_Directions for Using._--When the apparatus is required for practical use, sufficient water must be poured into the boiler to fill it about one third, through an opening, _F_, which must be afterwards closed by the screw plug. Then apply the lighted lamp. In a short time steam will issue from _G_; and the mercury in the thermometer, kept carefully immersed, will rise rapidly until it attains a stationary point, which is the boiling temperature. The observation should now be taken and recorded with as much accuracy as possible, and the temperature of the external air must be noted at the same time by an ordinary thermometer. The water employed should be pure. Distilled water would therefore be the best. If a substance is held mechanically suspended in water, it will not affect the boiling-point. Thus, muddy water would serve equally as well as distilled water. However, as it cannot be readily ascertained that nothing is dissolved chemically when water is dirty, we are only correct when we employ pure water. =94. Precautions to ensure correct Graduation.=--Those who possess a boiling-point thermometer should satisfy themselves that it has been correctly graduated. To do this, it is advisable to verify it with the reading of a standard barometer reduced to 32 deg. F. The table of "Vapour Tension" (given at p. 62) will furnish the means of comparison. Thus, if the reduced reading of the barometer, corrected also for latitude, be 29.922, the thermometer should show 212 deg. as the boiling-point of water at the same time and place; if 29.745, the thermometer should read 211.7; and so on as per table. In this way the error of the chief point of the scale can be obtained. Other parts of the scale may be checked with a standard thermometer, by subjecting both to the same temperature, and comparing their indications. The graduations as fixed by some makers are not always to be trusted; and this essential test should be conducted with the utmost nicety and care. Admiral FitzRoy writes, in his _Notes on Meteorology_:--"Each degree of the boiling-point thermometer is equivalent to about 550 _feet of ascent_, or one-tenth to 55 feet; therefore, the smallest error in the graduation of the thermometer itself will affect the height deduced materially.