SURFACE WATERS OF THE NORTH ATLANTIC OCEAN. 
79 
method of checking the purity of the samples, and the accuracy of the work as it 
progressed. The following table gives the mean values of these differences ;— 
Mean diff. A 
p- 
No. of 
samples. 
^ Cl - j) s 
{fro mille). 
r. 
R. 
0 to 5 
5 
-b-17 
±•06 
±•03 
5 „ 10 
9 
+ •17 
•05 
•02 
10 „ 15 
7 
+ •13 
•04 
•01 
15 „ 20 
A 
bfc 
+ •15 
•02 
•01 
20 „ 25 
A 
-X 
+ •12 
•07 
•03 
25 „ 30 
12 
+ •13 
•06 
•02 
30 ,, 35 
183 
+ •03 
— 
— 
Over 35 
270 
•00 
•07 
•00 
The actual values of the differences correspond, of course, to the difference of 
density and chlorine ratios found, and indicate that while the values of p obtained 
from the chlorines are substantially correct within limits of error of observation for 
salinities above 30 pro mille, the values for fresher waters are too high by an amount 
rising to nearly 0‘20 in waters below 5 pro mille salinity. So far as this inquiry is 
concerned, these errors are altogether negligible, as isohalines lower than 30 y)ro mille 
rarely appear on the charts, and when they do the number of observations is never 
more than sufficient to merely indicate their position roughly ; Indeed it seems very 
unlikely that in oceanic waters, sufficiently fresh to make it worth while to take the 
differences into account, the distribution of salinity would require so great refine¬ 
ments for its elucidation. 
If we accept the mean values of A in the third column as correct, and allow for 
them accordingly, the second differences afford a measure of the comlfined errors of 
the chlorine and Sprengel determinations ; the values of r give the probable error of 
one observation of the difference A, which amounts to about 0'05. The quantity R 
is the probable error of the values of A. 
Table I. gives jDarticulars as to the samples, and the results of the various determina¬ 
tions. The first column gives the “ Laboratory Number” assigned to each sample on 
its receipt; the next the name of the vessel; the next two the date and hour of 
collection ; the next two the latitude and longitude at time of collection ; the next 
the surface temperature observed; the next the amount of chlorine estimated ; the 
next the salinity calculated, by table (35), from the preceding ; the next the 4,8^5 from 
Sprengel tube. The last column gives the amount of SO3 in grammes per kilo. 
The working charts, upon which the temperatures and salinities were plotted, were 
drawn on Mercator’s projection, to a scale of 1° long. = ‘37 inch ; each chart was 
divided into four sheets. After the plotting of the observations was completed the 
lines were drawn in in pencil and carefully revised three times, first independently, 
and then in comparison ^vith the charts for the months preceding and following. The 
