1819.] and onthe Laws of the Communication of Heat. 115 
instruments is the same for each.* Thus in an extent of more 
than 130°, the difference of the two scales which we compare 
is sufficiently small to be confounded with the errors of the 
observations.+ 
Nothing is easier than these kinds of experiments, as long as 
we do not go higher than the temperature of boiling water. But 
when we wish to prosecute this examination at higher tempera- 
tures, we experience great difficulties, depending partly upon 
our finding no longer any fixed temperature, and partly upon the 
great rapidity with which the liquid masses, in which the expe- 
Timents are made, cool down. For this reason, and several 
others which it is useless to state, we are obliged to have 
recourse to more tedious and complicated processes. Those 
which we adopted, after studying carefully all the causes 
of error which are likely to occur, appear to us to attain the 
greatest possible precision in researches of this nature. We 
dispense, however, with giving a detailed description of them 
here, because they differ very little from those which we have 
already made known in a memoir inserted in the second volume 
of the Annales de Chimie et Physique, p. 240.{ Our results, 
* In order to enable the reader to judge of the small deviations of the partial 
determinations, we shall state here some of those which were taken between — 30° 
and — 36°, 
Air thermometer corrected for 
the dilatation of the glass. 
ODO a otet plane ajak' aid « sci n/clajsia ole siete gar kOe 
Sede Weed cieedine ole pc seuss sence cia.) —)Garoe 
=o aE tw aisivs claet asia alaperpataaiale -- — 33°40 
SUBD Peveladesvctweseeaetes | '—\se ND 
= BNF OS vung Slo weedde s'seldspicsidsiscscs \— Gtk 
Mercurial thermometer. 
Serle UW eld a cle nial Scttia ts efates winds wees — 31:04 
ATTAIN a ol Sohsicenta ties STR AeE «~- — 80°59 
at SOAS) O teecer estes ops Weatdadscs) = 25°64 
Mean — 32:452 ; Mean — 32°420 
+ The very considerable number of experiments which we shall have to sfate 
in this memoir does not permit us to enter into details relative to each of them. We 
shall, therefore, satisfy ourselves with giving the results obtained in each case, sup- 
pressing the intermediate calculations which led to them. 
Of all the means indicated in this memoir for measuring the dilatation of air, 
we shall only recapitulate the following, which has been most frequently employed. 
The air is inclosed in a tube perfectly dried, placed horizontally in a bath of 
fixed oil, the temperature of which is gradually elevated. This tube terminates in 
the outside of the bath in a very fine tube, whose capacity is oniy a negligible 
fraction of the total yolume. When the air has acquired the requisite temperature, 
the fine point of the tube is shut by means of a blow-pipe, It is then withdrawn 
from the bath, and when it has recovered the temperature of the air, the, point is 
broken off under mercury. A yortion of this fluid of course enters into the tube. 
By comparing the weight of it with that of the mercury which fills the whole tube, 
we can easily determine the dilatation of the air, taking into consideration always 
the difference of pressure. 
This process requires only a slight change, when we operate on a gas different 
from air. The point of the tube must then be bent, and plunged into a capsule 
filled with mereury. While the temperature is increasing, a portion of the gas is 
driven out into the external air; but when the covling begins, the mercury makes 
H 2 
