496 
POPULAR SCIENCE REVIEW. 
Dr. Miller, of King's College, than whom no one is better 
entitled to be heard on matters of this kind, gives the following 
as tests which a good thermometer should answer : — “ When 
immersed in melting ice, the column of mercury should indicate 
exactly 32° F. ; when suspended with its scale immersed in 
the steam of water boiling in a metal vessel (the barometer 
standing at thirty inches), the mercury should remain stationary 
at 212°. When the instrument is inverted, the mercury should 
fill the tube and fall with a metallic click, thus showing the 
perfect exclusion of air. The value of the degrees throughout 
the tube should be uniform. To ascertain this, a little cylinder 
of mercury may be detached from the column by a slight jerk, 
and on inclining the tube, it may be made to pass from one 
portion of the bore to another. If the scale be properly 
graduated, the column will occupy an equal number of degrees 
in all parts of the tube." 
Not one thermometer in fifty of those commonly sold in the 
shops will stand these simple and (if accuracy is really to be 
thought of) indispensable tests. In numerous shop -windows 
stacks of thermometers may be seen, which though all in close 
proximity, indicate the most discordant results instead of 
perfect harmony or even respectable uniformity. Dr. Miller 
adds the following cautionary remarks : — “ If a thermometer 
be graduated immediately after it has been sealed, it is liable 
to undergo a slight alteration in the fixed points of the scale, 
owing to the gradual contraction of the bulb, which does not 
attain its permanent dimensions until after the lapse of several 
months. This contraction is probably due to the pressure of the 
atmosphere. From this circumstance the freezing-point may 
become elevated from J to a degree ; and thus the graduations 
throughout the scale indicate a temperature which is higher 
than the true one by the amount of the error. In some ther- 
mometers the bulb, as Pierre has shown, does not at once 
contract to its proper dimensions ; and thus a temporary dis- 
placement of the graduation is caused every time such instru- 
ments are heated to 212°." Kegnault has found that some 
mercurial thermometers which agree in this indication at the 
freezing and boiling points, differ at intermediate positions ; 
and that these differences frequently amount to several degrees, 
a circumstance which the observer in question thinks may be 
due to the unequal expansion of different kinds of glass. 
Enough has been said to show the delicacy of these instru- 
ments, and it must be quite obvious that at such a price as 
one, two, or three shillings, it is impossible to manufacture an 
instrument with any solid claims to reliability. Instruments of 
this character will indicate relative changes, but absolute values 
of value can only be obtained from carefully made (i. e. expen- 
