REPORT ON THE COMPOSITION OF OCEAN-WATER. 
101 
Summary. 
Bromine per 100 parts of chlorine in water from : — 
I. 
II. 
III. 
IV. 
Surface. 
Medium depth. 
Deep-Sea. 
Arran. 
0-3394 
0-3414 
0-3398 
0-3403 
r 0 — 0 ■ 0 0 1 0 
±00009 
±0-0007 
±0-0006 
Deviation from the general mean, i.e., 
from '3402, 
-0-0008 
+ 0-0012 
-0-0004 
+ 0-0001 
The deviations, as we see, lie within, or do not materially exceed, even the <£ probable ” 
errors of the means as calculated from the law of frequency of error. This, after the 
irregularities observed in the test analyses (page 97), is so surprising a result that I find 
it necessary to state expressly that the above report includes all the analyses that were 
made. 
The ratio of bromine to chlorine (or total salt), as far as can be seen from our analyses, 
is the same at all depths, the weight of bromine per 100 of chlorine being always 
= 0 '3402 ±0 '0004. 
I hasten to add that this small fraction 0'0004 does not include a constant method 
error, which, according to the test analyses tabulated on page 97, would appear to lie 
somewhere about — 0'0034, so that perhaps we should come nearer the truth by adopting 
the value ±0'3436, or say ±0'344 a somewhat uncertain probable error.* 
As the weights of the first precipitates were decidedly more constant than those of the 
second, it would perhaps be better, for a mere - comparison, to take them as our guide. 
Taking the four means and referring each to 100 of chlorine, we have, — 
I. 
II. 
III. 
IV. 
(1) Original numbers, 
. 
62-99 
62-32 
62-675 
55-265 
(2) Per 100 of Chlorine, 
0-3207 
0-3224 
0-3210 
0-3205 
Mean of (I. II. IIL), . 
. -3214 
Deviation from the mean, 
-0-0007 
+ 0-0010 
-0-0004 
-0-0009 
Probable error in value (2), . 
±0-000 20 
0-000 36 
0-000 28 
0 000 30 
* The results of C. Schmidt for a water “ a ” off the Norwegian coast, and a water “b” W.N.W. from Bergen, and 
that of Thorpe and Morton for the Irish Channel, when referred to 100 of chlorine, I calculate to be, 
Schmidt. Schmidt. Thorpe and Morton. 
a. b. 
Bromine per 100 of chlorine, 0-2451 0-2600 0-3269 
Thorpe and Morton came near my number ; Hunter’s and v. Bibra’s analyses (see p. 89) are worthless. 
