MINERAL CONSTITUENTS OE METEORITES. 
191 
that it affords the means of analysing the separate minerals selected by it ; but even 
thus the difficulties are considerable. The quantity of material at one’s disposal amounts 
generally to a very few grains, and of these but a small proportion consists of the desired 
mineral ; in fact half a gramme is as much as one is generally able to obtain in a state 
of purity. 
The analysis of so small an amount of a silicate is a difficult problem. To break up 
the silicate by fusion with alkalies implies the introduction of foreign and non-volatile 
ingredients; to estimate the silica by its loss on treating the mineral by ammonium 
or hydrogen fluoride were to lose what, in the analysis of such small quantities, is the 
necessary check afforded by the summing of the percentages of its constituents. 
To distil the silica in the form of silicium fluoride, and then to determine it either as 
silica or as potassium fluosilicate, suggested itself as a means by which this check might 
be secured. A long series of experiments, undertaken in order to ascertain the best 
process for thus determining the silica, has resulted in a completely successful application 
of the method ; and by it so small an amount as two-tenths of a gramme of enstatite or 
of augite has been analysed with a satisfactory result in the Laboratory at the British 
Museum. 
II. On a new Method of analysing Silicates that do not gelatinise with Acids. 
The method adopted for the analysis of silicates in small quantity, to which reference 
has already been made, was the following. 
Hydrogen fluoride formed from picked fluor-spar was conducted into water. When 
the saturation had reached the point at which the liquid gave off fumes, the acid solu- 
tion was treated in a platinum dish with potassium fluoride so long as any precipitate 
was formed. After decantation the acid was distilled, the first and last portions being 
omitted, and the distilled acid preserved in a platinum bottle. A leaden bottle, even 
when lined with pure gutta percha deposited from its solution in benzole, appears to be 
attacked by the acid. 
A small platinum retort of a capacity of 30 cub. centims. (fig. 1) has fitted into it a 
tubulated stopper (fig. 2) reaching nearly to the bottom of the retort ; a small tube ( b ) 
Eg- 4. 
enters the straight tube (Jc) of the stopper (a) at an angle above the neck of the retort, for 
the delivery of hydrogen. The straight tube can be stopped either by a small platinum 
2 c 2 
