70 
MR. H. N. DICKSON ON THE CIRCULATION OF THE 
A label was affixed to each, bearing a number (1 to 30), and the distmguisbing 
number of its box : these labels were secured by a thick coating of shellac varnish. 
The corks, which were carefully selected, were soaked in very hot melted paraffin 
wax, and allowed to drain for a considerable time at a temperature above melting 
point. Except in cases where the samples had only been a short tune in the bottles 
(as in, e.g., the boxes supjdied to E.M.S,. “ Teutonic ”), the corks were stirred together 
in hot water and dried between each voyage, and all defective or dirty corks replaced. 
I have not found any difficulty with this method of the nature described by 
M. Knudsen (26), nor have I any reason to suppose that the paraffin could give rise 
to error in any of the analytical determinations. The trouble caused by particles 
getting into the pipettes is entirely avoided by taking care to remove superfluous 
paraffin in the manner described, and in the absence of paraffin particles of cork 
occasionally find their way into the samples. 
As soon as the analyses of each box of samples were examined and booked for 
charting, the l)ottles were emptied, thoroughly washed in several changes of warm 
water, and allowed to drain for at least twelve hours. In view of Instruction a 
(Form A) it was thought unnecessary to attempt further drying. 
The details of working were simple. One Ijottle was filled daily in accordance with 
the instructions given in Form A (Appendix II.), and the details as to position, &:c., 
filled up in Form B in the line numbered to correspond with the number on the 
bottle. On the return of the ship to the home port the box was at once sent to 
Oxford, along with F(U’m C, containing particulars as to forwarding of a fresh box. 
III. The Determinations of Salinity, dc. 
The estimations of salinity of the samples were made, in the first instance, from 
the amount of halogen contained. My own experience of this and other metlu^ds 
led me to agree entirely with the views so strongly expressed by Professor Petterssox 
as to its superiority ; especially in dealing with large quantities of material. The 
mode of operation adopted was practically the same as that described in a paper pub¬ 
lished recently by Professor Petterssox (27), but a number of details throwing light 
on the accuracy of the work are of some importance. It is to be distinctly understood 
that no attempt lias been made to obtain the highest possible degree of accuracy of 
which the methods are capable ; the olqect was rather to work in such a way that the 
values obtained should Ije sufficiently close for the geographical purposes of the 
investigation, and that it should be possible to deal with the large quantities of 
material in the limited time at my disposal. 
The chlorine determinations were made volumetricallv with silver nitrate, usino; 
potassium chromate as index. The amount of sea-water used for each was 10 cub. 
centims., the titration being ])erformed with a silver solution of about one-fifth normal 
strength. The pipette and burette used were made by Geissler, the former being 
