1892-93.] Dr J. Gibson on Composition of Sea- Water. 315 
On the Chemical Composition of Sea-Water. 
By John Gibson, Ph.D. 
(Read July 3, 1893.) 
The question as to the perfect uniformity or otherwise of the 
chemical composition of sea-water is one of great difficulty and 
complexity. It is quite certain that, broadly speaking, great 
differences do not exist, so that a conclusion can only be arrived 
at by the application of methods of analysis of great refinement, 
and then only if the limits of experimental error inherent in these 
methods be satisfactorily determined. 
Thus the late Prof. Dittmar, when discussing the results of the 
analytical work in his “ Challenger ” Eeports (vol. i. p. 26), makes the 
following statement : — “ When we compare the percentages of the 
several components with the respective means, we frequently meet 
with differences which lie decidedly beyond the probable limits of 
the analytical error ; hence the variations must be owing partly to 
natural causes. Unfortunately, whatever these causes may be, they 
must in their effect on the numbers be presumed to be, to a certain 
extent, of the nature of observational errors, and to this extent 
they are in our reports inseparably entangled with the analytical 
errors.” 
The fact is that direc chemical analysis, meaning thereby the 
quantitative determination of the several constituents of sea- 
water, has hardly reached that degree of perfection which is 
requisite. Moreover, the labour involved in making one full 
analysis of sea-water is very great, and a satisfactory conclusion 
would require a very large number of such analyses to be carried 
out with every refinement. As this is beyond our reach, it becomes 
a matter of great importance to arrive at some method whereby the 
existence of slight differences in chemical composition may be 
indicated and recorded in such a manner as to admit of comparison 
with each other, even though the method does not give us directly 
a perfectly full and clear knowledge of the nature of the differ- 
ences which it indicates, but only their cumulative effect. I have 
