128 Proceedings of the Royal Society of Edinburgh. [Sess. 
ordinary limits, may be taken as independent of the temperature at which 
the measurements are effected. It is largely due to this fact that the 
differential densimeter offers such decided advantages over the usual and, it 
may be added, tedious pyknometer methods. In the same paper it is also 
shown that the relative densities of sea- water and distilled water vary 
appreciably with variations in the common temperature of the two waters. 
Hence the necessity for first determining the true relative density of the 
sea- water, which is subsequently to be used as a standard, at some fixed 
temperature. This necessary determination may obviously be effected 
either with a pyknometer or, if desired, equally well in a room maintained 
at the selected standard temperature, with the aid of the densimeter itself. 
In this present instance the former method was, as we have already stated, 
adopted, and that chiefly because it was desired as far as possible to 
ascertain how nearly the ordinary pyknometer and the newly proposed 
densimeter methods might agree. 
Some few of the samples of water obtained in the Bay of Biscay, the 
east coast of Spain, the Straits of Messina, and in the Red Sea were, after 
their densities had been measured by the densimeter, bottled and taken on 
to Colombo, where their densities were re-determined with a pyknometer 
at the temperature 29° C. In carrying out these pyknometer determina- 
tions the densities were found by a direct comparison, not with distilled 
water, but with the standard sea-water which had been used throughout. 
In this way the difficulty which arises in the former case owing to dis- 
similarities in the coefficients of expansion of distilled and sea-water is 
avoided. The following example will serve to illustrate our mode of 
checking the values obtained at sea with the densimeter, by means of 
pyknometer measurements carried out in the laboratory at Colombo 
Example : 
Water No. 57. (See Table II.) 
Water required to fill pyknometer at 29° C. = 7 9825 grms. 
Standard water „ „ „ „ = 7*9724 ,, 
Density of standard water at 18° C. = 1*02767 „ 
.*. at 18° C. the density of water No. 57 = 7 ^ 97^4 x 1*027 67 = 1-02898. 
The densimeter on board ship gave the density of this water as 1*02904; 
the difference between the two independent values is therefore *00006 only : 
and this may, for most purposes, be regarded as negligible. The points 
which have been dealt with in the immediately preceding paragraphs will, 
it is hoped, be rendered sufficiently clear by the data and results set forth 
in Table I. 
