298 
DR. TURNER ON CHLORIDE OF BARIUM. 
methods, owing to a trace of the chloride being dissolved by the strong acid 
solution. 
It may be inferred, as a mean of the four first experiments, that 100 parts 
of silver correspond to 132.83 parts of chloride of silver. The proportion 
stated by Berzelius is 100 to 132.75 ; and it is estimated at 100 to 132.72 by 
Dr. Thomson. All these results, therefore, are closely correspondent. 
From one of the experiments (No. 4.) just mentioned, it is manifest that the 
precipitation of chloride of barium by nitrate of silver does not involve any 
appreciable source of error. To be quite certain, however, as to this fact, 
chloride of barium was mixed with nitrate of silver in excess, and the precipi- 
tate carefully washed. It was then boiled in distilled water, and the fluid ex- 
amined for silver and baryta ; but not a trace of either could be detected. It 
dissolved completely in ammonia, and the addition of sulphuric acid did not 
cause the slightest turbidity. 
In five analyses made by precipitating chloride of barium by an excess of 
nitrate of silver, I obtained the following proportions. 
Chloride of Barium. Chloride of Silver. 
Exp. 1. 100 yielded 137-45 
2. 100 137.54 
3. 100 13770 
4. 100 137-62 
5. 100 137-64 
Though all these analyses were made with great care, the last two were the 
most successful, as being less influenced by errors of manipulation than the 
others. Instead, therefore, of taking the mean of the five, which is 100 to 
1 37.6 1 , I adopt the mean of the two last experiments, which is 100 to 137-63. 
In one of these the precipitate was washed with distilled water only, and in 
the other with water acidulated with nitric acid. Berzelius in his experiments 
on this subject found that 100 parts of chloride of barium corresponded to 138.06 
in one experiment, and 138.08 in another. This is the only material difference 
between us which I have yet had occasion to notice. It induced me to recon- 
sider every part of my experiments ; but as I am unable to detect the slightest 
inaccuracy in the two analyses from which my result was derived, I cannot 
