SUEFACE \YATEES OF THE NORTH ATLANTIC OCEAN. 
75 
4S15. 
Sample. 
Tube. 
1 . 
2 . 
Diff. 
1803 
1 
1005T2 
1005-12 
-00 
1805 
3 
07-74 
07-74 
-00 
1809 
4 
20-15 
20-11 
H--04 
1817 
5 
20-06 
20-05 
+ -01 
1830 
6 
17-91 
17-97 
--06 
But these can scarcely be taken as representing the certainty of the method in 
dealing with large numbers of samples. On comparison with the chlorines a Sprengel 
determination was every now and then found to give a widely diflerent salinity, and 
repetition almost invariably proved the latter to be wrong. Speaking generally, 
single Sprengel determinations are much less trustworthy than single chlorines, the 
possible sources of error in obsei'vation are more numerous, and the errors arc much 
less easily detected, while the work is slower and more troublesome. It is worth 
while to note that thorough rinsinor with distilled water l^efore and after each observa- 
tion, and washing through twice with each sample before filling, is not sufficient to 
keep the tubes thoroughly clean. A transparent, apparently gelatinous, film is slowly 
deposited on the wadis of the tube, diminishing its volume and veiy slightly increasing 
its weight. It was found necessary to wash the tubes thoroughly at intervals with 
soda and strong nitric acid, drying them in a current of hot air. The following shows 
the effect of this treatment on a couple of tubes a.fter being in use some time :— 
Tube. 
No. 3. No. 4. 
Washed with distilled waiter and dried in 
hot air : after being in use some time . . 19'4069 20‘5916 
Washed wuth soda and water, and dried. . 19'4051 20’5901 
differences cpiite sufficient to aftect results unfavourably, although the presence of the 
film causing it could not he detected by examination of the tube. The occurrence of 
this deposit was unfortunately not definitely recognised until the work had reached 
its later stages, and it is possible tliat its presence may slightly aftect the ratios about 
to be discussed. 
Grouping the pairs of values according to the chlorines, and averaging, we get the 
following table ;— 
No. of 
Mean 
Mean 
4815 
Cl. 
samples. 
Cl. 
• 4 S 15 . 
from ( 1 ) below, 
17-00 to 17-99 
98 
17-640 
23-635 
23-635 
18-00 „ 18-49 
57 
18-243 
24-478 
24-475 
18-50 „ 18-99 
43 
18-734 
25-160 
25-160 
19-00 „ 19-49 
128 
19-327 
25-991 
25-986 
19'50 „ 19-99 
162 
19-675 
26-489 
26-480 
20-00 and over 
41 
20-220 
27-239 
27-245 
529 
L 2 
