Household thermometers from reliable makers are usually correct to within 1 or 2 degrees at room temperature and below, although sometimes they are several degrees in error.

Any thermometer which has 32° F. or 0° C. on its scale may be easily tested at this point by scraping a tumbler full of clear ice, saturating this with ice-cold, pure water, and placing the thermometer bulb in this mixture until it reads as low as it will go. Clean snow saturated with water may also be used, but if the snow is left dry it may be much colder than 32° F. If the thermometer tested reads 32° F. or 0° C, it is correct at this point. If higher or lower than this, it is too high or too low by the amount of the difference observed. Such a test is reliable to a tenth of a degree if carefully made.

For other temperatures there are no tests which are quite as convenient or reliable as for the ice point. The steam point, 212° F. or 100° C, is used in the testing of thermometers in the laboratory, but the steam temperature depends on the barometric reading, which varies with the weather, and with the altitude of the place where the water is boiled. For places within 500 feet of sea level, the temperature shown by a thermometer immersed in a steam bath over briskly boiling water, or in the water itself if the same is pure, should be between 210° and 212° F., or between 99°, and 100° C. For higher altitudes the temperature will be lower, as may be seen from Table I. The temperatures given in this table are averages only and variations of 1° F. or 0°.6. C. may take place from day to day because of changes in the barometric pressure.

If a tested clinical thermometer is at hand, a fairly accurate test at about 100° F. may be made. A thermometer which is correct at the ice point and at about 100° F. will probably be correct at other temperatures. Clinical thermometers should be tested by a competent testing laboratory, such as that at the United States Bureau of Standards.