974 
THE VOYAGE OF H.M.S. CHALLENGER. 
one another. The percentages of soda being too largely affected by the cumulative 
error of the other determinations, had better be left out of consideration. But even if 
we do so, we often meet with fluctuations which are too great to be taken as arising 
from analytical errors, and consequently must correspond to differences in the actual 
composition. I have taken great pains in trying to explain these differences by natural 
causes, but have not been very successful. The final results of my inquiries may be 
summed up as follows: — 
“ From my analyses (which I do not pretend exhaust the subject) it would appear 
that the composition of sea water salt is independent of the latitude and longitude 
whence the sample is taken. Nor can we trace any influence of the depth from 
which the sample comes, if we confine ourselves to the ratio to one another of 
chlorine, sulphuric acid, magnesia, potash, and bromine. I emphasise the bromine 
because, while present in very small proportion, it is taken up preferably by sea 
plants/and consequently must be presumed to be more liable than any of the major 
components to at least temporary local diminution. And yet my analyses of the three 
mixtures of Challenger waters, and of the Arran water referred to, gave identical values 
for the bromine present per 100 of chlorine. But the determinations of the lime in the 
'same set of waters make it most highly probable that the proportion of this component 
increases with the depth. Referring to 100 parts of halogen calculated as chlorine, we 
find for the quantity of lime : — 
In deep-sea waters — 
Mixture III., .... 3-0307 
In surface waters — 
Mixture I., . . . . 3-0175 
Difference, .... 0-0132 
In medium depth waters — 
Mixture II., .... 3-0300 
In surface waters, .... 3-0175 
Difference, . . . 0-0125 
and either of the two differences is five to six times as great as even the absolute sum of 
the probable errors of the respective two terms. A discussion of the quantities of lime 
brought out by the 7 7 analyses had given a similar result, but exaggerated the difference 
between deep-sea on the one hand and shallow or medium depth on the other. 
“ But there can be no doubt that, if I had applied even as exact a method in the 77 
analyses as I did subsequently in the special investigation on the lime, I should have 
arrived at a greater difference than 0 - 013 between certain individual samples. 
The result under discussion received a valuable confirmation from the alkalinity 
