REPORT ON THE COMPOSITION OF OCEAN-WATER. 
107 
The precipitate was washed with strong hydrochloric acid, and then once or twice 
with water, and ultimately dried for hours at 200° C. Yet the salt, when tested 
with aurine, proved decidedly, though slightly, acid. The acid could not he completely 
washed away with alcohol, so that I almost believed that in the case of chloride of 
sodium the point of aurine-neutrality corresponded to some point like NaCl + SNaOH 
where 8 is a very small number, when I found that recrystallisation from a hot solution 
by evaporation, and a subsequent slight washing with hot water, deprived the salt of its 
acid reaction. This is an instructive example of the difficulty of preparing perfectly pure 
specimens of even such a common thing as common salt. 
After the Torn0e method had thus been tested, I applied it to a large number of 
Challenger samples. In each case 250 c.c. of water were operated on, and the point 
of neutrality determined repeatedly by zig-zag titration with acid and alkali. At first 
we used decinormal liquids ; but we soon found it more convenient to use solutions 
containing [HC1] grms. and f 2 [KOH] grms. respectively per litre, so that 1 c.c. of 
each corresponded to exactly 1 mgrm. of carbonic acid. The results are tabulated in the 
chapter on the “ Alkalinity of Ocean Water.” 
The Method for the Determination of Carbonic Acid. — This method is an adaptation 
of one proposed years before by Borchert and by Classen. It consists in this : — 250 c.c. 
of the sea-water are placed in a flask surmounted by an inverted Liebig’s condenser, the 
upper end of which communicates with a set of absorption-apparatus charged with a 
known sufficient volume of standard baryta water. After addition of some sulphuric 
acid, a current of purified air is passed through the liquid in the flask, which is then 
heated to boiling and kept boiling for about 1 5 minutes, when, according to experience, 
the whole of the carbonic acid has been conveyed to the baryta. The baryta-liquors 
are united in a graduated cylinder, allowed to settle, an aliquot part of the liquor is 
decanted off clear, and titrated with standard oxalic acid. I do not describe the very 
practical absorption apparatus, because I ultimately discarded it, as will be seen in the 
sequel. 
When I made trial of this method — with solutions in 250 c.c. of pure boiled-out 
water of known weights of pure ignited carbonate of soda, varying from 2 to 50 mgrms. 
— the result in the majority of (some 20) trials proved 1 to 2 mgrms. too high. I do 
not mean to insinuate that Tornpe’s results are likely to be vitiated by similar positive 
errors ; it is quite possible that my experiments may have been infected by some constant 
error in manipulation which Tornpe avoided. Whether or no I am unable to say, because, 
while engaged in the necessary blank experiments for elucidating this point, I happened 
to hit upon a most essential improvement in the apparatus, which I thought would, and 
when tested was actually found to, eliminate the worst of the errors of the method. My 
improvement consists in this, that the gas which comes out of the condenser is passed 
