RESEARCHES IN THERMOMETRY. 583 
tremity a capillary tube, which was sealed off at ¢ after the two pots had been 
filled for an hour with finely broken ice. The mercury in the pressure tube 
h, 9, f, was brought to e on a front capillary tube, ¢ being the uppermost division 
of a small scale, or to some contiguous division. Below e the instrument 
dilates considerably, so as to allow of an excess of air being introduced into the 
helix by pressure. At / there is a joint of marine h 
glue; and at g, a lateral tube gives access to the i 
mercury supply, which can be raised or lowered as 
desired, or cut off, or otherwise adjusted, by a clamp 
on the india-rubber tubing through which it flows. 
Between / and y were cemented two standard half 
metre scales constructed for me at Kew, and two 
corrected thermometers ; there were thermometers 
alsoat dandc. The bulb of thermometer 2, which 
was the subject of comparison, occupied the centre 
of figure of the helix ; it was accompanied by a sub- 
ordinate thermometer for the purpose of calculating 
the exposure correction. The openings of the two 
pots were closely screened with tin plates during 
the experiments. 
The readings were made when the principal 
mercurial thermometer was stationary, or almost so, 
Fig. 4, 
some being taking during an ascending, others during a descending phase ; the 
five ancillary thermometers and the pressure were read as nearly as possible at 
the same time. At every “reading” the exactness of contact of the mercury 
with e was secured by suddenly pinching and relaxing the india-rubber tubing 
VOL. XXIX. PART II. 7G 
