120 Dulong and Petit on the Measure of Temperatures, [FEs. 
terminating it by a larger tube, we ayoid the error of the unequal 
capillary action from the different temperature of the two 
columns. ; 
_ The horizontal tube lies (all its length) upon a strong iron bar, 
M_N, in the form of a T, which is itself supported very firmly 
by its three feet on a very thick wooden table. The upper face 
of the bar was carefully smoothed, and carries two levels at night 
angles with each other, which are regulated by means of screws 
placed at the four corners of the table. 
Near each of the vertical tubes rises an iron bar, carrying a 
ring with a screw, which takes hold of the tube, and keeps it 
— in a fixed position. (Not to overcharge the figure, the bar on 
the side of the tube, A B, only is drawn. It terminates, as is 
seen, by an arch of iron, the point of which, R, is intended for a 
mark.) 
The bent tube being thus completed in all its parts, it remained 
to place the apparatus in such a way as to communicate to each 
of the two columns the requisite temperature. This was easily 
done for the column, A B, which was to be kept at zero. It was 
surrounded with a large cylinder of tin plate, cemented at the 
bottom round the iron bar, and which was filled with pounded 
ice to the height of the mercury in the tube. There was a small 
window, F, in this cylinder, which was opened to disengage a 
little the pieces of ice, in order to be able to perceive the top of 
the column of mercury at the time of observation. Accurate 
thermometers, plunged at different times imto this column, 
showed that it was always exactly at zero. 
‘The part of the apparatus, which was to contain the bath 
destined to heat the column, A’ B’, was of difficult execution. 
A box, the bottom of which was of a piece with the sides, could 
not have answered, because it could not have admitted the 
column, A’ B’. It was requisite likewise that the bar, M N, 
. should traverse this box, and that the small spaces between the 
bar and the sides of the box should be filled with an impermeable 
lute. To satisfy all these conditions, we got a cylinder of 
copper, whose bottom could be removed at pleasure. It termi- 
nates above in an edge, on which the cover is put. It has 
likewise at its bottom two opposite appendices, R R’, S 8%, 
having each the form of a horizontal semicylinder, in the inside 
of which passes the bar, MN. An exact idea of it may be 
formed by inspecting fig. 2, which represents a section of it 
made by a vertical plane, parallel to the direction of these appen- 
dices. The fcrm of the bottom is represented in fig. 3. It was 
united to the sids of the box by a great number of steel screws 
forced very tight. But this pressure not being sufficient to pre- 
vent the liquid from running out, thin slips of cord were intro- 
duced between the two metallic surfaces. 
The advantage of the appendices is to allow us to lute at a 
considerable distance from the fire. But in spite of this precauy 
