1(10)

THERMOMETERS, LlQUID-lN-GLASS: UDC 536.5:53.088

CALIBRATION

Key words: Calibration, test method, liquid-in-glass thermometers

CONTENTS 1. SCOPE
1. SCOPE 1 The scope of this method is to function as a guideline for the
2. FIELD OF APPLICATION 1 calibration of liquid-in-glass thermometers by means of the
3. REFERENCES 1 comparison technique.
4. DEFINITIONS 2
4.1 GENERAL DEFINITIONS 2
4.2 DEFINITIONS FOR LIQUID-IN-GLASS 2. FIELD OF APPLICATION
THERMOMETERS 2
The method can be applied for the calibration of liquid-in-glass
5. SAMPLING 3
thermometers in the approximate range of -100°C to 500°C.
6. METHOD OF TEST 3
The method does not apply to minimum and maximum ther-
6.1 PRINCIPLE 3
mometers.
6.2 APPARATUS 3
6.2.1 Reference thermometer 3 For the purpose of this method it is important to distinguish
6.2.2 Thermostat 3 between the 3 types: total-immersion thermometer, partial-
6.2.2.1 Baths 3 immersion thermometer and complete-immersion ther-
6.2.2.2 Temperature calibrators 4 mometer. (See chapter 4.2 Definitions for liquid-in-glass
6.2.3 Equalising blocks 5 thermometers.)
6.3 PREPARATION OF TEST SAMPLES 5
6.3.1 Cleaning 5
6.3.2 Visual inspection 5 3. REFERENCES
6.3.3 Rejoining of separated columns 5
1. The International Temperature Scale of 1990 (ITS-90), H.
6.4 PROCEDURE 6
Preston-Thomas, Metrologia 27, pages 3-10 (1990).
6.4.1 Selection of calibration temperatures 6
6.4.2 Mounting 6 2. Standard Practise for Preparation and Use of Freezing
6.4.2.1 Immersion 6 Point Reference Baths, ASTM E 563-76.
6.4.3 Readings 6
3. Precautions in the Use of Nitrate Salt Baths, Health and
6.4.3.1 Parallax errors 6
Safety Executive, Her Majesty's Stationery Office.
6.4.3.2 Sequence of calibration tempera-
tures 7 4. International Vocabulary of Basic and General Terms in
6.4.3.3 Readings at a calibration tempera- Metrology, BIPM, IEC, ISO and OIML, 1984.
ture 7
5. Bureau International de Poids et Mesures. Proces-Ver-
6.4.4 Correction for emergent stem 7
beaux de séances du ClPM 49, 26 (1981). Version in
6.5 EXPRESSION OF RESULTS 8
English published in Metrologia 18, page 44 (1982).
6.6 UNCERTAINTY 9
6.7 TEST REPORT 10 6. Guidelines for the Expression of the Uncertainty of
Measurement in Calibrations, WECC Doc. 19-1990.
7. Kaufman, Alvin B., Temperature Calibration Baths,
Measurements & Control, April 1990, page 121.

8. Wise, Jacquelyn, Liquid-in-Glass Thermometer Calibra-

tion Service, NlST Special Publication Service 250-23,
1988, U.S. Government Printing Office, Washington, 1988.

Published by NORDTEST Tekniikantie 12 FIN-02150 Espoo FINLAND Tel + 358 9 455 4600 Fax + 358 9 455 4272
ISSN 0283-7226 Proj. 951-90 www.nordtest.org
NORDTEST METHOD NT VVS 102 2

9. Liquid-in-glass Laboratory Thermometers - Principles of 4.2 Definitions for liquid-in-glass thermometers

Design, Construction and Use, ISO 386.
10. Standard Specifications for ASTM Thermometers, ASTM
E 1.
11. Standard Test Method for Inspection and Verification of
Liquid-in-Glass Thermometers, ASTM E 77.
Auxiliary scale
A short scale including a reference temperature, such as the
ice point, to provide a means for checking the thermometer for
a change in calibration with time. This scale is added when a
suitable reference temperature is not included in the range of
the main scale.

4. DEFINITIONS
Bulb Stem Auxiliary Scale Expansion Chamber
4.1 General definitions r

The general definitions are mainly taken from [4]. I&+& Contraction Chamber Immersion Line Main Scale

Calibration Fig. 1. A solid-stem liquid-in-glass thermometer.

The set of operations which establish, under specified condi-
tions, the relationship between values indicated by a measur-
ing instrument or measuring system, or values represented by Bulb
a material measure, and the corresponding known values of a
measurand. The reservoir for the thermometer liquid.

Notes
1. The result of a calibration permits the estimation of errors
of indication of the measuring instrument, measuring sys-
tem or material measure, or the assignment of values to
marks on arbitrary scales.
2. A calibration may also determine other metrological
properties.
3. The result of a calibration may be recorded in a document,
sometimes called a calibration certificate or a calibration
report.
4. The result of a calibration is sometimes expressed as a
calibration factor, or as a series of calibration factors in the
form of a calibration curve.
Conventional true value (of a quantity)
A value of a quantity which, for a given purpose, may be
substituted for the true value.
Contraction chamber
An enlargement of the capillary bore which serves to reduce a
long length of capillary or to prevent contraction of the entire
liquid into the bulb. This chamber is introduced below the main
scale or between the main scale and the auxiliary scale.
Expansion chamber
An enlargement at the top of the capillary to provide protection
in case of overheating.
Enclosed-scale thermometer
A thermometer having a capillary stem adjacent to a separate
strip bearing the scale, both being enclosed in a protective
sheath.

Note A conventional true value is, in general, regarded

as sufficiently close to the true value for the dif-
ference to be insignificant for the given purpose.
Example Within an organisation the value assigned to a
reference standard may be taken as the conven-
tional true value of the quantity realised by the
standard.
TraceabiIity
The property of a result of a measurement whereby it can be
related to appropriate standards, generally international or
national standards, through an unbroken chain of compari-
sons.
Solid-stem thermometer
A thermometer having a thick-walled capillary stem on which
the scale is etched or marked.
Complete-immersion thermometer
A thermometer designed to indicate temperatures correctly
when the whole thermometer, including the expansion cham-
ber, is subjected to the temperatures being measured.
Partial-immersion thermometer
A thermometer designed to indicate temperatures correctly
when the bulb and a specified part of the stem are exposed to
NORDTEST METHOD NT VVS 102 3

Used with a good measuring bridge and a stable standard

resistor, uncertainties in the range of millikelvins can be ob-
tained.
A stable Pt-100 probe with a measuring bridge and a standard
resistor or with a good digital thermometer or with an ohmmeter
can be a useful reference thermometer. Uncertainties of a few
Complete
millikelvin have been reported.

Liquid-in-glass and other direct-reading thermometers of suf-

ficient accuracy and stability are convenient due to their ease
of use.

Fig, 2. Liquid-in-glass thermometer types of immersion.
the temperature being measured. The remaining part of the
stem, referred to as the emergent stem, will be at the ambient
temperature, usually different from the temperature being
measured. Such thermometers are generally marked with an
immersion line to indicate the proper depth of immersion.
6.2.2 Thermostat
The requirements for the thermostat are:
- Sufficient dimensions to contain both the thermometer and
the reference thermometer.
- Sufficient stability compared with calibration time and the
uncertainty requirements.
- Good temperature uniformity so that the temperature dif-
ference between the thermometer and the reference ther-
mometer is small compared with the uncertainty require-
ments.

Total-immersion thermometer 6.2.2.1 Baths

A thermometer designed to indicate temperatures correctly For the most demanding calibration work the stirred liquid bath
when the bulb and just the portion of the thermometer contain- is preferred. The liquid and the stirring ensure good tempera-
ing the liquid are exposed to the temperature being measured. ture uniformity in the bath and good thermal contact with the
thermometers.

Special calibration baths are commercially available. They

5. SAMPLING have the necessary depth, and the liquid to air surface is small.
Not applicable. Some general purpose baths can also be used, but the larger
the liquid to air surface area the higher is the risk of instability,
and the larger the bath the higher is usually the "from corner
to corner" temperature difference.
6. METHOD OF TEST
A bath with a large amount of liquid is usually very stable, but
6.1 Principle
it takes a long time to change the calibration temperature, while
The thermometer is placed in a thermostat together with a a small volume bath changes calibration temperature fast but
reference thermometer. The indication of the thermometer is requires a better control system to be stable.
compared with the reference temperature which is measured
with the reference thermometer. Different liquids are used depending on the temperature range:

Ethyl Alcohol -112 °C to 23 °C

Methyl Alcohol -97 °C to 23°C
6.2 Apparatus Ethylene Glycol/Water 1:1 -30 °C to 110 °C
Ice point bath 0 °C
6.2.1 Reference thermometer
Water +1 °C to 95 °C
Any type of thermometer can be used as reference ther- Vegetable oils up to approx. 200 °C
mometer provided that it covers the required temperature Silicone oils -50 °C to 300 °C
range and that its calibration is traceable to a national stan- Petroleum oils 100 °C to 300 °C
dard. Further the calibration and reading uncertainty of the Potassium Nitratel
reference thermometer must be small compared with the Sodium Nitrite, approx. 1:1 mixture 170 °C to 550 °C
desired calibration uncertainty.
The temperature range of the alcohols can be extended up-
Standard Platinum Resistance Thermometers (SPRT) fulfilling wards, but care has to be taken as the fumes are toxic and
the requirements in [1] are used for the most demanding work. explosive.
NORDTEST METHOD
Different grades of petroleum and silicone oils must be used
to cover the stated ranges. The oils must not have flash points
below the highest calibration temperature nor must they be too
viscous to prevent adequate stirring at the lowest calibration
temperature. At the high end of the temperature range, the
fumes can be rather unpleasant.
Experience with the use of vegetable oils for calibration baths
is limited. [7] gives some examples of flash points and experi-
ence gained with oils such as olive, corn, soybean, peanut,
rape seed and others.
Use of the salt bath with Potassium Nitrate and Sodium Nitrite
requires some precautions. The nitrate salt alone cannot be
made to burn or explode. It is, however, a powerful oxidizing
agent and combustible material coming into contact with mol-
ten nitrate is readily ignited and often burns with explosive
violence. The molten nitrate should therefore never come into
contact with organic materials such as oil, grease, cotton-
waste, rags, sacking, paper, wood, clothing, food and graphite.
Water introduced accidentally into molten salt baths will give
rise to a 'steam explosion'. In several instances the injuries
have proven fatal. Due to fast oxidation, aluminium or alumi-
nium alloys should not come into direct contact with molten
nitrate.
Further information about precautions and the use of salt baths
can be found in [3].
Fig. 3. Ice point.
NT VVS 102 4
At lower temperatures the liquid-in-glass thermometers may
be immersed directly in the salt, but at higher temperatures
above approx. 400 °C attack of the glass may occur. This can
be avoided by placing the thermometers in steel wells im-
mersed in the molten salt. Thin-walled steel tubes closed at
one end are suitable for this purpose. Neither should quartz
encapsulated thermometers come into contact with molten
nitrate as the quartz will be matted due to the start of a
devitrification process.
The ice point (Fig. 3) is the equilibrium between ice and air
saturated water, and it was once a defining point on the
temperature scale. Today it still plays a major role in thermo-
metry as a secondary fixed point that can be readily achieved
by almost any laboratory with a minimum of resources.
In its simplest form it consists of a dewar with a mixture of
demineralised or distilled water and crushed or shaved ice
made from demineralised or distilled water. As the ice is
melting continuously and as water has its maximum density at
4 °C, one has to make sure that there is ice and not only water
around and below the tip of the thermometer. Such a bath can
give the temperature 0 °C with an uncertainty better than
±0.01 °C. The more sophisticated ice points contain a stirring
mechanism constructed in such a way that it keeps the water
saturated with air. The uncertainty in these baths can be better
than ±0.002 °C. A description of the preparation and use of the
ice point bath can be found in [2].
6.2.2.2 Temperature calibrators
The thermostat and the reference thermometer can be com-
bined into one instrument, by different manufacturers called a
temperature calibrator or a dry well calibrator. The temperature
is maintained in a chamber in which interchangeable tempera-
ture equalising blocks can be fitted one at a time. The block
contains one or more bores for the thermometers to be cali-
brated. The probe (the temperature sensitive part) of the
reference thermometer is located in the chamber but outside
the block and it is connected to a built-in digital instrument. The
resolution of these instruments is often 0.1 °C, but 0.01 °C has
also been seen.
The stability of these calibrators is often better than 0.1 °C, but
the lack of a stirred liquid can cause problems with the tem-
perature difference between the reference thermometer and
the thermometer to be calibrated. To improve thermal contact
it is important that the bore in the block fits closely the ther-
mometer to be calibrated. The depth of the bore is limited, and
this often makes the calibrator less suitable for long liquid-in-
glass thermometers.
Further there is a time lag between the probe of the reference
thermometer and the thermometer to be calibrated. After
changing to a new temperature, the built-in instrument will
show that the temperature is stable before the temperature of
the thermometer to be calibrated is stable.
It is recommended that the user establishes a thorough base
of experience of his calibrator before using it for demanding
calibrations.
NORDTEST METHOD
6.2.3 Equalising blocks
The advantage of an equalising block in a bath is primarily its
ability to smoothe the small temperature fluctuations to a few
millikelvins. The temperature differences are usually small in
a stirred bath provided that the distances between the ther-
mometers are small. This extremely good stability should not
make one forget about other sources of error. If the equalising
block is touching a cold wall in the bath, errors greater than the
fluctuations can easily arise due to gradients. And when a
resistance thermometer (reference thermometer) is used in an
equalising block, the error due to self-heating is generally
greater than when the thermometer is used directly in the
stirred liquid and this may impair the advantage of the equalis-
ing block.
A drawback of using equalising blocks is that they make the
change to a new temperature slowly. The temperature at the
beginning of the change quickly approaches the set point, but
final stabilisation takes time.
For any equalising block as well as the bore in the block of a
temperature calibrator, it is important that the bores fit closely
both the thermometer under calibration and the reference
thermometer to ensure good thermal contact.
6.3 PREPARATION OF TEST SAMPLES
6.3.1 Cleaning
If necessary, the thermometer is cleaned with water, water and
soap or alcohol and rinsed with demineralised water in order
to avoid contamination of the liquid in the baths and of the other
calibration equipment.
6.3.2 Visual inspection
The thermometer is viewed under a microscope or magnifying
glass with 8 to 15 times magnification. It is not possible to make
a complete list of all possible defects, but below are listed some
of the most important to look for:
a) Separated columns. Try one of the methods described in
6.3.3 to rejoin the liquid.
b) Gas bubbles in the bulb or in the liquid column. Try one of
the cooling methods in 6.3.3 in order to place the gas above
the column.
c) Foreign matter in the capillary or in the bulb might tend to
make the thermometer readings erratic. If glass chips or
mercury oxide are found in the capillary or bulb, they can
cause the thermometer to give different readings at differ-
ent periods of time. Such a thermometer should be re-
jected. If the foreign matter is in the top end of the scale,
the thermometer may be calibrated to a maximum tem-
perature below the position of the foreign matter, but infor-
mation of the position and the type of the foreign matter
observed should be given in the certificate so that the user
is warned and can be observant of any change in its
position.
NT VVS 102 5
d) Broken capillary tube or broken scale can be seen in
enclosed scale thermometers. Such thermometers must
be rejected.
6.3.3 Rejoining of separated columns
Separation of the liquid column is common particularly after
shipment. With care it is often possible to rejoin the column.
Various joining methods can be tried:
a) Ligthly tap the thermometer against the palm of your hand,
a block of paper or a rubber stopper.
b) Apply centrifugal force, but not a flick, and be careful to
avoid striking anything. This can best be done by holding
the bulb alongside the thumb, protecting it with the fingers
and with the stem along the arm. Raise the arm above the
head and bring it down quickly to alongside the leg.
c) If the separated part of the column is small and the ther-
mometer has a contraction chamber, cooling can be ap-
plied to bring the separated part into the contraction cham-
ber. By ligthly tapping the thermometer, the separated part
is brought into a position where it only has contact with one
side of the contraction chamber. If the bulb is then heated,
the separated part should remain in the contraction cham-
ber and be rejoined with the rest of the liquid when it enters
the contraction chamber.
d) If none of the above methods have been successful or the
thermometer did not have a contraction chamber, another
cooling method may be tried. Cool the bulb so that all the
liquid enters the bulb leaving none in the stem. The above
methods may need to be applied to assist this. The column
may rejoin when it is warmed to room temperature. Cooling
can be done in various ways using: either the ice point,
calibration bath with alcohol, dry ice or liquid nitrogen.
Since the last two methods and perhaps the calibration
bath give temperatures below the freezing point of mercury,
special care has to be taken. Cool only the bulb and never
the stem or mercury column. Care must be taken to warm
the top of the bulb first, so that pressures in the bulb due
to expanding mercury may be relieved. The warming can
be done by rolling the thermometer between two fingers
around the top of the bulb.
e) Should none of the above methods give the desired result,
a more drastic method is to apply some heating to allow
the rejoining to occur in the expansion chamber at the top.
Great care is necessary to avoid filling the expansion
chamber to more than 2/3, otherwise a pressure large
enough to burst the bulb might build up. This method
should not be applied if it requires the thermometer to be
heated above 250 °C, and the bulb should never be heated
in an open flame.
The methods mentioned above are the most important. [8]
gives some additional methods especially for thermometers
containing organic liquid.
After rejoining, the thermometer should again be visually in-
spected for bubbles and small separations.
NORDTEST METHOD NT VVS 102 6

6.4 PROCEDURE Sometimes exceptions are made to the main rules.

6.4.1 Selection of calibration temperatures If the calibration equipment does not permit the main rules to
be followed, the temperature of the emergent stem must be
Calibration is performed at the temperatures selected by the
determined with an auxiliary thermometer and a stem tempera-
user. If the measuring range or the auxiliary scale of the
ture correction must be applied. (See Subclause 6.4.4 Correc-
thermometer includes 0 °C, the user is advised to have a
tion for emergent stem.)
calibration performed at the ice point as this point can be easily
reproduced by the user and serve as a future check point. If the user has given information about how the thermometer
is used, this measurement situation is simulated as far as
When the user requests a calibration covering the total range
possible during the calibration measurements, and corrections
of the thermometer, the number of calibration points should be
to the readings are applied where necessary. (See Subclause
chosen based on the required uncertainty between the calibra- 6.4.4 Correction for emergent stem.) The attempted depth of
tion points. In order to obtain these uncertainties, the user must
immersion is indicated on the certificate.
use the corrections in the certificate. For normal accuracy 5
calibration points or points at every 100 scale divisions cover-
ing at least 80% of the range of the thermometer are recom-
mended. For the highest accuracy, points at each 50 scale
6.4.3 Readings
divisions should be chosen [9]. Good lighting is required when reading a thermometer. Inter-
polation should be carried out between the centres of the
In [8] the recommendations are also dependent on the range.
graduation lines. Interpolation between the edges will give
For thermometers graduated below 200 °C and when an un-
erroneous results.
certainty of approximately one-half of a scale division is re-
quested, the interval between calibration points should not
exceed 100 scale divisions. If uncertainties of one or two- 6.4.3.1 Parallax errors
tenths of a scale division are desired, the calibration interval
Parallax errors will arise if the line of sight is not perpendicular
should be reduced to 40 to 50 scale divisions. For thermome-
to the liquid column. Using a binocular or a monocular with a
ters graduated above 200 °C, a 40 to 50 scale division calibra-
close-up lens placed on a heavy stand and set at the same
tion interval is required to obtain corrected temperature values
height as the top of the liquid column will ease the readings
with uncertainties of approximately one-half scale division.
and help to avoid parallax errors. The existence of parallax
And a 20 to 25 scale division calibration interval is necessary
errors can be checked as follows.
for accuracies of approximately one or two-tenths of a scale
division. In case of an enclosed-scale thermometer the graduation line
immediately above the meniscus can be seen through the
The recommendations in [11] are consistent with the above.
capillary tube. If parallax errors exist, the graduation line
seems to be broken by the capillary tube. The liquid column
on the left side of Fig. 4 has no parallax error.
6.4.2 Mounting
For solid-stem thermometers the graduation lines are placed
The thermometer is placed in the thermostat close to the probe in front of the liquid column. If the thermometer liquid is
of the reference thermometer. Several thermometers can often mercury, a reflected picture of the graduation lines can be seen
be calibrated at a time. in the mercury. By looking at the graduation line immediately
below the meniscus and adjusting the line of sight until the

6.4.2.1 Immersion
The main rule is that thermometers are immersed according
to their type, which means:
- Partial-immersion thermometers are immersed to their im-
mersion line or as otherwise indicated on the thermometer.
- Total-immersion thermometers are immersed so that just a
small length of the liquid column is left emergent from the
bath to allow reading of the position of the meniscus.
- Complete-immersion thermometers are immersed com-
pletely in the bath liquid. Readings are taken through a
window in the bath.
- Should the type of thermometer not be obvious from its
construction, it is calibrated as a total-immersion ther-
mometer, and this is indicated on the certificate. Fig. 4. Enclosed-scale thermometer.
NORDTEST METHOD
Fig. 5. Solid-stem thermometers.
graduation line covers its own reflection, the parallax is
avoided. The liquid column on the left side of Fig. 5 has no
parallax error.
6.4.3.2 Sequence of calibration temperatures
Calibration is generally performed by starting with a measure-
ment at the ice point then continuing with the other tempera-
tures starting with the lowest calibration temperature and
advancing to the higher calibration temperatures. Finally the
ice point is measured once more to give an idea of a possible
drift of readings.
6.4.3.3 Readings at a calibration temperature
When the temperature is sufficiently stable, the reference
thermometer (R) and the thermometers to be calibrated (Xi)
are read in the following order (Fig. 6) at constant time inter-
vals:
R, X1, X2 .... Xn, R', Xn' .... X2', X1', R"
The thermometers are gently tapped before reading to prevent
sticking of a falling meniscus.
Fig. 6. Sequence of readings.
NT VVS 102 7
The mean of the three readings of the reference thermometer
is the basis for the average calibration temperature. The mean
of the pairwise readings of the thermometers to be calibrated
is used as their reading at the average temperature. By taking
readings in this sequence, the risk of a slight drift in tempera-
ture is minimised.
When calibrating in the ice point, readings of the reference
thermometer are not necessary.
6.4.4 Correction for emergent stem
When it is necessary to calibrate or use a total-immersion
thermometer partially immersed or a partial-immersion
thermometer totally immersed, a correction should be made
for the difference of the temperature of the emergent stem
from that which is appropriate for the specified immersion
cond it ion.
The immersion conditions for partial-immersion thermometers
can be specified by the user or e.g. by a standard [10] and
include both the depth of immersion and in some cases the
emergent-stem temperature.
The correction c1 should be evaluated from the formula:
where
k is the apparent thermal expansion coefficient of the liquid
in the particular type of glass from which the stem is made;
for purposes of evaluation, the value of k may be con-
sidered to be dependent on the arithmetic mean of t 1 and
t;
N is the number of °C equivalent to the length of the emer-
gent-stem and is the difference between the thermometer
indication and the actual or extrapolated scale value corre-
sponding to the specified immersion level;
t is the average temperature of the N degrees of the ther-
mometer stem;
t1 is the temperature of the thermometer bulb.
When total-immersion thermometers are partially immersed,
the correction c1 is added to the indication of the thermometer
when the mean emergent-stem temperature is lower than the
bulb temperature and subtracted when it is higher.
When partial-immersion thermometers are totally immersed,
the correction c1 is subtracted from the indication of the ther-
mometer when the mean emergent-stem temperature is
higher than the specified ambient temperature of the emergent
stem and added when it is lower.
When partial-immersion thermometers are used at the pre-
scribed depth of immersion in the medium, there may be a
difference between the actual and the specified temperature
of the emergent stem. In this case a correction to the indication
c2 should be evaluated from the formula:
c2 = k * N * (ts - tf)
NORDTEST METHOD NT VVS 102 8

ruxf
where

k is the apparent thermal expansion coefficient of the liquid thermometers

in the particular type of glass from which the stem is made;
for purposes of evaluation, the value of k may be con-
sidered to be dependent on the arithmetic mean of ts and
,Faden
tf ; immersion level

N is the number of °C equivalent to the length of the emergent

stem, and is the difference between the thermometer indi-
cation and the actual or extrapolated scale value corre-
sponding to the specified immersion level; Llquld column
of thermometer 1
tf is the average emergent-stem temperature in use during
Cali bration;

ts is the specified average emergent-stem temperature. (A) (B) (C)

The corrections are considered to be correct within ±10 %.
For thermometers graduated in °C, most types of glass and
over the common temperature ranges, the following values of
k are to be used:
Mercury k = 0.00016 °C -1
Organic liquids k = 0.001 °C -1
More information about the dependence of k on temperature
and the type of glass can be found in [8] and [9].
Complete-immersion thermometers are mainly used for me-
teorological purposes i.e. indoor and outdoor air temperatures.
If they are gas-filled, there will be a significant difference
between complete and partial-immersion as a result of the
effect of the temperature on the gas pressure in the ther-
mometer. The difference in readings under the two conditions
is particularly significant at high temperatures. Corrections are
difficult to apply due to lack of knowledge of the internal
pressure. [8] gives some advice.
Before any correction can be applied, the average temperature
of the emergent column has to be determined. A traditional
method for measuring the emergent-stem temperature is to
use a Faden thermometer with the correct bulb length. Faden
thermometers have long bulbs and they are attached along-
side the stem of the thermometer and overlap the emergent
column length. The top of the Faden thermometer bulb should
be at the same height as the meniscus in the thermometer to
be calibrated. Thus, the Faden thermometer does the avera-
ging. When a Faden thermometer is used, the number N is
determined as the length in °C parallel to the bulb of the Faden
thermometer. Fig. 7 (A) and (B) show a total-immersion ther-
mometer partially immersed. The bulb of a Faden thermometer
measures the average temperature t over N degrees of the
main thermometer. Fig. 7 (B) can also be interpreted as a
partial-immersion thermometer being immersed to the correct
length, but a Faden thermometer measures the average of the
emergent-stem temperature tf, thus corrections can be made
if the temperature is different from the specified emergent-
stem temperature ts.
If the bulb of the Faden thermometer is too short or no Faden
thermometer is available, place two or three mercury-in-glass
Fig. 7. Determination of emergent column temperature.
thermometers along the stem and average their readings by
weighting them for the length along the emergent stem in order
to determine t or tf.
If for example two thermometers are used of which one covers
n1 degrees on the thermometer to be calibrated and its reading
is t1 °C, and the second thermometer covers n2 degrees and
its reading is t2, the weighted average temperature of the
emergent stem is then found as:
6.5 EXPRESSION OF RESULTS
The result is stated in a table in which all numbers are rounded
to the same number of decimals as the uncertainty d).
The table has 4 columns.
a) Temperature t (°C)
t is the conventional true value of the temperature and is
calculated as the mean of the 3 readings R of the reference
thermometer corrected for all known systematic errors.
b) Reading tf (°C)
tf is the reading of the thermometer. It is calculated as the
mean of the 2 readings Xi and Xi', and corrected for
emergent-stem temperature if required.
c) Correction tc (°C)
tc is calculated from the following equation:
tc = t - tr
When the thermometer is used after calibration, the correc-
tion tc can be added to the reading tf of the thermometer to
give the best estimate of the true temperature:
t = tr - tc
 NORDTEST METHOD NT VVS 102 9

A. Those which are evaluated by applying statistical methods

to a series of repeated determinations.

B. Those which are evaluated by other means.

The WECC guideline treats all components as random vari-

ables even the influence quantities which may affect the
Temperature Reading Correction Uncertainty measured value in a systematic way. The results of using the
°C °C °C ±°C WECC guideline are consistent with the results of using the
0.00 0.20 -0.20 0.10 BIPM recommendation.
100.57 100.25 +0.32 0.16
230.9 228.5 +2.4 1.o In the short description below, we do not distinguish between
category A and B, and as for the WECC guidelines we treat all
components as random variables.

The principle of the combination method is that the size of each

6.6 UNCERTAINTY uncertainty component s i is evaluated at a 1 sigma level.

There are at least 4 main sources of uncertainty for the total The 1 sigma values are then combined by means of the
calibration uncertainty. Each source can have several uncer- following equation:
tainty components involved.
st = --+3
a) The calibration uncertainty of the reference thermometer
as stated on its certificate. ut is then found from:

b) The uncertainty of determination of the reference tempera- ut = k * st

ture.
This is very dependent on the type of instrumentation used.
In case of a direct-reading thermometer, the components
are the reading uncertainty and the drift since last calibra-
tion.
In case of a Pt-100 with an ohmmeter, the drifts of the
Pt-100, the ohmmeter, the reading uncertainty and the
calibration uncertainty of the ohmmeter must be con-
sidered.
In case of a SPRT measured at the triple point of water or
ice point for determination of its W at the temperature, its
drift can be neglected, but contributions from the triple or
ice point, the measuring bridge and the standard resistor
have to be considered.
c) The uncertainty of the thermostat (instability and gra-
dients).
d) The reading uncertainty of the thermometer to be cali-
brated.
+-
The reading uncertainty is normally 20 % (1/5) of the
smallest graduation. In the case of coarse graduation lines,
the reading uncertainty may be greater. Use of a monocular
and good lighting may give smaller reading uncertainties.
For this Nordtest method k = 2 is used.
An example of the calculation of uncertainty is given below.
Each component is listed with an estimate of its size. The
distribution factor for each component is then estimated and
the 1 sigma value is found by dividing the component's size
with its distribution factor.
a) The calibration certificate for the reference thermometer
states the uncertainty to be ±0.02 °C. The calibration
laboratory also uses k = 2, thus the 1 sigma value becomes
±0.01 °C.
b) The reading uncertainty of the reference thermometer is
equal to its resolution 0.01 °C. The distribution is estimated
to be rectangular, thus the distribution factor is √ 3.
The history of the reference thermometer shows that the
drift between two calibrations has never exceeded 0.02 °C.
Again the distribution is estimated to be rectangular.
c) Temperature fluctuations in the thermostat have an ampli-
tude not exceeding 0.01 °C. The variations follow approxi-
mately a sine function. In such a case the distribution is
U-shaped and the distribution factor is √ 2. The type A
uncertainty is strongly related to this component and is
estimated to be included in it.

The uncertainty components are combined to a total uncer- The temperature gradients in the thermostat have a limit of
tainty as described by BIPM in [5] or WECC in [6]. 0.005 °C with a rectangular distribution.

Both descriptions assume that corrections have been made
for all known systematic errors.
The BIPM recommendation defines two categories of uncer-
tainty according to the way in which their numerical values are
estimated:
d) The graduation of the thermometer to be calibrated is
0.1 °C. Interpolation to 1/5 of the graduation (0.02 °C) is
carried out. The distribution is assumed to be rectangular.
The correction for the emergent-stem temperature was
0.2 °C assumed to be accurate within 10 %.
NORDTEST METHOD NT VVS 102 10

a) Name and address of the issuing calibration laboratory.

Component Value Factor 1 sigma
value
b) Accreditation number of the laboratory if the laboratory is
±°C ±°C
accredited by the national accreditation body.
a) Ref. therm. calibration 0.02 2 0.01
c) (Unique) identification number of the calibration certificate.
b) Reading uncertainty of 0.01 1.73 0.006
reference thermometer d) Date of receipt of the instrument or device calibrated,
date(s) on which the calibration was carried out and date
Drift of ref.therm. 0.02 1.73 0.012 of issue of the certificate.
c) Thermostat instability 0.01 1.41 0.007
e) Number of pages comprising the certificate.
Thermostat gradients 0.005 1.73 0.003
f) Statement by the accreditation body if the laboratory is
d) Reading of thermometer 0.02 1.73 0.012 accredited by the national accreditation body.
to be calibrated
g) Name(s) and signature(s) of authorized person(s).
Correction for emergent- 0.02 1 0.02
stem temperature h) Suitable identification of the client.

Root of sum of squares st= 0.029 i) Identification of the instrument or device calibrated.

j) The conditions (e.g. environmental) under which the Cali-

The uncertainty is then calculated as: brations were performed.
ut = 2 * st = ±0.059 °C
k) Identification of this and any other procedure employed
and of any deviation from the procedures.

6.7 TEST REPORT I) Statement of the traceability of the calibration results.

The Test Report called a Calibration Certificate must, if rele- m) Calibration results and the associated uncertainty of the
vant, include the following information: calibrations.