It is to be understood that in the foregoing discussion of the errors of the mercury-in-glass thermometer, we have been treating of the de termination of temperatures to such a degree of precision that the final error is not to exceed (say) C. No such elaborate care is re quired, if the only object of the measurement is to determine the temperature to the nearest degree, or half-degree.
Passing now to the consideration of the ordinary thermometers that are used about the household and by amateur meteorological ob servers, it may be pointed out, first, that in the manufacture of a thermometer that is to be sold at retail for (say) 50 cents, it is not commer cially possible to engrave a special scale for each instrument. In making cheap thermom eters it is customary to stamp out the scales in large numbers and then to blow the bulb of each instrument to such a size that the scale will be as nearly as practicable adapted to the finished thermometer. This can be done, by an experienced glass-worker, with greater accu racy than might be supposed; but it is evident that no high degree of precision can be at tained in this way. The scale and the rest of the thermometer being adapted to each other as nearly as is commercially practicable, the thermometer is adjusted with respect to the scale by exposing it to some known temperature (say 70° F.) in the vicinity of the tempera tures at which it is most likely to be used and then securing it in such a position that the point on the stem to which the mercury rises comes opposite the proper mark on the scale. Such a theremometer will give readings that are not greatly in error at temperatures near the one at which it is standardized ; but at other temperatures any two such thermometers will necessarily diverge by an amount which de pends upon the judgment and skill of the work men who blew the bulbs and who endeavored to give them capacities adapted to the sizes of the degrees upon their respective graduated scales.
For further information concerning the methods that are used in precise thermometry consult Guillaume,
are most commonly employed for this purpose are air, hydrogen and nitrogen ; and thermom eters containing these several gases are re spectively called eair thermometers,* ((hydro gen thermometers" and
thermom eters.* See THERMOMETRY.
Alcohol Thermometer.—A thermometer in which the temperature is indicated by the expan sion of alcohol (instead of mercury) ; coloring matter of some kind being dissolved in the al cohol, so that the column of fluid in the stem of the instrument may be distinctly visible. Al cohol has a larger coefficient of expansion than mercury, and hence, for the same sizes of bulb and stem, the degrees are longer upon a ther mometer containing it. Alcohol can also be used at temperatures that are low enough to destroy an ordinary thermometer, by the freez ing of the mercury. No great degree of preci sion can be attained with the alcohol thermom eter, however, partly because the liquid wets the glass and thereby causes the instrument to read too low when the temperature is falling, and partly for other reasons. For the measurement of temperatures approaching the freezing point of mercury (37.8° F. below zero) the Interna tional Bureau of Weights and Measures pre fers a thermometer filled with toluene to one that is filled with alcohol; the toluene ther mometer being apparently capable of yielding much more accurate results. Owing to the fact that alcohol boils at a much lower temperature than water, the alcohol thermometer can hardly be graduated by the method given for the mer cury instrument, since exposure to a tempera ture of F. would cause the alcohol to have a vapor pressure so high that the bulb would be likely to burst. These thermometers are, therefore, graduated, most commonly, by direct comparison with a standard mercury-in glass instrument. The expansion of alcohol by heat is not strictly proportional to that of mercury and hence if the scale of the mercury thermometer is taken as the standard, the de gree marks upon the alcohol thermometer will not be spaced at uniform intervals. These spaces are in fact smaller at low temperatures than at higher ones, as will be seen by examin ing any good alcohol thermometer that is adapted for observihg a considerable range of temperature.