588 REGNAULT ON THE 
portion of the mercurial columns was necessarily outside the 
cover of the retort, and had not the temperature of the vapour ; 
a correction was consequently required to be made for it. This 
correction would have been easily made if the mean tempera- 
ture of the columns had been known with sufficient exactitude. 
When the stem is in the open air, the temperature of the different 
points of the column outside the retort varies according to the 
distance at which those points are from the lid. To avoid all 
sources of uncertainty, MM. Arago and Dulong made use of 
thermometers the stems of which were bent at right angles, im- 
mediately on leaving the iron tubes, and enclosed in horizontal 
glass tubes in which a current of cold water circulated, the tem- 
perature of which could be ascertained by a small thermometer. 
By means of this artifice the length of the column, the tempe- 
rature of which is uncertain, is very much diminished ; but there 
remains always a considerable portion of the stem at the place 
of bending which cannot be surrounded by water, and of which 
the temperature remains unknown. Thermometers bent in this 
manner are besides, for many other important reasons, incon- 
venient. That their bore may be exact, it is necessary that they 
have their scale only on the horizontal part of the stem; conse- 
quently the correction must be made for a very considerable 
number of degrees. The position of the fixed points is likewise 
more difficult to determine accurately in bent thermometers than 
in such as have straight stems. These considerations deter- 
mined me to employ thermometers with straight stems, so 
placed, that only a portion of the scale extended beyond the 
retort. A small thermometer, placed in the middle of the co- 
lumns which were not immersed, indicated a temperature which 
was taken as the expression of the mean temperature of the mer- 
curial columns not immersed. It remained to be ascertained 
whether the corrections made in this manner were sufficiently 
exact, or whether they ought to be modified in order to obtain 
their real value. To decide this question I made the following 
experiments :— 
A very delicate thermometer, perfectly free from air and having 
a range from 0° to 110°, had a bulb at the extremity of its stem, 
into which a portion of its mercury could be poured*. With the 
greatest care in melting ice, the freezing-point on this ther- 
* The invention of this thermometer, which has very lately been claimed by 
a M. Walferdin, is really due to Wollaston the astronomer, who first suggested 
it.— Ebr, 
