RESEARCHES IN THERMOMETRY. 581 
V. CoMPARISON OF THE MERCURIAL WITH THE AIR THERMOMETER. 
It has been shown in Sections I., II., and III. that every mercurial thermo- 
meter has its individual peculiarities, so that we cannot construct a table of 
comparison necessarily applicable to more than one mercurial instrument. The 
individuality referred to can be readily understood ; for PrERRE showed long 
since* that the fault lies more in the bulb than in the stem, and it is clear that 
no two bulbs can be blown of similar figure, uniform or equal thickness, and 
identical chemical composition. 
The air thermometer, which was at first tried in these investigations, con- 
sisted of a horizontal horseshoe-shaped glass reservoir, having a capacity of 
about 115 cubic centims. ; it was closed at one end, and communicated at the 
other with a capillary tube, which was bent twice at right angles, and carried a 
stopcock. The reservoir lay in a bath, shortly to be described, in which a 
constant temperature could be maintained, and the 
bulb of the mercurial thermometer stood vertically in ( 
the centre of figure of the reservoir. The drawing 
(fig. 3) shows the arrangement of this apparatus. The 
portion of the capillary tube beyond the stopcock 
communicated with an excellent mercurial air-pump. 
Now the reservoir of such an instrument, if placed 
in ice, should contain a definite proportion of normal 
dry air; and this, if heated and allowed to partially 
escape through the stopcock, should remain less in 
proportion to the rise in temperature. According to this method, the expansion 
is not measured by pressing up a column of mercury, but by pumping out the 
air which is residual at a definite temperature. 
It is unnecessary to give minute details respecting the operations conducted 
with this apparatus, because it proved unsuitable for its purpose. About 
twenty comparisons were made with thermometer 2 before finally abandoning 
the method, and the following points were elicited. 
Tn the first place, the following reduced comparisons were made :— 
Fig. 3. 
eae emote mee Therm. 2.+ 
80:94 ft 04. 115-84 
86:26 At 06 95:33 
91:42 if 09 1277 
97-88 12 08 50:35 
Zero observations to connect, if possible, the above numbers to a common zero, 
* Ann. Ch. Phys. (3) v. 428, et seq. 
+ These numbers have not received PogeENnDORFY’s correction. 
