82 
Proceedings of the Royal Society 
be necessary for sending down a thermometer and bringing it up, 
it is impossible to bring up water from great depths in warm 
climates without sensible change of temperature. 
In the case of outflow thermometers, the delicacy of the instru- 
ment is limited by the size of the mercury drops. In the inverting 
thermometers of Negretti and Zambra an error may arise from the 
difference of volume of the mercury in the stem at the temperature 
at which it was inverted, and at that at which it is read. In 
an extreme case this may amount to as much as 0*4° F. ; it can, 
however, be allowed for. 
In Six’s instruments there is a possible error from looseness of 
the indices, in consequence of which they are apt to be shaken out 
of their places by any jarring of the line. Errors from this source 
can be avoided to a great extent by attaching the thermometer to 
the line by means of an elastic or india rubber stop. 
All the self-registering instruments are liable to error from the 
effects of pressure. The pressure inside a thermometer is never 
greater than that of the atmosphere when it was sealed up. It may, 
however, be exposed outside to a pressure of 500 or 600 atmo- 
spheres. The effect of this difference of pressure on the outside and 
inside of the glass envelope is to make it tend to collapse. The 
bulb of the thermometer is squeezed, and its volume in consequence 
diminished. The liquid which it contains is thereby forced into 
the stem, and its apparent volume is greater than it would have 
been had there been no excess of pressure on the outside of the 
instrument. The temperature of the instrument is measured by 
the apparent volume of the liquid which it contains ; hence the effect 
of pressure is to raise the observed -temperature above the true 
temperature. Parrot and Lenz,* in 1832, made a series of experi- 
ments on the effect of pressure on thermometers. They experi- 
mented at pressures up to 100 atmospheres, and observed differences 
between the apparent and the true temperatures of as much 
as 20° C. They found that for the same instrument the com- 
pression was simply proportional to the pressure. They used 
a thermometer as a manometer. After this date it was usual 
to attempt some kind of protection for self-registering thermo- 
meters. Those with straight stems, such as Walferdin’s mini- 
* Mem. del. Acad. Petersb. 6 C Serie ii. p. 264. 
