1892-93.] Dr J. Gibson on Composition of Sea- Water. 
321 
Mr Dickson further showed that the difference observed by him 
varied with the density, being zero for observations made in pure 
water. To the value for this difference he gave the form 
a = -0042 („So-l). 
Mr Dickson’s observations were made with waters collected in 
the English Channel having values for near 1028. By applying 
to the results obtained with the particular hydrometer which he 
employed, when reduced to qSq, a constant correction of +0*12, and 
using these corrected results along with chlorine determination 
made by the Volhard-Dittmar method, he obtained values for D 
agreeing very closely indeed with those obtained by me. 
His lowest value for D out of 42 determinations being 1*449 and 
his highest 1*458; the probable value for D in two sets of observa- 
tions being 1*4548 and 1*4554 respectively. 
A consideration of the bearing of these several investigations 
led me to attempt a recalculation of the values for \~X ^ 
number of typical cases taken from the “ Challenger ” Reports, 
using Dittmar’s values for the chlorine per kilo., Buchanan’s 
specific gravities (48^5.53) reduced by Dittmar’s tables to qSq 
plus Dickson’s correction of -f0*12 and my value for D, viz. 
1*4560. In almost every case the result was a correction of the 
values for \ “ X given in Dittmar’s Report, Table I., of approxi- 
mately -f 0*100. 
This correction altogether alters the apparent significance of these 
values. The large number of samples which in Dittmar’s Report 
are associated with values for x “X? ranging from, say, - *180 to 
- *030, give, after applying this correction, values for f according 
Avithin reasonable limits with the values for x determined experi- 
mentally by Dittmar. On the other hand, those samples with values 
from zero to -h 0*245 in Dittmar’s Report all show a difference be- 
tween the calculated and the experimental result which distinctly, 
and, indeed, in many cases very greatly, exceeds the limits of ob- 
servational error. I have also recalculated the values for D in a 
number of typical cases with the result that a large proportion of 
the “ Challenger ” samples give values for D approximating closely 
to that found to be characteristic of the sea-water derived from the 
Atlantic Ocean between 50° and 80° N. lat. 
18/3/95 
VOL. XX. 
X 
