Professor Adams on the Theory ofthe Thermometer. 127 
A mercurial thermometer, in the present improved state 
of that instrument, must be constructed according to the fol- 
lowing facts. Ist. The tube in which the mercury is to 
range between its extreme points, must be divided into por- 
tions of equal capacity; since the bore of no tube is equal: 
in all its parts. Such a division may be most convenient- 
ly made by means of an instrument invented by M. Gay 
Lussac.* 2d. It must be as free as possible from air and 
vapours, by being hermetically seated-under the most fa- 
vourable circumstances. . It is not designed here to enter 
into the details of the practical construction of the ther- 
mometer. 3d. When it is wished to'render the instrument . 
very sensible, the bulb should be cylindrical or conical 5 
otherwise, the bulb may be spherical. 4th. There must be 
at least two fixed points; viz. that of mediing ice, (not freez- 
ing water) and that of boiling water. 5th. The water used 
in determining the fixed points, must be distilled, and must 
be boiled in a ‘metallic vessel.t 6th. The boiling point 
for‘water must be determined, when the barometer, after be- 
ing reduced to the level of the sea, and to the temperature of 
melting ice, stands at the height of thirty inches. If, after 
the reductions above mentioned are made, the barome=- 
ter is not at thirty iaches, the boiling point must be determin- 
ed by making an allowance at the rate of 1° Fah. for a dif- 
ference of 0.589 of an inch of barometric pressure.* 
It admits of mathematical demonstration, that thermom- 
eters constructed with the conditions above specified, are 
strictly comparable with each other. 
It is said above, that the thermometer viewed as a meas- 
urer of temperature, is, like other standards of measure- 
ment, an instrument of a conventional nature. The length 
of the English yard, was adjusted by the length of the arm 
of king Henry I. and the original metallic rod is preserved 
* Bict, Traité de Physique, Tom. I. 46—8. 
‘ 
+ Bict, I. 43. 
* The expansion and contraction of the glass in thermometers, is at least 
theoretically speaking, a cause of error in their results. But as this cause, 
of error is common to all thermometers, and affects them all in very nearly 
the same degree, and is itself small ; its practical effect will be scarcely dis- 
cernible. lt would not be difficult, were it of any use in practice, to give a 
formula for the correction of this error. 
