90 
THE VOYAGE OF H.M.S. CHALLENGER. 
-ilvt-r become KujC'l i parts of chloride. Hence every [Br — Cl] grms. of loss of weight 
indicate [Br] grins, of broipine. According to Stas’s determinations, Br = 79*95 ; 
01 = 35*46; Ag=107*93. Hence Br— Cl = 44*49 ; and 
AgBr 
Br-Cl 
4*2230 : 
Br 
Br-Cl 
= 1 *7970 . 
To he able to work this method to the best advantage, I began by ascertaining the 
minimum weight of nitrate of silver which is sure to precipitate the maximum weight 
■ ■I bromine presumably present in a litre of sea-water. And here, unfortunately, I 
f. 11 into an error which rendered a whole string of subsequent analyses almost use- 
lr>s. Two test analyses of synthetically prepared mixtures agreed fairly well, and had 
apparently brought out the fact that 100 c.c. of deci-normal silver solution, i.e., about 
' tli of the volume required for the total halogen, brings down the whole of the bromine 
from one litre of sea- water. This exact proportion of precipitant accordingly was used in 
the duplicate analyses of 14 sea- water mixtures, before it was discovered by additional 
synthetical trials, that the proportion named falls considerably short of what is really 
irv. Thr r*->ults of the I l bromine-determinations thus became, of course, value- 
less in an absolute sense; but as all the work had been done in a strictly uniform 
fashion, the results still remained available to some extent for seeing whether or not the 
proportion of bromine in sea-water salts is subject to considerable variation ; and it is 
only on this account that I refer to the aeeident in tlii' memoir. In all the 14 samples 
of v. iter, tie “chlorine” had been determined by my modification of Volhard’s process, 
and the numbers for the bromine had been reduced finally to 100 parts of “chlorine.” 
The 14 resultB when thus reduced, varied from 0*280 to 0*316; mean of them all =0*292; 
probable error of the individual determination =±0*007. Hence it would appear that 
th< ratio of bromine to total halogen in sea-water is subject to considerable variation; 
but even this result I could not accept as final, the Less so as the particular modus 
operand* which had 1" ipted, when critically looked into, was found to be infected 
irregular errors which deprived the few exceptionally high results of a considerable 
j*ortion of their value as evidence. 
T iifortunately the work, through sheer want of material, could not be repeated, or 
. •. . i r unu d on -imilar lines ; there were, however, still a sufficient variety and number 
< l.-nger water -amples in my possession to enable me to prepare mixtures fairly 
■ • 'iv. of 'irfricr, medium -depth, and great drpth respectively, and I accordingly 
i up n preparing -uch mixtures and analysing them. But I had first to make 
quite sure of my analytical method. 
II ■ • f und out that it i- not pos-ible to eliminate the whole of the bromine from a 
• v. • r, without producing an inconveniently large haloid of silver precipitate, 
to cone the bromim converting it into the silver-salt,- by dis- 
