RESEARCHES IN THERMOMETRY. 573 
vection, and the middle temperature is not improbably too low; yet the error 
will now have but little effect, because the correction for short lengths is 
comparatively small. . 
A complete investigation of the exposure correction would require to take 
into account (1) the law of conduction of a mercury-glass rod, (2) the law of 
thermal access by convection of more or less heated air at the rod’s surface, 
and (3) the law of relation between temperature and the difference between the 
volume expansions of mercury and glass. Lastly, there is the practical diffi- 
culty of ascertaining the temperature of a mercury-glass rod by means of an 
ancillary instrument placed at its edge, and the circumstance that unannealed 
glass is not likely to be isotropic. It is clear from Mouvusson’s investigation 
that, taking conduction into account, we diminish the correction; upward 
convection of warm air must act inthe same way ; and it appears very probable 
that the difference between the coefficients of expansion of mercury and glass 
increases as the temperature (or exposure) increases. These effects are 
embodied in the formula which, taking REGNAULT’s as a basis, I have deduced 
from very numerous experiments, and which therefore includes the convection 
effect. 
Ill. Zero MOVEMENTS. 
The movements in the zero of a thermometer, when the pressure upon it is 
constant, are due primarily to a difference of temperature between some given 
initial state and some state brought about thereafter. 
The bulb of a thermometer consists of glass, that is, of a mixture in various 
proportions of more fusible, less crystalline, basic silicates with less fusible, 
more crystalline acid silicates. During the operations of blowing it becomes 
richer in silica, and hence of a more crystalline nature. The crystalline por- 
tion, in all probability, takes many years to complete its separation—however 
rapid at first—from the amorphous constituents; and this separation should 
be attended with some slight contraction of volume. The mixture is also 
especially sensitive to the influence of temperature, more particularly soon after 
its manufacture; and thus it exhibits—after heating, for example—a “set,” 
the recovery from which is comparatively slow. 
Movements in the zero of a thermometer may be investigated in two ways, 
according as we make (1) time or (2) an immediate temperature disturbance 
the leading feature of our study. 
Movement of the Zero with Time. 
The ascent of the zero of a thermometer kept at the ordinary temperature 
appears to have been observed in some of the earliest instruments, and to have 
