CONTINUOUS ELECTRIC CALORIMETRY. 
95 
the stem. It is probable that a similar method would give the most accurate 
results in all .cases, but if the current heating is small its variation may generally 
be neglected, so that no special apparatus is required. 
(21.) Insulation of Thermometers. 
Defective insulation due to moisture condensed in the tubes is sometimes a source 
of error in accurate work at the ice-point with thermometers of high resistance, 
if the tubes are not sealed. To avoid this, the instruments may be fitted with a 
small inner tube leading to the bottom, through which dry air may be forced 
occasionally. A better plan, which I first adopted in 1893, for accurate work at 
low temperatures, is to seal the platinum leads through the glass so that the whole 
thermometer is air-tight. In this case the platinum leads may conveniently 
terminate in glass cups, and may be connected to the external leads by mercury or by 
fusible alloy,* as indicated in fig. 4. If the tubes are made of lead-glass, there is no 
a 
Fig. 4. Hermetically-sealed Thermometers. 
B, brass tube; C, compensator; D, plaster of Paris ; Cf, glass tube; M, mica discs; P, pyrometer leads. 
difficulty in fusing the four leads through the tube. The joint will even stand 
sudden exposure to high-pressure steam without cracking, if properly annealed. All 
the thermometers employed in my experiments on the temperatures of steam in the 
cylinder of a steam-engine (‘Proc. Inst. C.E.,’ November, 1897), were made in this 
manner in order to make the joint perfectly steam-tight at high pressures. The 
method cannot be applied to porcelain tube pyrometers, but in this case the employ¬ 
ment of a high resistance is out of the question for other reasons. I recommended 
hermetic sealing for the Kew thermometers (‘ Nature,’ October, 1895), but it was 
not adopted, as the instruments were of low resistance, intended primarily for high 
temperature work. With such an instrument it is easy, as the recent report of 
the Kew Observatory shows (‘ Proc. Poy. Soc.,’ November, 1900), to obtain an order 
of accuracy of a hundredth of a degree on the fundamental interval, which is all 
that is required for work at high temperatures; but it would be unreasonable to 
expect to be able to work to the thousandth of a degree, except under the best 
conditions, with the most perfect apparatus and the most skilful observers. 
* I originally employed common solder, but the fusible alloy, first suggested by Griffiths, appears to 
make a good connection and is more easily managed, though it could not be used in the steam-engine 
experiments. 
