REPORT ON THE COMPOSITION OF OCEAN-WATER. 
3 
remarked that the above numerical results refer to surface waters exclusively; but the 
proposition concerning the ratios might have been extended a priori, and without fear of 
going far wrong, to deep sea waters, even if it had not been proved by my own analyses. 
horchliammer’s results naturally guided me, when I had to arrange my programme for 
examination of the 77 specimens of water collected by the Challenger, which were handed 
to me for “ complete analysis.” 
As the individual samples never amounted to more than about 2 litres (in many cases 
only to half that quantity), anything like a determination of the minor components was 
simply out of the question. I at once decided upon confining myself to determining, with 
high precision, the chlorine, sulphuric acid, soda, potash, lime, and magnesia, and thus 
furnishing, if nothing better, at least a useful extension of Forchhammer’s great work. 
My original programme included also the determination of the ammonia and organic 
nitrogen ; but, after having lost a considerable amount of time over attempts in this 
direction, I became convinced that these determinations would run away with an undue 
proportion of my precious material. Perhaps also they would ultimately have proved of 
little value, because, of what originally was organic nitrogen, it could not be expected that 
the wdiole, or even the greater part, had survived as such, after the long time which had 
elapsed between the collecting of the samples and their analysis. 
The waters came to me in three lots, and the mode of procedure adopted was, passing 
from component to component, to determine one of these in the whole series of samples ; 
then similarly the second, &c. Each determination, for a given sample, was made at 
least twice, often three or four times. The value ultimately adopted, however, was 
not in all cases the mean of the individual results : when I found, for instance, that two 
analyses would not agree sufficiently, a third and, as a rule, a fourth were made (which 
generally led to this, that three agreed while one differed from the rest), and then the one 
widely deviating number was allowed no influence in the calculation of the mean. I do 
not consider it necessary in this Report to give my individual determinations. I content 
myself with recording the finally adopted value, with as good estimates as I am able to 
make of the presumable errors. It would obviously be absurd, from two to four deter- 
minations, to calculate the probable error in the usual way. 
No analytical method is free from inherent (i.e., other than purely observational) 
errors ; hence, after having once fixed upon a certain modus operandi, I made it a point 
subsequently to adhere rigorously to it, even at the expense of a little extra precision, 
which might have been gained by a modification suggested during the progress of the in- 
vestigation. Only in one case (that of the determination of the potash) did I allow myself 
to break this rule, having succeeded in improving so essentially upon the original process, 
that I should have considered it wrong to adhere to it. Hence all my results (on the 
basis of the methods to be presently described) are susceptible of subsequent experimental 
rectification. 
