1819.] and on the Laws of the Communication of Heat. 163 
already taken place at 300°, how far the dilatation of glass is 
from being uniform. 
The same process seems capable of being employed to measure 
the expansion of iron, by enclosing the mercury m an iron ves- 
sel. But our attempts in this way not having completely 
succeeded, we had recourse to the following method. In a glass 
tube of 18 millimetres in diameter and 6 decimetres in length, 
and shut at one of its extremities, we introduced a cylindrical 
rod of iron, which was contained in the axis of the tube by four 
smail traverses of a length, almost equal to the diameter of the 
tube. After having cemented to the extremity of the tube 
another capillary tube, we filled it entirely with mercury, which 
was boiled for a sufficient time to drive out completely the air 
and humidity. By exposing it then to different temperatures, 
and determining the weight of the mereury which was driven 
out, it is easy to deduce the dilatation of the iron; for the volume 
driven out represents obviously the sum of the dilatations of the 
mercury and the metal, diminished by the dilatation of the glass. 
To make the calculation, it is necessary to know the volumes of 
these three bodies at the temperature of freezing water; but that 
of the iron is obtained by dividing its weight by its density at 
zero. We deduce in the same manner the volume of the glass 
from the quantity of mercury which fills it at zero. That of the 
mercury is obviously the difference of the first two. 
The process which we have just pointed out may be applied 
likewise to other metals, taking the precaution merely to oxidize 
their surface, to hinder the dissolving action of the mercury. It 
is obvious that the coating of oxide is so small that it can pro- 
duce no change in the result. This method succeeded very 
well with us for copper, and we regret that we had not time to 
try the other metals. When we know with precision the expan- 
sion of a solid body, we easily deduce from it that of others, by 
studying the expansion of a metallic pyrometer formed by the 
union of two rules united in an invariable manner by one of their 
extremities. In this way we measured, in a new series of expe- 
riments, the dilatations of a platinum rule by uniting it to a rule 
of copper. The dilatation of this last metal already found 
was verified by uniting a rule of copper to one of glass. 
We haye collected in the following table the results obtained 
by these different researches. It contains the mean dilatations 
of iron, copper, and platinum, taken at first between 0° and 100°, 
and then between 0° and 300°. We have not given any inter- 
mediate determination, because the sole object which we had in 
view was to assign the degrees in which the different thermome- 
tric scales differ from each other. But in order to make the 
results more evident, we have deduced from each dilatation the 
temperature which would result if we suppose the expansion of 
the body uniform. They. are the temperatures shih thermo- 
meters constructed of these bodies would indicate. 
L2 
