REPORT ON THE COMPOSITION OF OCEAN-WATER. 
93 
very nearly 10 mgrm. equivalents of haloid, the solution precipitated by 100 c.c. of 
decinormal silver, plus a few drops extra to be sure of an excess of silver, the precipitate 
collected, washed, and chlorinated to determine the bromine. The modus ojperandi was 
exactly that explained in the introduction as being the one finally adopted for the water- 
samples, except in the case of No. 4, when the haloid-precipitate, instead of being 
worked by decantation, was all collected on a filter and the chlorination effected in a 
bulb-tube instead of in a boat within a tube. 
No. of Experiment. 
Bromine Taken. 
Bromine Found. 
Error. 
mgr ms. 
mgr ms. 
mgrms. 
4 
63-63 
63-68 
+ 0 - 05 
6 
63-54 
62-50 
-1-04 
15 
63-92 
63 78 
-0-14 
16 
63-88 
62-72 
-1-16 
17 
63-86 
62-52 
-1-34 
18 
63-86 
63-40 
-0-46 
Mean error, 
-0-68 
The weighing error corresponding to this mean = -f 0'37 mgrm., which I suspect is 
owing to the fact that chloride of silver, when fused in chlorine, retains a slight excess of 
halogen, which is not expelled by a current of air of short duration such as we used to apply 
at the end. The mere inconstancy in the weight of the apparatus, according to my judgment, 
cannot have been more than ±0‘2 mgrm. at the most, and besides it would not always have 
affected the result in the same direction. Admitting that the method is subject to an 
inherent negative error, this error (excluding No. 4) would amount to about ^_ths of the 
bromine to be determined. But the experiments are too few to admit of such an 
interpretation. 
Attempts to Determine the Minimum of Silver required for Precipitating 
the Bromine from 1 Litre of Sea-Water. 
(1.) Preliminary Trials with Natural Sea- Water. 
In each of the following trials 1 litre of some sea- water (or sea-water mixture) was 
mixed with 100 c.c. of the decinormal silver, the precipitate separated by filtration or 
decantation and chlorinated. In general the mother-liquor was again worked up with 
