148 “REPORT—1845. 
ing together these quantities, a graduation originally arbitrary becomes a 
comparable measure corresponding to the capacity of the tube. We obtain 
by this means a table of correction which gives the true volume of the tube 
corresponding to each mark. 
It is necessary, in order to obviate the parallax on reading from the surface 
of the mercury, to use a small moveable mirror (Plate IV. fig. 1), which is 
placed on the opposite side of the tube. If the pupil of the eye seen through 
the tube in the mirror appears halved by the mark corresponding to the con- 
vexity of the mercury, the reading may be considered as exact. If the 
volume of the measured gas be read, as must always be the case, from the 
highest point of the convexity of the mercury, we must add to the correc- 
tions a small constant quantity deduced from the value found in the plate, 
and which may be named the fault of the convexity, the necessity for which 
will be rendered obvious by the following consideration :—If the reading of 
the volume of the mercury during the measurement of the instrument be at 
the mark a, the capacity a ab is not measured, but only the volume eged 
(fig. 2). Now on using the instrument, if we read a volume of gas at the 
same mark a, while the convexity takes the place dg 6, this volume as read 
does not correspond toe ge 6, but tothe real capacity egeb + deged. 
Hence the quantity deg ec is not measured by the reading, and must there- 
fore be added to the volumes observed, which otherwise would be too small. 
This quantity may be ascertained by an experiment, and serves for all fu- 
ture corrections. Ifa dilute solution of bichloride of mercury be placed in 
contact with the convexity, it disappears immediately, on account of the for- 
mation of a thin layer of protochloride of mercury which adheres to the glass. 
The mercury now shows the horizontal surface f6. The quantity caae is 
obviously equal to fe aa f, which may be measured directly by the divi- 
sions on the tube. Hence the quantity cd ée must be equal to2 x aaPf, 
which is the quantity that must be added to the observed volume on every 
reading. 
Another source of error may arise from air bubbles, which are apt to at- 
tach themselves to the glass during the filling of the tube, and being loosened 
when gas is admitted, render the latter impure. If these bubbles of air be 
visible to the naked eye, it is easy enough to separate them by means of a 
wire ; but the walls still remain covered with microscopic bubbles which 
cannot be removed in this way. In order, therefore, to prevent altogether 
this danger to the experiment, it is necessary to clean very carefully with 
unsized paper the walls of the tube after every experiment, and to introduce 
the mercury by means of a funnel with a long neck ending in a narrow 
opening at the lower end, and placed at the bottom of the tube. The mer- 
cury flowing from this funnel adheres to the walls of the tube, with a perfectly 
clear mirror-like surface. 
Especial care must be taken that air neither enters nor escapes during the 
combustion of the gas in the eudiometer. This evil is perfectly avoided by 
pressing the open end of the instrument, during the explosion, upon a per- 
fectly smooth sheet of caoutchouc placed under the mercury in the pneumatic 
trough. However, it is quite necessary to take care that the caoutchouc has 
not carried down with it any air, which might easily find its way into the 
eudiometer by the diminished tension of the gas. The caoutchouc is there- 
fore moistened with a solution of corrosive sublimate, and very slowly sunk 
into the mercury ; the protochloride of mercury formed between the mercury 
and the caoutchouc causes such complete adhesion as to exclude all air. 
Finally, the reading can only be made exact by using the mirror formerly 
described, and estimating the position of the level of the column of mercury 
