REPORT ON THE COMPOSITION OF OCEAN-WATER. 
1G9 
besides it was the one which in my hands had given the less irregular results ; hence, when 
I came to resume my absorption! e brie work, I at once decided upon confining myself to 
experiments with air. But the air method might be brought into a variety of forms 
other than the one I had adopted in my experiments. One might, for instance, start 
with a known quantity of water and a known quantity of air, shake them up together in 
a vessel at a constant volume, not greater than necessary but exceeding that of the water, 
and determine the ensuing changes of composition and tension in the unabsorbed residual 
gas. This method for a while seemed to me to be particularly promising, the more 
so as it obviously lends itself to successive determinations at different temperatures 
with the same quantity of water; bpt I could not see my way towards devising an 
adequate apparatus, and therefore fell back upon the original form of the method, 
from which, however, I took care to eliminate what I had been led, rightly or wrongly, to 
regard as its principal sources of error. 
When water in wdrich air is absorbed is boiled in a Jacobsen (or similar) apparatus, 
the “ vacuum ” present at the commencement is, of course, quickly lost, and at the end 
of the process the water boils under a pressure of something like one-fourth or one-third 
of an atmosphere, and at a temperature at which either of the two coefficients of absorp- 
tion may still have a very appreciable value. Hence, in successive experiments a gas- 
residue, variable in relative magnitude and in composition, may be retained by the water. 
To avoid this error, all that needs to be done is to suck out and remove the gas, as quickly 
as it is liberated, by a mercurial air-pump. This, in short, is the method which we 
adopted in the final experiments now to be reported on. The boiling-out flask (A 
PI. III.) differed from Jacobsen’s only in this, that the bulbs were replaced by a plain 
upright tube, surrounded by a cold-water jacket, and communicating by its upper end 
with a mercurial air-pump. As we had not a Geissler air-pump in the laboratory, and 
I thought that Sprengel’s apparatus (which could easily have been procured) would 
work too slowly, I designed a new form of the instrument, which Mr. Lennox had no 
difficulty in constructing. It is represented on PI. IIP, and as the drawing is made to 
scale, I can be very brief in its explanation. The pump proper is seen to consist of 
two cylindrical glass reservoirs, united by a stout |J-tube, so that the one is above the 
other, and is fixed vertically to a wooden stand. The upper reservoir, by a narrow 
side tube fused in somewhere near its lowest point, communicates with the exit end of 
the gas-tube of the modified Jacobsen’s flask, the bent capillary tube soldered on to 
its top serves to discharge the sucked-in gas into a test-tube inserted into a small mercurial 
trough. The lower bulb, by a Y-tube inserted into its neck by an india-rubber stopper, 
can be made to communicate either with a copper ball containing a supply of sufficiently 
compressed air, or with a Sprengel water-pump, as used for accelerating Situations. A 
perfect vacuum is obtained by the sucking power of the vertical column of mercury sus- 
pended in the (J-tube adding itself to that of the partial vacuum of the water-pump. 
(PHYS. CHEM. CHALL. EXP. — PART I.— 1884.) 22 
