1833.] Expansion of Metals. 133 
it was found however at the onset that the heating of the bar in dry 
air, although surrounded closely by the copper tube, was a most tedious 
process, whereas it was effected immediately by contact with the steam, 
which, condensing on the colder surface of the metal, delivered its 
latent heat, and did not issue from the vent until the whole apparatus 
had been effectually brought to the boiling point. It was therefore a 
fortunate circumstance that a leak in the inner tube, in the course of 
the first experiment, threw open a communication for the steam to the 
inner chamber : this was afterwards enlarged by piercing a hole through 
the copper, immediately in front of the steam injection pipe 6. The 
only inconvenience produced therefrom was, that a little steam escaped 
from the two ends where the bar necessarily projected under the 
microscopes. This was however obviated by packing with cotton, and 
screening the object glasses with paper. The steam issued in plentiful 
clouds from the thermometer apertures f and g. 
Having thus described the apparatus as it stood during the experi- 
ments, I must be allowed to add a few words on the capabilities of the 
several parts of it: and first, of the micrometers. The northern mi- 
croscope was immovable, bearing fine cross wires in its field, to which 
the centre of the corresponding dot on the bar was brought by the 
lateral screw of the camel K. The cross wires of the southern micro- 
scope on the contrary had a range of about a tenth of an inch, which 
it subdivided by 20 revolutions of the centesimally-divided screw-head 
into 2000 parts. The micrometer was therefore sensible to the 20000th 
part of an inch, or more rigidly, each division of the index was equal 
to 554,, inch, and the error of reading did not amount on several trials 
to more than one or two such divisions. 
Secondly. Of the thermometers. There was some difficulty in pro- 
curing good instruments with naked bulbs, and it was necessary to 
remove common ones from their metal scales to adapt them to the appa- 
ratus, and to scratch the degrees on the tubes ; many thermometers were 
broken from this and other causes. As the precaution was taken of com- 
paring their boiling points, and their indications at the general tempera- 
ture of the air, with a standard instrument, no error on this score was to 
be feared beyond the necessary difficulty of reading off to fractions of a. 
degree, where the instruments were only divided to every two degrees. 
The mean thermometric error cannot however be estimated at more than 
0.2 of a degree, which upon a range of 140 degrees will not affect the 
resulting dilatation more than ;;%aths, or about 2 in the sixth place of 
decimals. The fact is that the bar itself was a much more delicate 
measurer of the mean heat of the apparatus than the thermometers. 
