SURFACE V/ATERS OF THE NORTH ATLANTIC OCEAN. 
73 
the end-points with the sea-water and with the solntions ; but on repeated strength 
determinations of the same silver solution with the same NaCl solution, it was found 
that the uncertainty did not exceed OH cub. centim. of the burette, giving as limits 
Gramme Cl in 1 cub. centim. . . . "007747 
„ „ „ . . . "007718 
•000020 
This satisfied me as to the cmisistency of the strength determinations, l)ut in order to 
estimate their accuracy I compared tlie mean given l)y the four solutions ("007724) 
with the mean of two gravimetric determinations obtained by weighing the AgCl 
precipitated from 30 cut), centims. of the silver solution, whicli was "007722. A 
further comparison was made by mixing a numl)er of samples together in two lots, 
A and B, and determining the chlorines by the Dittmar-Yolhaid method, which gives 
a better-defined reaction susceptiljle of “zig-zag” repetition, for comparison with the 
chromate method, with tlie following results— 
Dittmar-Volhard 
. 1. 
A. 
19-476 
B. 
19-523 
2 
19-470 
19-521 
Chromate .... 
. 1. 
19-48 
19-48 
jj .... 
2 
19-51 
19-50 
The following are eight duplicate 
solutions :— 
determinations made wfith different 
Sample. 
No. 14. 
Cl. 
Silver solution. 
No. 15. 
No. 17. 
Diff. 
592 
19-38 
— 
19-36 
--02 
594 
19-48 
— 
19-47 
--01 
597 
19-51 
— 
19-49 
—-02 
599 
19-51 
— 
19-49 
—-02 
621 
— 
18-44 
18-45 
+ •01 
624 
— 
19-67 
19-69 
+ -02 
626 
— 
19-35 
19-38 
+ •03 
630 
— 
19-52 
19-56 
+ •04 
silver 
This gives fir "02 as the probable error of any one chlorine estimation. Repeats of 
the twelve samples 1775—1779, 1787—1790, 1794, 1796, 1797, and 1800, with No. 40, 
No. 45, and No. 46 gave a probable error for each Cl = fii "024. 
These experiments, and the further experience gained in the course of the work, 
VOL. CXCVI.-A. 
L 
