130 
Proceedings of the Royal Society of Edinburgh. [Sess. 
From the above table it may be seen that the average difference 
between the densimeter and pyknometer values (neglecting No. 87, which 
was probably misplaced in the case of bottles), is ±‘00019. This may be 
taken as ±0 # 7 per cent, of the total salt content ; whence it appears that 
the extreme variation between the densest water (No. 83), and the lightest 
(No. 3), is equal to ±4 per cent. From the point of view adopted by the 
exacting physicist, it is therefore evident that unless the density determina- 
tions are refined to an accuracy of 5 decimal places, which is difficult even 
on land, the information which such determinations afford will not be of 
the highest order. We are, however, under the impression that the degree 
of accuracy attained by us with the densimeter at sea just fulfils the 
actual requirements of those who engage in the study of ocean currents 
and matters allied therewith. 
As a result of the whole of this present investigation, the conclusion 
was reached that the following distinct and desirable advantages may be 
obtained by the use of the differential densimeter : — 
1. The apparatus is simple, easily managed, and can be set up in quite a 
small cabin ; in fine weather a reading can be taken in four minutes or less. 
2. The use of many cumbersome cases of bottles is avoided, as well as the 
uncertainty attending the storing of samples for a more or less lengthy 
period. 
3. The density values obtained are independent of the temperature at 
which they are measured ; and in order that the densities sought may be 
stated for any given temperature, it is only necessary to determine once for 
all the true relative density of the standard sea-water at that temperature. 
4. In fine weather observations upon the perfectly fresh and natural 
sea-water may be carried out as often as may be desired. When rough 
weather prevails the samples of water may be temporarily stored, their 
densities measured as soon as opportunity arises, and the bottles again set 
free for future occasions. 
That the measured densities were practically independent of the 
temperature at which the observations were made, may be seen by 
referring to Table II. In the column headed “ Specific gravity,” two 
slightly different and bracketed values are give for certain waters ( e.g . 
Nos. 24, 30, and 44) ; the first of these values was obtained when the tempera- 
ture of the surrounding air was greatly different from that which pre- 
vailed during the second measurement ; between the two determinations 
the water was suitably stored. As the differences are sometimes in one 
direction and sometimes in the other, and are, on the average, equal to ±7 
in the fifth decimal place, such discrepancies are to be assigned to un- 
