REPORT ON THE COMPOSITION OF OCEAN-WATER. 
91 
tillation of the water with chlorine, and collecting the liberated bromine (and the excess 
of re-agent) in pure caustic ^potash. To test this method, a large number of trials were 
made with synthetically prepared sea-waters, containing known weights of pure bromide 
of potassium; but the results were unsatisfactory. Even when six equivalents of chlorine 
were used to expel one equivalent of bromine, the residue still contained a remnant of 
undecomposed bromide, whose bromine, it is true, was found capable of being eliminated 
by one or two repetitions of the process ; but the method then becomes rather trouble- 
some, besides being uncertain as to its exhaustiveness. In addition to this, the determina- 
tion of the bromine in the distillate proved not so easy as I had expected. The reduction 
of the bromate is, of course, easily effected by means of sulphurous acid ; but the haloid 
of silver precipitate obtained from the reduced liquor is often contaminated with a 
dark-coloured (sulphur?) compound, which has to be removed by hot nitric acid : a risky 
operation under the circumstances. 
Several attempts to separate the bromides from the bulk of the other salts by 
means of alcohol gave still more discouraging results. I therefore at last came back 
to the original method, which, as the result of a number of trials, assumed ultimately 
the following form : — 
A decinormal solution of nitrate of silver is prepared by dissolving 17 grms. of the anhy- 
drous salt in 20 c.c. of nitric acid of 1 '4 specific gravity, and enough water to produce 1 litre 
of solution. This solution serves for the precipitation of the bromine, and its titre must be 
sufficiently exact to enable one to maintain in all the analyses a constant ratio between 
silver added and silver-equivalent of the total halogen present ; so that, assuming the 
method to be infected with some latent inherent error, the results retain at least their 
comparative value. For the same reason an exact determination of the “ chlorine ” always 
precedes that of the scarcer halogen. The modus operandi is as follows : — 
One kilogram of the sea-water is weighed out, and mixed with a volume of the silver- 
solution, equivalent to exactly 1 jnth of the total halogen present, the mixture shaken and 
put aside in a dark place. What the right value for n is will be explained by and by. 
After the precipitate has settled down, which generally takes about twelve hours,* the 
supernatant liquor is syphoned off, the precipitate washed by decantation, and the 
washings are added to the decantate. The precipitate is then transferred to a basin, 
and dried over a water-bath, in the absence of light. The dry residue is then 
collected in a porcelain boat, which has previously been tared within, and along with a 
piece of combustion-tubing, which is about to 2 inches longer than the boat itself. 
Within this tube the dehydration of the haloid is effected by fusion in a current of dry 
air, and the chlorination by repeated heating in a current of dry chlorine until the 
weight is constant. As the latter operation is generally accompanied by a slight 
effervescence (which would otherwise lead to a loss of from 1-3 mgrms. of chloride), 
* In a neutral mixture the precipitate remains suspended for days as an intractable milk. 
