| 
5 
. 
. 
: 
Z 
. 
. 
4 
ANALYSIS OF MINERAL WATERS. 267 
The decomposition of the magnesian salt by ammonia would 
have the former advantage, as the muriate of ammonia would 
be expelled at the end of the process by heat; but this de- 
composition, it is well known, is only partial. Sub-carbonate 
of ammonia causes a more abundant precipitation of magnesia, 
but still its action is likewise partial, a ternary soluble salt be- 
ing formed after a certain quantity has been added. It seem- 
ed probable, that this might be obviated, by adding the sub- 
carbonate of ammonia as long as it occasioned any precipita- 
‘ tion, then evaporating the clear liquor to dryness, expelling 
the muriate of ammonia, and any excess of ammonia, by heat, 
re-dissolving, and again adding the sub-carbonate of ammonia 
to decompose the remaining magnesian salt. Proceeding in 
this way, I found that a copious precipitation took place on the 
second addition, and even at the fourth a small quantity of 
precipitate was thrown down. But the decomposition, after 
all, was not perfect, for the quantity of magnesia obtained was 
not equal to what was procured by other methods. 
Sub-carbonate of soda or potash has been usually employed 
to precipitate magnesia from its saline combinations. The 
precipitation, however, is only partial, unless an excess of the 
precipitant be employed (and even then, perhaps, is not alto- 
gether complete); and as this exces cannot easily be estima- 
ted, it introduces a source of error in estimating the quantity 
of muriate of soda at the end of the operation, against which it 
is not easy to guard. 
The method proposed by Dr Woxtaston, of precipitating 
magnesia from its solution, by first adding carbonate of am- 
monia, and then phosphate of soda, so as to form the insoluble 
phosphate of ammonia and magnesia, is one much more per- 
fect ; the whole of the magnesia appears to be precipitated, 
and as a method, therefore, of determining the quantity of this. 
L112 base, 
