly 

 after expulsion by ignition of the powder. Roscoe's figures 

 for water, if not for moisture, must be affected by error, prob- 

 ably arising from the unsuspected oxidation of vanadium. 



Analyses of Roscoelite. 



Amount 

 used. 



Si0 2 __ 



Ti0 2 



Y 2 3 



A1 2 3 



FeO 



MgO 



K 2 



Na 2 



Li a O 



H 2 below 



105° C... 

 H 2 105- 



280° C. . 

 H 2 above 



280° C... 

 Fl 



Hillebrand. 



45-30 



•77 



23*90 



1-59 

 1-64 



•25310. 



•2635#. 



•15600. 



4504 



•77 





24-00 





24-09 



1174 



11-34 



1-65 



10-41 



10-32 





•13 



none 





ft. tr. 



! 



4-29 





•80 

 2406 



157 



45-17 



•78 



24-01 



11-54 



160 



1-64 



10-37 



•06 



ft. tr. 



•40 



•17 



4-12 

 none 



Genth. 



47-69 



20-56 

 14-10 

 1-67 

 2-00 

 7-59 

 •17 

 tr. 



4-96 

 ign. 



99-86 98-76 



Eoscoe. 



SiCV— . 

 T,0.._. 



A1 2 3 _. 

 Fe 2 3 .. 

 Mn 2 3 __ 



CaO 



MgO ... 

 K 2 0'____ 

 Na a O___ 

 Water _. 

 Moisture 



41-25 



28-60f 



14-14 



1-13 • 



1-15 



•61 



2'01 



8-56 



•82 



1-08 



2-27 



101-62 



From the column of means of my own analyses the follow- 

 ing ratios are obtainable : 



Si0 2 



v 2 o 3 . 



A1 2 3 



FeO 



MgO 



K 2 



H 2 0. 



•753 

 •159 

 •113 

 •022 

 •041 

 •110 

 •229 



The entire absence of manganese and of calcium in my own 

 and Genth's samples tends to confirm the suspicion that Ros- 

 coe's material was far from pure. It is to be remarked, how- 

 ever, that my figures for vanadium agree quite closely with his 

 and differ widely from Genth's.J 



Discrepancies of this kind are not necessarily to be ascribed 

 to faulty analyses. It is well enough known that in any one 



*4-94 per cent oxygen used for complete oiidation instead of 5*27 needed 

 for all V" as V 2 3 and Fe as FeO. 



f Equivalent to 23-59 per cent V ? 3 . 



X This agreement is probably more apparent than real for the reason that 

 Roscoe's figure would be materially lowered by regarding the iron of his analysis 

 as FeO instead of Fe 2 3 in conformity with my analysis. 



