81 
“The carboute acid was determined by the ‘well known method. 
08 the weighed flask and Ca Cl tube. 
The other constituents were determined in the weed manner, 
se dissolving the mineral in HCl, etc. In making the two fol- 
lowing analyses, the most extreme precautions were used to en- 
su 
Ber being weighed. Thus, the silica, after being weighed 
at: left i in it by the HCl separated, weighed, and added to the HCl 
solution, their weight being deducted from that of the silica. In 
one case, this weight amounted to half a per cent. of the sub- 
stance used. Difficulty being found in separating the Al and Fe, 
by means of KO, in one experiment the #e and 41 were weighed 
together, and the x aig by deducting the amount of Be af- 
terwards found in _ ma 
IL. Mean. Oxygen 
Si 7 27°32 27-51 14-29 
Al 10°30 10°17 10°23 4°78 
Fe 2T-99 97°55 2777 617 
Na 1:23 ea 1:23 39 
8°69 8-73 8-71 14 
Ga 12-25 12-98 12°61 ; 
| é 10°32 10°09 10-29 
98°35 
The first thing to be observed bere. is that the quantity | * : 
found i is just sufficient to form Ca€ w b.the Ga found, 12°61 ¢ 
rally supposed that all the iron contained i in minerals which give 
Pp green powders is in vn form of FeO. This error of which 
so many examples may b e found in mineralogical works, has 
Its origin in the great difficulties always met with in the separa- 
tion of the two oxyds of iron from one another, and also in the 
small difference between the equivalents of the two oxyds,—9 per 
went. of protoxyd making only 10 per cent. of peroxyd. 
The suggestion of a scientific friend that the mineral might 
still soo both oxyds, led to a renewal of my resear 
The stock of elutriated material, originally prepared, being 
Sie, it was necessary to repeat the analysis from the be- 
ela on another specimen, which was accordingly done, in 
manner 
