HtLLEBKAND.] NOTES ON ROSCOELITE. 73 
From the column of means of my own analyses the following ratios 
are obtainable: 
SiO* 753 
V 2 3 159 
A1 2 3 113 
FeO 022 
MgO 041 
K 2 110 
H 2 229 
The entire absence of manganese and of calcium in my ow r n and 
Gentli's samples tends to confirm the suspicion that Eoscoe's material 
was far from pure. It is to be remarked, however, that my figures for 
vanadium agree quite closely with his and differ widely from Genth's. 
Discrepancies of this kind are not necessarily to be ascribed to faulty 
analyses. It is well enough known that in any one species of mica 
various molecules must sometimes be assumed to exist in different pro- 
portions, and the general formula for such a species can only be arrived 
at by comparison of a series of analyses of different varieties. Hence, 
in view of the lack of any simple ratios, the deduction of a definite and 
final formula from my data is not justifiable. Further analyses are 
needed of new and very pure material from other locations, even if 
these be not far removed from the source of the p resent material. 
Nevertheless, in the hands of an expert very unpromising data may 
often be made to afford positive indications, and that this is true in the 
present case the following discussion by Prof. F. W. Clarke clearly 
shows : 
Chemical Constitution of Roscoelite, by F. W. Clarke. 
The ratios given in the foregoing new analysis, nsed directly, lead to the follow- 
iug empirical formula for roscoelite: 
H458Ko 2() Fe 2 . 2 Mg^Al 226 V : j,KSi 7S 302724. 
Here H to K and Mg to Fe are as 2 to 1. Between O and Si, however, the ratio is 
not simple, and lies below the orthosilicate and above the trisilicate proportion. 
Sim-e in many micas the groups Si0 4 and Si: 3 8 are replaceable, that suggestion may 
be followed out here ; and then the formula reduces to 
R / (i78 R // ,; 3 AWV3i8(SiO0^ ; (Si3OH) 7 ,. 
From this expression, applying Clarke's mica theory, the mineral may be regarded as 
a molecular mixture of the three compounds 
1. 2. 3. 
,Si0 4 =FeK .Si :} 8 -KH 2 ^SiO^KH, 
Al— Si0 4 =MgH . Al— Si30 8 =KH 2 Al— Si0 4 =V 
^SiOiEEEMgH, \si 3 8 =Al, x Si0 4 =V, 
in the ratio 21 : 22 : 159, or nearly 1:1:8. Upon reducing the analysis to 100 per cent, 
