REPORT ON" THE COMPOSITION OF OCEAN- WATER. 
165 
experiments with the ordinary Jacobsen’s apparatus. The apparatus was accordingly 
given up. 
Seeing that all our results, especially those of the oxygen experiments, exhibited 
greater irregularities than could be accounted for by liberal allowance for all the sources 
of error which I could think of as possible, I concluded that there must be something 
fundamentally wrong in the method itself independently of our mode of executing it, and 
I made an attempt to determine the coefficient of absorption of oxygen synthetically in 
the following manner : — A round-bottomed flask of some 800 c.c. capacity was provided 
with a quill-sized neck and Geissler stop-cock, and its exact weight and capacity 
determined by means of a fine balance. This flask was charged with water, which was 
boiled until the air might be assumed to have been expelled, the stop-cock turned off, the 
apparatus allowed to cool, and weighed to ascertain the weight of the water, for its own sake 
and as a necessary datum for the subsequent calculation of the empty space in the flask. 
The necessary supply of oxygen was contained in a graduated (J -shaped tube, connected 
with a moveable mercury reservoir so as to render it possible to keep the pressure at exactly 
one atmosphere, and immersed in a large water-bath maintaining a constant temperature. 
To make a determination the flask was totally immersed in a large water-bath of the 
proper temperature, connected with the oxygen reservoir by means of capillary tubing, 
the stop-cocks turned open, and absorptiometric equilibrium presumably established by 
shaking the flask constantly, while an assistant kept the pressure inside at exactly that of 
the atmosphere. After the necessary (obvious) readings, the temperature in the bath was 
raised or lowered so many degrees, and another set of readings at that temperature taken, 
and so on. The results were very discouraging, falling far below what by any possibility 
could be admitted to be the values of yS : sought. My explanation of this was that a 
given mass of water to be saturated with oxygen must be shaken violently with the gas 
(which, with the apparatus adopted, was impossible), and that the data for calculating the 
volume of the residual gas in the empty part of the flask were too uncertain. This latter 
source of uncertainty might have been removed by a modification in the apparatus, which 
will readily suggest itself to everybody ; but both Mr. Lennox and myself found it 
necessary to defer the continuation of the investigation for a time. 
Before reporting on the manner in which it was subsequently resumed, I propose 
first to utilise our final formulae for n u (3 U and /3 2 , for reducing the numbers derived 
from the pioneering work to a few integer temperatures, and in this form place them 
before the reader. The differences of temperature which entered the calculations were, 
as a rule, less than ±2° C. 
In the following tabular statements, the first line always gives (as “final”) the 
value demanded by the respective interpolation formula, while the succeeding lines 
report the numbers brought out by the experiments named in Column I., each result 
standing under the assumed temperature to which it is reduced. 
