conditions expected to prevail during the use of these thermometers or for specified stem temperatures. This use of like standards eliminates the need for many of the precautions necessary when dissimilar thermometers are compared and permits the direct comparison of the indications of similar thermometers as long as

temperatures are essentially the same for all of the thermometers under comparison.

All of these liquid-in-glass standards should be calibrated with reference to the IPTS-68 through comparisons with a standard platinum resistance thermometer.

In most instances, the determination of the liquidin-glass thermometer scale corrections is accomplished by comparing its scale indications with a known temperature from a platinum resistance thermometer or mercury-in-glass thermometer standards. This comparison is achieved by placing the standard and the thermometers under test in a series of stirred liquid comparison baths (See Sec. 5.1c). A discussion of the equipment needed, the procedure followed, additional corrections that may be necessary, and the choice of calibration points is given in the following sections.

5.1. Equipment

a. Ice Bath

An ice bath can be easily assembled and consists of a container, a siphon tube, ice and distilled water. A Dewar flask, approximately 36 cm deep and 8 cm in diameter, can serve as a container for the ice. A vessel of this type is preferable, since the melting of the ice is retarded by the insulating properties of the Dewar flask. A siphon is placed in the flask to enable excess water to be removed as the ice melts. The clear or transparent portion of commercially purchased ice, or ice made from distilled water, can be used. The ice is shaved or crushed into small chips measuring 2 to 5 mm. The flask is one-third filled with distilled water and the shaved ice is added. This mixture is compressed to form a tightly packed slush and any excess water is siphoned off. Before the bath is used, adequate time (15 to 30 min.) should be given for the mixture to reach a constant temperature throughout.

Ideally there should be as much ice in the flask as possible, with the small spaces between the chips filled with distilled water. It will be necessary periodically to add ice and to remove the excess water while the bath is being used to maintain this ideal consistency. If care is taken to prevent contamination of the ice and water, the ice point can be realized to better than 0.01 °C by this means.

b. Steam Bath

A schematic drawing of a steam bath is shown in figure 4. Steam produced from the boiler circulates within a double-walled steam jacket permitting free

pressure above atmospheric in the steam jacket. After the corrected pressure reading is obtained, the temperature of the steam can be derived from the values given in Table 2. With a barometer accurate to 0.1 mm Hg, this procedure is capable of an accuracy of 0.002 to 0.003 °C (0.004 to 0.005 °F). The Fortin type barometer will usually suffice for all but the most exacting measurements.

The steam bath can also be used as a comparison bath, with the temperature of the steam being determined at the time of test by means of a previously standardized thermometer. This method, which does not require the use of a barometer, may be preferable, particularly when a platinum resistance thermometer can be used as the standard.

At NBS two types of stirred liquid baths are used for comparison calibrations in the range -110 to 540 °C (-166 to 1004 °F). Each is equipped with a stirring unit to provide a uniform temperature throughout the medium and a controlled current to the heating coils for proper temperature regulation. A 5 to 7 cm thickness of insulation surrounding the baths and an insulated cover is provided to help minimize heat loss. Fitted into the top cover is a holder containing the thermometers. This holder can be rotated, enabling each thermometer to appear in the field of view of a vertically adjustable telescope attached to each bath.

FIGURE 5. Stirred liquid bath.

SUPPORT

The bath illustrated in figure 6 is designed for use with a medium that is solid at room temperature. A bath of this type is used at NBS for high temperature calibrations. Two coaxial cylinders are arranged in such a way as to permit the medium to circulate between the walls of the two cylinders and through the inner cylinder by means of openings at the top and bottom. The stirring propeller is situated near the bottom of the inner cylinder, leaving the majority of the space for the reentrant tubes into which the thermometers are inserted. The heater coils are wound on the outside of the outer cylinder. Both types of

FIGURE 6. Stirred high temperature bath using liquid having freezing point above room temperature.

baths are designed to shield the thermometers from direct radiation from the heating units.

For calibrations in the range 1 to 99 °C (34 to 210 °F) water is used as the bath liquid. One grade of petroleum oil is used between 100 and 200 °C (212 to 392 °F) and a second grade between 200 and 300 °C (392 and 572 °F). The oils must not have flash points below the highest test temperatures, nor be too viscous to prevent adequate stirring at the lower test temperatures.

Calibrations from 1 to -110 °C (30 to -166 °F) are made in a cryostat similar to the one described by Scott and Brickwedde [6]. The cryostat, shown in figure 7, consists of an inner Dewar flask, D, which contains the bath liquid. This flask is surrounded by liquid nitrogen contained in the outer Dewar flask, C. The rate of heat transfer between the bath liquid and the liquid nitrogen is retarded by evacuating the space

the side tube, M, which is connected to a vacuum system. Vigorous stirring of the bath liquid is maintained by the propeller, I, which circulates the liquid around the walls of the stirrer tube, P, similar to the flow of the medium in the high temperature bath. The temperature of the bath is thermostatically controlled by heater coils, J, wound on the outside of the stirrer tube. The thermometers are immersed inside the stirrer tube, thus shielding them from the heater coils.

The bath liquid is a five-component mixture containing by weight 14.5 percent of chloroform, 25.3 percent of methylene chloride, 33.4 percent of ethyl bromide, 10.4 percent of transdichloroethylene, and 16.4 percent of trichloroethylene. This mixture freezes at approximately 150 °C (-238 °F), but readily absorbs moisture and becomes cloudy at somewhat higher temperatures. For this reason calibrations are not performed in this bath below -110 °C (-166 °F).

A comparison measurement can be made at approximately 196 °C (boiling point of nitrogen) and -183 °C (boiling point of oxygen). A silvered Dewar flask with a narrow transparent vertical strip is used as a container for the liquid nitrogen or oxygen. The liquid is agitated by bubbling nitrogen or oxygen gas in the corresponding liquid through a glass tube with an outlet near the bottom of the flask.

5.2. Determination of Scale Corrections

Thermometers submitted to NBS for test are generally calibrated by starting at the lowest test point requested and advancing to the higher test points. In most cases the ice point is the lowest point. Ice is packed firmly around the thermometer which has been placed in an ice bath and immersed to the proper depth as determined by the type of thermometer. The thermometer is gently tapped before reading to prevent the sticking of a falling meniscus. Care is taken not to tap too vigorously, as this may cause the mercury to rebound to an erroneously high reading.

The thermometer scale corrections quoted in the Report of Calibration apply as long as the ice-point reading remains the same as observed during the NBS calibration. Subsequent changes in the ice-point reading of the thermometer will result from small changes in the glass of the thermometer bulb which affects its volume. The volume of the capillary also changes, but the volume of mercury contained in the stem is so small, in comparison to the amount of mercury in the bulb, that changes in the stem volume can usually be ignored. As a result, changes in the ice-point reading of the thermometer, taken after an exposure of not less than 3 days at a temperature of approximately 23 °C (73 °F), will be reflected by similar changes in readings at each point along the scale. Therefore, when the correction at the ice point is found to be higher (or lower) than that observed at the time of calibration, the other reported scale corrections will be higher

brated are not taken after each test point on the scale. For thermometers graduated below 300 °C or 600 °F, ice points are taken before the comparison calibrations and 3 days (with the thermometer at room temperature) after the last test point to assure that the bulb has recovered within acceptable limits. With thermometers graduated above 300 °C or 600 °F, an ice point is taken and it is immediately tested at the highest test point on the scale. After a rest period of 3 days at room temperature a second ice point is taken. The agreement between the first and second ice point must be within the stated accuracy limit of the thermometer, or it is deemed unsuitable for calibration and no further testing is done.

The ice point is the only test temperature taken in the manner described above. Comparison measurements are made with the temperature of the bath slowly increasing at a steady rate, which does not exceed one scale division in 3 to 10 minutes [7]. All thermometers are read using a 10 power telescope, which is attached to each calibration bath and perpendicular to the thermometers. The telescope is a necessity if errors due to parallax are to be avoided.

When a liquid-in-glass thermometer is intended to be used as a standard, an ice point should be taken after heating it at each test temperature. The adjusted scale correction is determined by adding the ice-point reading to the current correction at the corresponding test temperature. The example shown above is modified in this way.

When this thermometer is used as a standard, the ice point should again be taken after each calibration point. To obtain the actual temperature of the bath, the ice-point correction is added to the appropriate adjusted scale correction and thermometer reading. If the calibration temperature is 25 °C, the thermometer reads 24.983 °C, and the ice-point reading taken after the calibration is +0.014 °C, the actual temperature is computed as follows:

Thermometer reading
Interpolated Adjusted Scale Correction
Ice-point correction

Actual Temperature of Bath Medium

24.983 °C

.001

.014

24.968 °C

This method eliminates the need to wait 3 days for the bulb to recover and avoids the error due to the

ice-point depression (See Sec. 8.1a) that appears when a thermometer is heated. The standard can be used at any time and at any temperature with no concern about the change in bulb volume, provided the ice point is taken after it is used.

It is advisable to use two liquid-in-glass standards when calibrating, in order to detect reading errors and maintain a cross check of the standards. The calibration procedure can best be described by following a hypothetical calibration of four thermometers, T, through T..

Table 3 shows hypothetical observations taken in obtaining the corrections applicable to the thermometers at 20 °C. For simplification, all of the entries in the table reflect perfect thermometer performance with a temperature rise of 0.01 °C between each observation and no observer error.