^{page 266 note *} The only source of error to which this step of the analysis is liable, is that which will arise if more barytes has been used in the first operation, than was necessary to precipitate the sulphuric and carbonic acids. It will be thrown down in the state of oxalate of barytes, and be converted into carbonate and sulphate, and thus give the apparent proportion of lime too large. This is obviated, of course, by taking care to avoid using an excess of barytes. To render the operation of the oxalate of ammonia as perfect as possible in precipitating the lime, the water should be considerably reduced by evaporation, taking care to avoid any separation of any of its ingredients.

^{page 270 note *} According to the result of this last experiment, 100 grains of crystallised muriate of magnesia would give 64 of real sulphate of magnesia, composed of 21.3 of magnesia, and 42.7 of sulphuric acid. This quantity of sulphuric acid is equivalent to 29.4 of muriatic acid. Hence 100 grains of this salt crystallised (in which state its composition, I believe, has not been determined) consist of 21.3 magnesia, 29.4 muriatic acid, and 49.3 of water.

^{page 271 note *} For the sake of comparison, and to ascertain the accuracy of different methods, I submitted a similar solution of muriate of magnesia and muriate of soda to analysis by sub-carbonate of ammonia. To the saline liquor, heated to 100°, a solution prepared by dissolving carbonate of ammonia in water of pure ammonia, was added, until it was in excess. A precipitation rather copious took place; the precipitate being collected on a filtre, the clear liquor was evaporated to dryness, and the saline matter was exposed to heat, while any vapours exhaled. Being redissolved, a small portion remained undissolved, and on again adding sub-carbonate of ammonia to the clear liquor, precipitation took place, rather less abundant than at first. This was repeated for a third, and even for a fourth time, after which the liquor was not rendered turbid. Being evaporated, the muriate of soda obtained, after exposure to a red heat, weighed 20.5 grains. The whole precipitate washed, being heated with sulphuric acid, afforded of dry sulphate of magnesia 4.8 grains, a quantity inferior to that obtained by the other methods, evidently owing to the less perfect action of the ammoniacal carbonate as a precipitant. A similar deficiency in the proportion of magnesia was found in the analysis of sea water by sub-carbonate of ammonia, as has been already stated; while, on the other hand, in its analysis by phosphate of soda and carbonate of ammonia, a larger quantity of muriate of soda was obtained than by the other methods, probably from the difficulty of avoiding an excess of phosphate of soda in precipitating the magnesia.

^{page 274 note *} The analysis of sea-water in a preceding paper, will afford an illustration of this (page 237.)