/ 
70 M. Berzelius m Compowiids 
form ; I thought it requisite to prepare the magnesia for these 
experiments, in a manner which sliould entirely secure its purity 
from silica. I dissolved some sulphate of magnesia, and added 
to it, first a few drops of caustic ammonia, and afterwards a few 
drops of oxalate of ammonia, but the liquid continued limpid. 
I then precipitated it by a boiling solution of pure carbonate of 
potash ; the precipitate, being carefully washed, was dissolved 
by water strongly impregnated with carbonic acid gas ; and this 
solution was next filtered and boiled for some time, till the mag- 
nesia was again precipitated. From 10 grammes of caustic 
magnesia thus purified, when dissolved by pure dilute sulphu- 
ric acid, and evaporated to dryness, the product being maintain- 
ed at a red heat for a quarter of an hour in a crucible of platina, 
by the flame of a spirit-of-wine lamp, I obtained 29.3985 grammes 
of sulphate of magnesia, soluble in water, without any appre- 
ciable residuum. 
In several experiments made before I had adopted this me- 
thod of purifying magnesia, 100 parts of caustic magnesia con- 
stantly produced 29S.2 parts sulphate of magnesia ; but the salt 
obtained, after its solution in water, left always a residuum, con- 
sisting of magnesia mixed with a strong dose of oxalate of man- 
ganese. 
According to the experiment which I have just detailed, 100 
parts of sulphuric acid are saturated by 51.55 parts of magnesia, 
which must therefore contain 19.954 parts of oxygen. But 
51.55 : 19.954 : : 100 : 38.708, agreeing very nearly with the re- 
sult of my former experiments, in which I had found 38.8 parts 
of oxygen. 
The experiment which, in my former investigation, had given 
39.8 parts of oxygen, consisted in decomposing a given weight 
of sulphate of magnesia by muriate of barytes. I thought it 
worth while again to examine the analysis, and for this purpose 
I made use of the pure sulphate of magnesia obtained in the 
preceding experiment. I then found, to my great surprise, not 
only that the precipitate gave indication of more sulphuric acid 
than the salt had contained, but also that the quantity of this 
precipitate varied in different experiments. 10 grains of sulphate 
of magnesia should have produced 19.2 grains of sulphate of bsu 
, rytes ; but I obtained from 19.64 to 19.81 grains. Having re- 
