The determination of the boiling-point is more difficult. The boiling-point varies with the pressure of the atmosphere. The normal boiling temperature of water is fixed at a barometric pressure of 29.922 inches of mercury having the temperature of melting ice, in the latitude of 45, and at the level of the sea. Of course, these conditions rarely if ever co-exist; and consequently the boiling-point has to be corrected for errors, and reduced for latitude. Tables of vapor tension, as they are called, computed from accurate experiments, are used for this purpose. Regnault's tables, the most recent, are considered the best.
An approximate boiling-point is first obtained by actual experiment. A copper boiler is used, which has at its top an open cylinder two or three inches in diameter, and of sufficient length to allow a thermometer to be introduced into it, without touching the water in the boiler. The cylinder is surrounded by a second one, fixed to the top of the boiler, but not entering it, the two being about an inch apart. The outer cylinder is intended to protect the inner one from contact with the cold external air. The thermometer to be graduated is placed in the inner cylinder, and held there by a thong of India-rubber. As the vapor of the boiling water rises from the boiler into the cylinder, it envelops the thermometer, and causes the mercury to ascend. As the mercury rises, the tube is gradually lowered, so as to keep the top of the mercury just visible above the cylinder. When the mercury becomes stationary, the position of the top of the column is marked on the tube; and the boiling-point, subject to corrections for errors, is obtained.
The freezing and boiling points being determined, the scale is applied by dividing the length between the two points into a certain number of equal degrees. This operation is performed by a machine called a dividing-engine, which engraves degrees of any required width with extreme accuracy.
The scale used in the United Kingdom, in the British colonies, and in North America, is that known as Fahrenheit's. Fahrenheit was a philosophical instrument maker of Amsterdam. About the year 1724, he invented the scale with which his name is associated. The freezing point of his scale is 32 degrees, the boiling-point 212 degrees, and the intermediate space is composed of 180 degrees. This peculiar division was thus derived. The lowest cold observed in Iceland was the zero of Fahrenheit. When the thermometer stood at zero, it was calculated to contain a volume of mercury represented by the figures 11,124. When plunged into melting snow, the mercury expanded to a volume represented by 11,156; hence the intermediate space was divided into thirty-two equal portions or degrees, and thirty-two was taken as the freezing-point of water. [Footnote 246] Similarly, at the boiling-point, the quick-silver expanded to 11,336. Fahrenheit's scale is convenient in some respects. The meteorological observer is seldom troubled with negative signs, the divisions of the scale are numerous, and tenths of degrees give all the minuteness usually requisite.
[Footnote 246: Mr. Balfour Stewart has lately concluded a series of experiments at the Kew Observatory, by which he has accurately determined the freezing-point of mercury. The experiments, conducted with great care, have shown that the freezing-point of mercury, like that of water, is constant, and that it denotes a temperature of -37.93 F. The freezing-point of mercury will now be used as a third point in graduating thermometers which are intended to register extreme temperatures.]
In 1742, Celsius, a Swede, proposed zero for the freezing-point, and 100 degrees for the boiling-point, all temperatures below freezing being distinguished by the negative sign (-). This scale is known as the Centigrade. It is in use in France, Sweden, and in the south of Europe; it has the advantage of decimal notation, with the disadvantage of the negative sign.
Reaumur's scale is in use in Spain, Switzerland, and Germany. It differs from the Centigrade in this, that the freezing and boiling points are separated by 80 degrees instead of 100 degrees.
It would not be difficult to construct a scale which should combine all the advantages of Fahrenheit's and of the Centigrade. Freezing-point should be fixed at 100 degrees; and boiling-point should be fixed at as many hundred divisions or degrees above 100 degrees as might be agreed on by practical men as most convenient. The advantages of decimal notation would thus remain as in the Centigrade scale, and the minus sign would be got rid of.
And now, having applied the scale, and having exercised every precaution, can we congratulate ourselves on possessing a perfect instrument? Disheartening as it may appear, the standard instrument of to-day may not be accurate to-morrow. It is more than probable that the freezing point will become displaced. This curious phenomenon has never been satisfactorily explained. Messrs. Negretti & Zambra, in their treatise on Meteorological Instruments, (a work which abounds with information of a most interesting nature,) say, in reference to displacement of the freezing point, that "either the prolonged effect of the atmospheric pressure upon the thin glass of the bulbs of thermometers, or the gradual restoration of the equilibrium of the particles of the glass after having been greatly disturbed by the operation of boiling the mercury, seems to be the cause of the freezing-points of standard thermometers reading from a few tenths to a degree higher in the course of some years." To obviate this small error, it is the practice of the makers in question "to place the tubes aside for about six months before fixing the freezing-point, in order to give time for the glass to regain its former state of aggregation. The making of accurate thermometers is a task attended with many difficulties, the principal one being the liability of the zero or freezing-point varying constantly; so much so, that a thermometer that is perfectly correct to-day, if immersed in boiling-water, will be no longer accurate; at least, it will take some time before it again settles into its normal state. Then, again, if a thermometer is recently blown, filled, and graduated immediately, or, at least, before some months have elapsed, though every care may have been taken with the production of the instrument, it will require some correction; so that the instrument, however carefully made, should from time to time be plunged into finely-pounded ice, in order to verify the freezing-point."