SURFACE WATERS OF THE NORTH ATLANTIC OCEAN. 
77 
clearer, we may take hypothetical values of Cl and SO3, such as might be oljtained 
from the examination of one samj^le, as follows 
100 SO 3 
Cl. 
SO 3 . 
Cl. 
19-97 
-00236 
11-82 
19-97 
-00234 
11-72 
20-00 
-00236 
11-80 
20-00 
-00234 
11-70 
20-03 
-00236 
11-78 
20-03 
-00234 
11-68 
Thus we know the value of the fraction to approximately the first decimal place, 
even with a low standard of accuracy in the sulphate determinations. It would 
seem, however, that the margin allowed is unnecessarily large, involving, as it does, 
differences of about 4 milligrammes in weighing in an ordinary sea-water determina¬ 
tion. The following duplicate series would indicate that an error of 2 milligrammes, 
or about per cent., covers most of the errors likely to be made in an analysis :— 
100 SO3 
SO 3 . CL ~ 
Sample. 
1. 
11. 
1. 
11. 
2314 
-00238 
-00237 
11-88 
11-84 
2318 
235 
235 
11-85 
11-86 
2326 
235 
235 
11-84 
11-86 
2331 
241 
240 
11-82 
11-77 
2337 
242 
242 
11-89 
11-89 
2338 
244 
242 
11-85 
11-78 
An examination of some of the values actually obtained confirms this ;— 
Cl. 
No. of 
samples. 
Mean of SO 3 
ProbaWe error 
of one determination. 
15 to 17-5 
15 
11-77 
± -047 
17-5 to 20 
20 
11-75 
± -055 
Over 20 
14 
11-74 
± -036 
i.e., the probable error of any one determination of ^ is d- '^15 approximately, 
or about ^ per cent. 
The values of —obtained for the 322 samples examined were first arranged 
according to the latitude from whicli the samples were obtained, with the following 
result:— 
