1914] Sumner, et al.: Physical Conditions in San Francisco Bay il 
be readily obtained by reference to Knudsen’s ‘‘ Hydrographic 
Tables’’ (1901). 
During the first year’s work, including all of the salinity tests 
upon which our tables (12 to 18) are based, decinormal silver solutions 
were used in the titrations and Schellbach’s burettes having a capacity 
of 25 ee. were employed for both silver and sea-water. The silver 
solutions were checked, from time to time, by comparison with deci- 
normal sodium ehloride solutions, which were prepared as accurately 
as possible. This was, of course, an inadequate method of verification, 
since errors were no less likely in the one set of solutions than the 
other. It would, however, have probably revealed the occurrence of 
any accidental inaccuracy in the preparation of the silver solution, 
or of any appreciable change in the coneentration of the latter. As 
will be narrated below, however, we have referred our results, or 
some of them, to another and more exact standard. 
The titrations were performed by Messrs. Schmitt and Johnston, 
their observations being occasionally checked by the senior author 
of this report. Each sample was commonly tested four times, twice 
by each of the persons first named. Whenever a discrepancy of a 
certain magnitude was found between their respective results, the 
test was repeated by both until the figures agreed with sufficient 
precision. In twenty-five cases selected at random from the first 
year’s titration work, there were fifteen cases of complete agreement, 
while the mean difference between the two observers was 0.32 per 
cent, 1.e., 382 ten-thousandths of the mean reading for the two. 
After we had completed the titration of the first vear’s series of 
water samples we received the standard outfit adopted by the Inter- 
national Council for the Exploration of the Sea, and it is intended 
that this shall henceforth be employed in the hydrographic work of 
the ‘* Albatross.”? It has been used for all of the titrations since 
February, 1913. The mean difference, in twenty random eases, be- 
tween the figures of our two observers, obtained by this method, was 
0.14 per cent. 
A considerable number of tests have been made by us for the 
purpose of comparing the results obtained with the standard outfit 
and with the titration methods first employed by us. 
Samples of some of the silver and sodium chloride solutions which 
had been used for these earlier titrations were tested by reference to 
the ‘‘standard sea-water 
tions of some of the water specimens from our hydrographic stations 
9 
of the Central Laboratory. Likewise, por- 
