253 
by microscopes is not less than 00005 millimetres. This ac- 
curacy is however illusory at temperature 0° C, unless the mer- 
curial thermometer used in finding the expansion of the bar, 
or an accurate copy of it, is preserved for subsequent use. For 
Regnault’s observations shew that thermometers constructed of 
different kinds of glass, though in perfect accordance with one 
another near 0° C and 100°C, may differ 0°°5 C at 50°C; as 
pointed out by J. Bosscha, jun. (Archives Néerlandaises, T. Iv.; 
Poggendorff’s Annalen, Ergdnzwngsband, v., 1871, page 465). 
At 162°C, the standard temperature adopted in this country, the 
difference may amount to 0°28 C, which implies an uncertainty 
of 0:0045 millimétres in the length of a bronze yard, or nine 
times the probable error of a single microscopic comparison of 
two such bars, and ninety times the probable error of a single 
comparison of two end yard bars by touch. 
The necessity for appealing to the thermometer used in the 
original comparison of a bar, in order to find the length of the 
bar at any given temperature, may be obviated by observing 
the expansion of the bar from 0° C to 100° C, and from 0° C' to 
about 50°C, by the original thermometer. The same obser- 
vations being afterwards made with a second thermometer, even 
if of different glass, and differing in its readings from the ori- 
ginal thermometer at points intermediate between 0° C and 
100° C, we have data from which we can deduce the temperature, 
as indicated by the original thermometer, from the reading of 
the new one, and thus obtain the true length of the bar at a 
temperature considerably distant from 0° C. 
Professor MAXWELL asked what the value of Whitworth’s 
method was. 
Professor M1LuEr replied that the instrument was so delicate 
that it was only of use when most carefully handled. 
Some further conversation occurred in which Mr Ellis, Prof. 
Maxwell, and Prof. Miller took part. 
