150 
On the Measure of Temperature . 
£Mat 
divided into fifths of a millimetre ( T i s inch), and the other with a vernier, 
which subdivides each of these into 20 parts : this will allow of our estimating 
hundredths ot a millimetre, inch) equal to the one hundred and twenty 
thousandth part of the length of the rods. This fraction answered, in most of 
»e experiments we conducted, to a change of temperature of about one degree- 
and as it is impossible to err by one part of the vernier, with the least practice in 
reading the divisions, we may see that each separate determination, even in the 
highest temperatures, has never been affected by an error of a single decree, 
huch a degree of exactness appeared to us to he sufficient in researches such 
as oul . s * Add to which, that to give greater sensibility to such an instru- 
ment, its size must be increased; and in that case the difficulty of establishing 
a uniform degree ot heat, would have occasioned much greater* uncertainty, as 
to the real amount of the expansion. 
The two rulers rest on four brass rollers, attached to a bar of iron. The 
whole apparatus is placed in a trough of copper, 14 drciinetres in length (55,1b' 
inch), 15 centimetres (5,91 inch) in depth, and 10 centimetres (3,94 inch) in 
breadth. In these new experiments, as in those which have been made on glass, 
w e used a fixed oil ; and we had recourse to the same means, to render the 
temperature stationary, during a period of time sufficiently long for the 
rulers to acquire the temperature of the liquid. There were also on each 
side of the vessel, arrangements of metallic vanes, which could be set in 
Sui, 8 ?)." perfectly to mix the different parts of the liquid mass, and to 
rn / ! ’ throughout, uniformity of temperature, and this without deranging the 
lulers. Finally, the copper trough was closed by a cover having four open- 
ings, in which were fixed thermometers, by which might be detected anv dif- 
ference oi temperature which might have occurred in the different parts of 
the mass. A thermometer placed horizontally between the bars, indicated the 
true temperature of the liquid. 
n'/wi! he rfi° f , 0 ; r co »P a rf*on we have taken the state of the rulers in the 
l-.atli, which had been left for several days in a room where the temperature 
id not sensibly vary. This is preferable to the employment of the freezing point, 
o ™Z,°, r^T” :, the hltter ^ * ‘™ d only under the preeaS 
is at “Tor 20oS« zero ^ Particularly when the surrounding air 
cd n 'n temperature sufficiently near that whicli teas requir- 
’ • J th f opuung a of the furnace were shut. The heated mass being, in this case 
a erable, the maximum of temperature was maintained Ion** enough to allow of 
the rulers acqumng the temperature of the oil, especially as the aSion conthm 
t . he „ surt ““ s <* “"‘act. This ionclusion tvas further coafim.ed b, 
t.ie steadiness in the indications of the vernier. ' 
to thTs^r method ia £lZ r J r0Ceeiia ? iu al1 l, f. we made according 
We prereuted us repeating timm with* many lubsLlt'cIT huUlm resultfwe” hare 
"o £ 
F* *£^££2^ Wng knmv e „ e to P us°f ihc ot“s 
glass ruler wifh B i , wipciauires. in a second series, we united a 
op:s„ n s z 3 whLh l ;i cl ' i V" **?■**• rbnfytrs 
a difficulty wind, does not occur in experiments on metals: it “thh. TkS 
of scr“s. e ’ S B,ft lus tell knotn which could only be effected by means 
effect the close union of a metallic and ^ Ure “ 
But the Xs did not lnn- r t" S c » IU P res . Si ™> ™ * vice to a heat of 3000. 
have been sufficiently wel^fixed to exetod . them after th e experiment, to 
substituted very thin Ztlnilo c , 0 a11 " us P ,Ci0r * of error. We, therefore, 
perfectly. The results given hv tho * Vei * anc [ oun( * them to answer the purpose 
determinations of the expansionVftrass tnf “ 7 co “ firmed our P rCTiom 
There will be tountUn it the meanexn ta ^ lc ’ resuIts all these experiments, 
between 0- and 
