Level of accuracy required for the calibration of glass mercury thermometers


Dave T

During a recent audit of our lab, the auditor noted that we did not state the level of accuracy required for the calibration of our glass mercury thermometers. According to our department SOP, the accuracy should be supplied by the manufacturer, or by the calibrations contractor if the manufacturer is unknown or if the manufacturer's numbers are not available. Most of the thermometers in our lab fall into the latter description. Here's the major problem - nowhere on the calibration certificates were there any indications as to the accuracy of the thermometers, or any pass/fail criteria. The contractor basically just "verified" the thermometers by recording the readings at different temperatures.

Here's my question(s): Would our department be in compliant with ISO9000 standards if we provided our own accuracy for the thermometers? Would it be okay for me to send out a single thermometer to be calibrated, use that as my standard, and calibrate my other thermometers against it with a blanket accuracy of, let's say 1% of their ranges. If not, is there any place or publications that address the determination of the accuracy of thermometers? Any help or suggestions would be greatly appreciated.

[This message has been edited by Dave T (edited 06 January 2000).]
Elsmar Forum Sponsor


Fully vaccinated are you?
I'm going to give Jerry a shot at this first - he's our current calibration expert.

nowhere on the calibration certificates were there any indications as to the accuracy of the thermometers, or any pass/fail criteria.
Not surprising if you did not specify that you wanted this info. Some companies charge more for 'extra' info and some do not, by the way.

...we did not state the level of accuracy required...
My only comment at this time is that this is a function of design/development to pass on. This is to say, they have to communicate to you the 'level of accuracy required' for a measurement such as a 'quality check'. From this you determine the equipment and the specifics of the calibration required which you communicate to the calibration resource (typically on the PO). Many issues can come up here.

Did they ask you anything about uncertainty?

...Would it be okay for me to send out a single thermometer to be calibrated, use that as my standard, and calibrate my other thermometers against it with a blanket accuracy of, let's say 1% of their ranges...
You can do that but they will all have to be 'common', so to speak. Such as all being the same model from the same company or such. And you will need supporting data.

Jerry Eldred

Forum Moderator
Super Moderator
My recommendation (if they were my thermometers) would be to somehow get documentation as to what their stated accuracy is. Depending on the particulars, liquid-in-glass thermometers are marketed under many different brand names. First would be to get a copy of the original manufacturers documentation. If you can't do that, you might get an Omega Engineering Temperature Handbook, Fisher Scientific (those are the two that most immediately come to mind; no intention to try to sell their products). Find a stated accuracy for your thermometers, make a copy of the page and retain that documentation. I think it important to have a document on file as to what their stated accuracy is (I will respond to your comment about creating your own specs later in my reply).

As for the certificate from the contractor. I have been a lab supervisor as a contractor doing calibrations. As far as I am concerned, they MUST state in their documentation what the tolerance limits that they calibrated your thermometers to. It may be that you have to pay extra to receive that documentation. But it is a must. My preference (and implicit in some standard requirements) is to have any or all cals done with BEFORE/AFTER data (in this case, only AFTER, as you cannot adjust liquid-in-glass thermometers). On the data sheet must be stated what measurement points are tested by the contractor, and must list the upper and lower tolerance limits. In order for a calibration (in this case, only a verification) to be valid, you must be checking to see if your thermometers are measuring within a specified set of tolerance limits. The contractor then must use standards that meet acceptable limits of uncertainty (for example a drywell(dryblock) calibrator, or other).

You mentioned that the contractor just recorded readings at different temperatures. Did the contractor list tolerance limits? And did the contractor state what measurement standard was used? Both of those are a must. If the contractor does not state tolerance limits for your thermometers, he/she cannot fairly state that they passed calibration (verification).

As for your question about establishing your own accuracy, using one as a standard..... If you don't have an established calibration program, and don't have trained, skilled people to administer it, without getting into a lot of detail, my gut feeling is that you could get yourself into trouble. You would have to basically document a whole program to be able to legitimately do that. Additionally, you would have to assure that the thermometer you use as a standard is well documented, characterized, capable of the desired accuracy, and you would need to purchase some standards (if you don't already have them). You would also minimally need to be able to place the standard and the thermometer being tested in an environment uniform and stable enough to compare them. An Ice Bath can be pretty economically produced. But you will also need to be able to create a uniform elevated temperature (preferrably at least two temperatures across the range of the thermometer in addition to the ice bath). There are a lot of variables that would need to be considered.

Please feel free to post or email further questions. In this short of a time, it is difficult to give a fair treatment.


David T.

Thanks for the replies, guys.

Marc, the auditors did not ask about the uncertainty of our equipment, but they did question our use of 'tolerance' on our calibration schedule listings. They preferred us to describe tolerance in terms of accuracy and precision. Is there a difference? I always thought that tolerance specifications of an instrument was the combination of accuracy and precision.

Jerry, I knew this wasn't going to be simple. I went ahead and got the manufacturer's specs for the ones that I could and took the rest off of the calibration schedule. As for the calibration contractors and their documentation, there were no thermometer specifications on the certs. I talked with the technician who does the calibration and he told me that without tolerances from the manufacturer, he cannot, as a contractor, set accuracy/tolerances himself, due to possible legal repercussions. Is this true?

This leads to me ask another question, if the tolerance or accuracy specifications are not available from the manufacturer, due to the age of the equipment, inadequate/no traceable identification of the instrument, etc., what actions would be appropriate short of removing the said equipment from use? If we wanted to set our own tolerances for equipment such as these, could we write in our calibation STM "calibrate at x tolerance...." and let it be? Or would it be a little more involved, like a validation of some kind?


Fully vaccinated are you?
I again want to caution I am not a calibration expert, but this is how I understand it:

Features Line Up:

Discrimination - Ability to tell things apart
Bias [per AIAG] Accuracy
Repeatability [per AIAG] Precision

For a thermometer, Accuracy is how far off the mark you are (Read 20 degrees, actual is 20.5 degrees). For Precision you have your R study.

Also see http://www.qs9000.com/ubb/Forum4/HTML/000047.html - APQP Aspects

In so far as your setting your own tolerances goes, that is typically done during design / development. You determine the measurement to be made, you determine the allowable tolerance for the measurement and then determine the instrument you need (typically there is the old factor of 10 sensitivity rule). As far as the equipment specifications, you can determine it's tolerance (+/- x degrees) during calibration. Do a study of X points and plot your results (gives you linearity).

You said: "calibrate at x tolerance...." Well, you don't calibrate to a tolerance. think of it this way: You buy a thermometer which comes with a stated 'tolerance' of + 0.05 degrees / -0.025 degrees and intervals of 0.1 degree. What does this tell you? It says that if you calibrate it along its range no measurement (actual temperature point tested) should (will?) deviate more than 0.05 above the indicated value (temperature) at that point or 0.025 degrees below the indicated value at that point. A correlation graph will show you the linearity along the range and the bias at any given point. You can also evaluate this variation from 'True Value' along the range and come up with a per cent of tolerance taken up by the variation.

The other question is whether or not the instrument is capable of the discrimination you need for the stated measurement. This is to say, if you have a dimension and you want to hold to +/- 1 degree. Using the 'factor of 10' rule, you need a thermometer which is accurate to +/- 0.1 degree.

QUIZ: Can you use (theoretically) the thermometer described herein for the measurement?

Does this help or confuse?

Jerry Eldred

Forum Moderator
Super Moderator
I'll respond to part of this issue, as Marc has pretty well covered a good portion of it already.

Having manufacturer's specs I think was most definitely a desirable step. That gives you at least some stake in the ground. So regarding those thermometers you were able to get those specs on, it leads me to the question I believe I brought up in the original reply (I'll get to that part after a ibt of rambling). If you don't have specs and the contractor doesn't have specs when they calibrate your thermometers, they must minimally state uncertainty. I have long been of the view in a situation like that, that if you have a thermometer with no specs and only a calibration certificate, it adds no value to your measurement process. It adds value when it tells you something about how well you can expect the thermometer to perform. If you have no tolerance specs for the thermometer, then you must at least have measurement uncertainty of the standards used by the contractor.

When I was a supervisor in the field, I would only allow my techs to calibrate equipment under certain conditions.

1. Measurement standards of adequate accuracy, range and resolution to give me a minimum 4:1 Test Uncertainty Ratio between specs of the measurement standard used and each parameter tested on the unit being calibrated.


2. If no specs were available, I would at least minimally require documentation somewhere either in my paperwork or accompanying the certificate stating what our measurement uncertainties were. If no tolerance limits were specified for the unit being calibrated, I would state what readings were made, and what MY measurement uncertainties were. I am very skeptical of contractor labs who "calibrate" without at least providing one of the above.


3. In RARE circumstances I would create a specification. This is one of those "don't try this at home, kids" circumstances. I have done that ONLY where my intuitive skill on the unit told me what it was capable of based on experience with similar technologies.

Back to the issue regarding the thermometers you found specs for. Having those specs, the contractor must cretify against those specs, verify the unit is within those tolerances, and must do so using written procedure, using measurement standards of at least four times greater accuracy.

The issue as to the capability of the thermometers to adequately measure your parameter is based on their calibrated tolerance compared with the spec or control limits of the product you use them on. That, I believe was covered in Marc's reply.

In my original reply, I asked about what specs the contractor had. I will elaborate on that. The contractor calibrated the thermometers to a procedure. If they were not calibrated to a procedure, that is a problem. The procedure must state how they determine pass or fail. I would recommend looking into how they determined pass or fail on calibration of these thermometers without tolerance limits documented. I have done many units that did not have tolerance limits. And in virtually every case, I documented what I did, that I received permission from the customer, and what the accuracy (uncertainty) of my measurement standards was.

Perhaps I am stirring the pot a bit. But I am curious as to by what criteria was the contractor able to certify the thermometers.

Hope I have been of some proactive assistance.



Fully vaccinated are you?
Date: Tue, 15 Feb 2000 12:58:01 EST
From: [email protected]
To: [email protected]
Subject: RE: Calibration of Glassware

Greg- I've been reading the emails from this list for months. I find the information invaluable. Until today, I haven't felt I've had anything to contribute. However, I am a glass thermometer manufacturer and I believe I can contribute a great deal on the subject James Taylor brought up with NIST Special Publication 819. Thank you for moderating this list. It is of great service to many of us.

This is a response to James Taylor.

NIST Special Publication 819 is an extremely valuable document. However, the first sentence you gave was too paraphrased. The actual statement is: "High quality liquid-in-glass thermometers require only one complete calibration in their lifetime and it is possible to avoid the usual requirement for complete recalibration of the instrument by the recalibration of a single previously calibrated temperature."

Unless the internal reference point of the thermometer is calibrated, usually the ice point, the thermometer will need to be completely re-calibrated. The NIST suggests this being done annually.

Also, the need for re-calibration is actually greater with a younger thermometer. The changes occur because of changes in bulb volume. These changes are greatest immediately after the thermometer is manufactured. As the thermometer ages you will actually see less and less change. The bulb becomes more stable with age.

The balance of the publication suggests taking ice point verifications after 3 months and then 6 months. Once the user sees stability, it can be decreased to a minimum of once each year. This can either be done in-house or by a calibration laboratory.

As a thermometer manufacturer and calibrator I would also suggest that a well made thermometer should NEVER age out of spec.

Deanne Miller Emory
Miller & Weber, Inc. ( An ISO-9001 Certified Thermometer Manufacturing
facility with ISO/IEC Accredited Calibration Laboratory)
email: [email protected]
website: www.millerweber.com
Thread starter Similar threads Forum Replies Date
C Pin Gauge Calibration using a Micrometer - Accuracy & Calibration Level Question General Measurement Device and Calibration Topics 10
I Higher level of packaging UDI-PI EU Medical Device Regulations 8
R Using RPN to Confirm Risk Reduced to an Acceptable Level Risk Management Principles and Generic Guidelines 12
qualprod Do sum of results of quality objectives should met a high level goal? ISO 9000, ISO 9001, and ISO 9004 Quality Management Systems Standards 10
D Maturity level and the way to obtain it Various Other Specifications, Standards, and related Requirements 2
T FDA level of concern vs IEC 62304 safety classification - 510(k) exempt device Other US Medical Device Regulations 13
F GM BIQS level 4 requirements Customer and Company Specific Requirements 1
K Level 1 PPAP Submission APQP and PPAP 11
H At what level (harm, hazardous situation, seq. of events, etc) is "risk" estimated? ISO 14971 - Medical Device Risk Management 12
F Site level CMRT reporting: how? Miscellaneous Environmental Standards and EMS Related Discussions 0
T NDT/NDE Level 2 Technicians Certification Oil and Gas Industry Standards and Regulations 20
T How can I determine the current level of exposure to talc in the workplace? Occupational Health & Safety Management Standards 6
M Packaging level EU Medical Device Regulations 2
B UKRP to what level should you audit Class I Technical Documentation? UK Medical Device Regulations 0
D What does a level 1 (PSW) PPAP actually promise? APQP and PPAP 19
A % of defects on the whole batch based on result from inspection under AQL Level II Inspection, Prints (Drawings), Testing, Sampling and Related Topics 6
R IATF and ASPICE level 2 Design and Development of Products and Processes 9
B ANOVA 3 level and 4 factors f and p values are not showing Using Minitab Software 0
N Level 3 PPAPs APQP and PPAP 4
J Part submission warrant for Level 1 PPAP APQP and PPAP 1
A Defining a lower ESD test level in IEC 60601 safety test IEC 60601 - Medical Electrical Equipment Safety Standards Series 5
was named killer Onsite Level II N.D.T. Training in Florida Training - Internal, External, Online and Distance Learning 0
D High level understanding of EUDAMED EU Medical Device Regulations 3
D Supplier Quality level category help - high level ISO 13485:2016 - Medical Device Quality Management Systems 6
silentmonkey Rationalising the level of effort and depth of software validation based on risk ISO 13485:2016 - Medical Device Quality Management Systems 10
Y What are different Special Inspection Level 1-4 and General spesification 1-3 ? AQL - Acceptable Quality Level 0
D Supplier Quality - How to classify a supplier level Medical Device and FDA Regulations and Standards News 10
E Received a Major finding during IATF Surveillance audit for loss of BIQS Level 3 (more than 6 SPPS in 6 months)...how should we address SYSTEMIC CA? IATF 16949 - Automotive Quality Systems Standard 11
J Level 3 KPI Excel Template Manufacturing and Related Processes 0
M Informational How to perform a clinical evaluation of medical devices – Part 2 – Level of clinical evidence and what sufficient clinical evidence means Medical Device and FDA Regulations and Standards News 9
R Bottom up approach versus system level ISO 14971 - Medical Device Risk Management 2
C Failure nets - Same level effects FMEA and Control Plans 0
H Graphical analysis of results - Confidence level bands nomenclature Gage R&R (GR&R) and MSA (Measurement Systems Analysis) 2
S Level of Clinical Evidence - MDR EU Medical Device Regulations 3
D GM BIQS level 5 requirements Customer and Company Specific Requirements 5
I Sampling processes - Who must define the AQL level? AQL - Acceptable Quality Level 9
R Identifying internal issues.. at what level? ISO 9000, ISO 9001, and ISO 9004 Quality Management Systems Standards 9
I What level of change in documentation requires re-training? ISO 9000, ISO 9001, and ISO 9004 Quality Management Systems Standards 5
M Level 3 PPAP submission of intentional defective parts APQP and PPAP 4
eule del ayre List of Level 3 PPAP requirements for automotive suppliers APQP and PPAP 20
I Document levels and approval requirements for lower level documents like work instructions, forms etc. Document Control Systems, Procedures, Forms and Templates 18
B AS9102 FAI & Lower Level Drawings - How should we perform the FAI? AS9100, IAQG, NADCAP and Aerospace related Standards and Requirements 1
Rameshwar25 What is a Control Plan at "Material Level"? IATF 16949 - Automotive Quality Systems Standard 4
N How to resolve discrepancies in Level 3 PPAP supplier dimensional reports? APQP and PPAP 11
M Creating a Plant Level Value Stream Map Process Maps, Process Mapping and Turtle Diagrams 1
K Defining Acceptance Quality Level, I need clarity on AQL 1.5, 2.5, 4.0 AQL - Acceptable Quality Level 5
P GM BIQS level 5 training? Training - Internal, External, Online and Distance Learning 2
Ron Rompen Dual level holes - Measurement method suggestions wanted General Measurement Device and Calibration Topics 9
A Level of details required for Class IIb Product Verification and Validation CE Marking (Conformité Européene) / CB Scheme 1
D System Level FMEA example wanted FMEA and Control Plans 2

Similar threads

Top Bottom