By Wu Peng, Senior Process Instrumentation Engineer · Last reviewed August 3, 2026
Calibration measures an instrument against a reference standard and documents the difference; it changes nothing on the device. Adjustment physically or digitally alters the instrument so it reads closer to true. Verification compares the calibration result against a tolerance and declares pass or fail. The three words describe three separate operations, and mixing them up costs money: buyers order the wrong service, auditors reject records, and out-of-tolerance instruments keep running because someone thought a sticker meant accuracy.
This guide uses the definitions from the International Vocabulary of Metrology (VIM, JCGM 200) and applies them to process instruments: transmitters, flow meters, sensors, and gauges. Laboratory-specific uses of the same words, such as centrifuge checks or clinical linearity studies, follow the same logic but their procedures are outside this scope.
Contents
- Three different operations
- What calibration means
- What adjustment means
- What verification means
- As-found and as-left
- Judging pass or fail
- Legal verification
- Quality system requirements
- Why the terms blur
- FAQ
Three different operations
The cleanest way to keep the terms apart is to ask two questions. Does the operation touch the instrument? And what does it produce?
Calibration touches nothing and produces data. Adjustment changes the instrument and produces a new state that must be measured again. Verification touches nothing and produces a judgment.
A fourth term, validation, asks a different question entirely: whether the requirement itself suits the intended use.
| Term | What it does | Output |
|---|---|---|
| Calibration VIM 2.39 |
Compares indications to a standard; no intervention | Certificate with errors and uncertainty |
| Adjustment VIM 3.11 |
Alters the instrument toward prescribed indications | Changed instrument; recalibration required |
| Verification VIM 2.44 |
Checks results against specified requirements | Pass or fail statement |
| Validation VIM 2.45 |
Verification that requirements fit the intended use | Fit-for-purpose statement |
VIM = International Vocabulary of Metrology, JCGM 200:2012, the definition source referenced by ISO/IEC 17025 laboratories.
What calibration means
The VIM definition (clause 2.39) describes calibration as a two-step operation. Step one establishes the relation between the values of measurement standards, with their uncertainties, and the corresponding indications of the instrument. Step two uses that relation to obtain measurement results from future indications. In plain terms: apply known inputs, record what the instrument says, and keep the comparison so readings can be corrected or trusted later.
The definition carries a note worth quoting in full, because many published guides contradict it: “Calibration should not be confused with adjustment of a measuring system, often mistakenly called ‘self-calibration’, nor with verification of calibration.” A calibration can end with the instrument exactly as it arrived. What you gain is knowledge: the error at each test point, the measurement uncertainty of that error, and an unbroken chain of comparisons back to national standards. That chain is traceability, and it is what separates a calibration certificate from a sticker.
How this plays out on a specific instrument class, with intervals and methods, is covered in our flow meter calibration guide.
What adjustment means
Adjustment (VIM 3.11) is the set of operations that makes the instrument provide prescribed indications: turning readings back toward true. The VIM lists the classic forms, zero adjustment, offset adjustment, and span or gain adjustment, which older technicians know as the zero and span screws on an analog transmitter.
On a modern HART transmitter the same work is done digitally, and it splits in two. A sensor trim applies a known input and corrects the digital measurement itself, either at a single zero point or at a lower and upper point across the range. An output trim aligns the 4–20 mA output with the digital value and needs no process input at all. One more operation gets confused with both: re-ranging, which changes only the 4 mA and 20 mA range points.
Re-ranging alters what the current output represents, but it corrects nothing and calibrates nothing.
The VIM adds two notes that define good practice. Calibration is a prerequisite for adjustment: you cannot correct an error you have not measured. And after an adjustment, the instrument must usually be recalibrated, because the old calibration no longer describes the device you now have. Adjustment without a follow-up calibration leaves an instrument in an unknown state with paperwork that describes its past.
What verification means
Verification (VIM 2.44) is the provision of objective evidence that an item fulfils specified requirements. For instruments, the specified requirement is usually a maximum permissible error: the calibration found an error of so much, the tolerance allows so much, therefore the instrument passes or fails. Verification adds no new measurement knowledge. It takes calibration data and renders a verdict.
That ordering matters. A verification without a calibration behind it is an opinion. This is also where validation splits off: validation (VIM 2.45) is verification where the question is whether the specified requirements themselves are adequate for the intended use. A temperature sensor can verify perfectly against a ±1 °C requirement and still fail validation if the process actually needs ±0.2 °C.
Requirements come from the application; class tolerances such as those for Class A and Class B RTDs are a common starting point.
As-found and as-left
Field practice ties the three operations into one visit, and the record keeps them separate. The as-found calibration documents the instrument exactly as it was operating, before anyone touches it. If the results sit inside tolerance, the work is done and as-found equals as-left. If not, the technician adjusts, then recalibrates; that second data set is the as-left record.
Here is a worked example for a 0–250 kPa pressure transmitter with a tolerance of ±0.25 percent of span, checked at five points. The numbers are illustrative, but the pattern is what a real certificate looks like:
| Point | Applied | As-found error | As-left error |
|---|---|---|---|
| 0% | 0 kPa | +0.02% span | +0.01% span |
| 25% | 62.5 kPa | +0.08% span | +0.02% span |
| 50% | 125 kPa | +0.15% span | +0.03% span |
| 75% | 187.5 kPa | +0.24% span | +0.03% span |
| 100% | 250 kPa | +0.32% span | +0.04% span |
Tolerance ±0.25% of span = ±0.625 kPa. The 100% point failed as-found (0.8 kPa high), so the transmitter received a sensor trim and a full as-left run.
The as-found data is not bureaucracy. It answers the question every quality manager eventually faces: was this instrument reading correctly during the last interval? If the as-found result is out of tolerance, every measurement since the previous calibration is suspect, and the record is what limits the damage. It also drives interval decisions: an instrument that keeps arriving well inside tolerance can justify a longer interval, and one that keeps drifting out needs a shorter one or a repair.
Judging pass or fail
The pass or fail verdict has its own fine print, because the calibration error comes with an uncertainty. ISO/IEC 17025 requires a laboratory making a statement of conformity to document its decision rule: how uncertainty is taken into account when declaring pass or fail. The ILAC-G8 guideline describes the common options.
Take the transmitter above with its ±0.25 percent tolerance and suppose the expanded uncertainty of the calibration is 0.05 percent of span. Under simple acceptance, a measured error of +0.23 percent passes, because 0.23 is inside 0.25. Under guarded acceptance, the limit shrinks by the uncertainty to 0.20 percent, and the same +0.23 percent result no longer qualifies as a pass, because part of its uncertainty band lies outside tolerance.
Neither rule is wrong; they allocate risk differently. What is wrong is not knowing which rule your certificate used.
Ask, and make sure it matches the risk of your application.
Legal verification
The word verification carries a second, stricter meaning in legal metrology, and international buyers meet both. In the OIML framework, verification is a conformity assessment performed under national law: an authorized body tests the instrument against legally fixed maximum permissible errors and, on passing, affixes a verification mark or seal and issues a verification certificate. Initial verification happens before first use; subsequent verification repeats at legally set intervals.
This applies to instruments in regulated service: custody transfer meters, retail scales, fuel dispensers, utility water meters. The output is legal validity, not measurement data. A verified water meter carries a stamp saying it may legally bill customers; it does not come with an error-versus-flow curve. Industrial calibration and legal verification therefore answer different questions, and a plant that exports or bills across borders often needs both: the national verification for the billing point, and traceable calibration data for its own quality system.
Quality system requirements
ISO 9001:2015 is where most plants first meet these words in writing. Clause 7.1.5.2 requires measuring equipment to be “calibrated or verified, or both, at specified intervals, or prior to use,” against standards traceable to international or national measurement standards. The same clause requires identification of calibration status and safeguards against adjustments that would invalidate results. Notice the standard’s own grammar: calibration and verification are listed as two things, and unauthorized adjustment is treated as a threat to both.
For purchased instruments, the practical checklist is short. Ask whether the certificate reports actual errors with uncertainty, or only a pass statement. Check that the laboratory’s accreditation scope covers the range and parameter you need. Insist on as-found and as-left data whenever an adjustment is possible.
And for billing or safety applications, ask which decision rule stands behind the word “pass.”
Application example
Electronics manufacturer, Turkey and Germany. The customer needed cryogenic pressure sensors for −196 °C service, ranged 0–100 bar with 150 bar overpressure capability, 4–20 mA on 24 VDC, with local display. The order condition was documentation: each unit had to ship with a calibration and a certificate of conformance, not just a factory test note. We supplied six SI-2088 sensors, each delivered with its calibration record and CoC.
Why the terms blur
If the definitions are this clear, why do so many published guides disagree? Because the calibration service industry uses “calibration” as a package word. When a service shop says calibrate, it often means the whole visit: measure, adjust if needed, measure again.
The VIM meaning survives inside that package as the measuring steps. Both usages are entrenched, and arguing about the word solves nothing.
Writing an unambiguous purchase order does. Specify the operations, not the label: “five-point as-found calibration with reported uncertainty; adjust if any point exceeds ±0.25 percent of span; as-left calibration after any adjustment; statement of conformity with the decision rule identified.” Every laboratory understands that sentence, whichever way it uses the word day to day. A field check against a portable reference meter between laboratory calibrations fits the same logic: it is a verification, it adjusts nothing, and a failed check triggers a real calibration.
Instranova supplies instruments, from pressure transmitters to complete flow meters, and can include a factory calibration record and a certificate of conformance when the order specifies them. Spans and outputs are configured to your specification before dispatch. Tell us the application and we configure one unit, not a shelf part.
The bench sequence behind those as-found and as-left data sets, with the tolerance bands and trim decisions worked in mA, is in our pressure transmitter calibration guide.
FAQ
What is the difference between calibration and verification?
Calibration compares an instrument to a reference standard and documents the errors and their uncertainty; it changes nothing. Verification takes that result and judges it against a specified requirement, usually a maximum permissible error, producing a pass or fail statement. Calibration produces data; verification produces a verdict based on that data.
What is the difference between calibration and balance verification?
A balance verification is the routine check where a known check weight is placed on a scale to confirm it still reads within tolerance. It adjusts nothing and involves few points. A balance calibration is the full documented comparison across the weighing range, with uncertainties, performed against traceable reference weights.
What is the difference between calibration verification and linearity?
Calibration verification tests known-value references at a few points to confirm an existing calibration still holds. A linearity study spreads test points across the whole measuring range to confirm the response stays proportional between those points. One asks whether the calibration is still valid, the other asks whether the response is straight.
What is the difference between calibration and inspection?
Inspection is a broad conformity check of an item, which may be visual or functional and may not involve measurement standards at all. Calibration is specifically the documented comparison of an instrument’s indications against traceable standards, with uncertainties. An inspection may include a verification; only a calibration produces traceable error data.
Request a quote
Tell us the instrument, range, tolerance, and the documentation you need: as-found and as-left data, certificate of conformance, or traceable calibration with uncertainty. We will quote the instrument and the paperwork as one package. Reach our application engineers or use the form below.
Written and technically reviewed by Wu Peng and the Instranova engineering team.