THE VOYAGE OF li.M.S. CHALLENGER. 
6 $ 
Passing to Series II. (S for ordinary temperature about 1/024), we have S 0 as before; 
8 1(1 ' 00 (11 SS, 1 i 90, 1*179 ; mean, 1*18G grm.), whence for the case of greatest error 
1000 S 0 = 1000 S-0-0G. 
With the mean 8’s we have 
1000 S 0 = 1000 S-0-04G. 
In the third series the one plunger experiment I made is liable to a correction of —0*02. 
T ■ determinations made by my difference method are obviously almost independent of 
the air-correction. 
To Sum uj >. — When we wish to reduce our numbers for S to the vacuum, subtract 
*033, and the result will be right apart from a residual uncertainty of about ±0*013, or 
• most ±0*03. My numbers for the relative volumes of sea-waters at different tempera- 
n . as is easily seen, relatively independent of this correction, but no doubt their 
precision would have been a little higher if they had been deduced directly from the 
special interpolation formulae for series I. and II. 
When I planned my experiments, I overlooked a very obvious modification of the 
I ts op< randi, which would have rendered the vacuum-correction almost insignificant. 
If. before each series of Bea-water weighings, we determine the loss of the plunger L 0 for 
• ;• of Borne convenient temperature r, we have for the corrected value S 0 for T S„ 
L +S 
S o = (l-t-8)(l+AAO 
. the second factor in the denominator allows for the expansion of the glass by the 
: m pc rat ure -change At = t — r. The value /.could easily be determined once for ever. 
We will therefore keep the factor in mind and say — 
S„=S— (8-1)8, 
r- the plain S stands for the uncorrccted sp. gr., and 8, as we see, multiplies a quantity 
: th- order 0*023 about , hence, instead of determining 8 we may adopt for it some pre- 
* nnined value, -ueh as 0*0012, which would make 0*0258= —0*000 03, and this value 
adopted even for all the S’s that practically come into consideration, without 
/■eng wrong by more than about 0*000 003 at the very outside. 
1 will now proceed to give a few tables intended to facilitate reduction of specific 
. *. - of • a-water from one temperature to another, and for translating specific 
gravities into salinities (x) and vice I'ersa. 
T !• VI. gives the correction which we must apply to a specific gravity when we 
* | from pure water of /' to pure water of T° as our standard. If ,S = 1000 + a;; 
