1906-7.] On the Application of a Differential Densimeter. 225 
ever, the instrument is used differentially, and the two waters are of 
almost the same salinity and density, this difficulty is obviated ; their 
coefficients of expansion will be practically identical, and consequently the 
ratio of the two densities will remain almost the same for all ordinary 
changes of temperature. Therefore, if we so arrange the apparatus that 
any two adjacent points along the tubes may assume the same temperature, 
all the necessary requirements will be fulfilled. Further, the temperature 
of one pair of adjacent points may differ within fairly wide limits from 
that of any other pair, without appreciably affecting the values obtained. 
In support of these statements the following experimental evidence may 
be adduced. 
The relative densities of two sea-waters, marked 0 V 0 2 , were determined 
by means of a pyknometer at the temperatures of 12°, 16°, and 20° C., 
and their ratios computed. The results obtained are set forth in 
Table VII. 
Table VII. — Showing Effects of Variations in Temperature upon 
the Relative Densities of Normal Waters. 
Temperature. 
0v 
e 2 . 
Ratio 8 / A 
12° C. 
8 = 1-02712 
A = 1-02945 
0-99774 
16° C. 
„ 1-02684 
it 1-02915 
6 
20° C. 
m 1-02661 
u 1-02889 
8 
Cl per litre 
19-994 grms. 
21-734 grms. 
Thus, for a difference of each degree within the limits of the tempera- 
tures named, the change in the measured value of the density would not 
be greater than 1 in 200,000. We also observe that, should the temperature 
of the laboratory differ from any standard temperature by it 8° C., the 
value obtained for the density will still be true to within zh '00005 ; in 
other words, for an extreme range of 16° C. above or below a given 
standard temperature the maximum error introduced from this source 
would amount to only ± 0001. 
In his report upon the specific gravity of ocean water, Mr Buchanan 
drew attention to the fact that for all true ocean waters the densities vary 
within the extreme limits of 1*0278 and 1-0240.* The difference between 
these two extremes is nearly twice as great as that in the case of the 
* Challenger Reports , vol. i., part 2, p. 1. 
VOL. XXVII. 
15 
