As Found and As Left: Meaning and Use of Calibration Results

As Found and As Left locate calibration results within an instrument’s history. They are not, by themselves, conformity statements: they identify conditions associated with service stages. A favourable final result may support return to use but does not automatically describe performance before a repair.

The report’s value depends on the documented sequence: what arrived, how it was prepared, when measurements took place and which operations changed the response. Asking only for a “before and after certificate” is insufficient if these steps remain undefined.

1. Define the states before ordering the service

In service terminology, As Found normally identifies results obtained in the agreed initial condition, before relevant interventions; As Left identifies results in the final condition in which the instrument is returned. Define the conditions behind these labels. “As received”, “before adjustment” and “before repair” may refer to different moments.

The VIM distinguishes calibration from adjustment: calibration establishes metrological relationships with associated uncertainties; adjustment changes the response. Calibration does not always require adjustment. As Found and As Left do not replace these definitions or require every provider to deliver two sets of tests.

Specify characteristics, points, configurations, conditions, required uncertainty, any conformity statement and authorisations in the order. Agree when the provider must stop and contact you, which initial measurements to collect and which interventions are already authorised. A generic “calibration with maintenance” order can leave precisely these matters unresolved.

2. Record the condition before changing the response

Identify the instrument, accessories, sensors, channels, relevant software version, settings and abnormalities observed on receipt. Record damage, user reports and conditions that may affect testing. For a measurement chain, the indicator’s serial number alone may not identify the system measured.

Distinguish warm-up, stabilisation and preparation within the normal method from operations that change the response, such as replacing a seal, changing coefficients or cleaning a sensitive surface. The distinction depends on the equipment and use: routine adjustment may itself form part of the agreed measurement conditions. Describe what happens instead of relying on the command’s name.

When pre-intervention data are needed, acquire them before changing the state. Do not perform unsafe tests merely to obtain an As Found label. If contamination, failure or damage prevents measurement, retain available observations, document the reason and agree the next step: “not measured” means neither “conforming” nor “zero error”.

3. Original chronology: activity, state, data and conclusions

Link each row to original records and responsible people. If service includes several intervention and measurement cycles, identify intermediate stages: the initial/final pair alone can conceal a meaningful sequence.

4. No intervention or missing initial data

Where no intervention occurs, only one calibration dataset may be available. The report must make the actual sequence understandable; there is no reason to invent a second independent acquisition. If the same results represent both the initial and returned conditions, explain that this is the same dataset under documented conditions.

If two acquisitions exist, identify their times, configurations and conditions even when values match. Numerical equality does not prove only one test occurred; two headings do not prove two tests occurred.

Where As Found is absent, ask why and what residual evidence exists: internal checks, reported issues, diagnostic data or maintenance records. Assess this information for what it actually contains; it does not retrospectively become an initial calibration. Avoid zero-filled fields, results copied from As Left or wording that conceals the missing measurement.

5. Compare genuinely comparable results

Before subtracting two errors, check quantity, unit, sign, point, configuration, conditions and correction treatment. Consider uncertainties and dependencies, such as a common reference. A change in results after intervention is not, by itself, a measurement of drift before the service.

Keep initial, intermediate and final results separate. An average mixing states before and after adjustment may describe an instrument condition that never existed in use. The consulted NIST volumetric procedures require separate initial and final results in specified cases: this is a relevant example of sequence management, not a universal pharmaceutical laboratory procedure.

Check coverage too: a final test on one channel or point does not demonstrate restoration of the entire chain. If components or settings change, assess which characteristics need new evidence and record any use restrictions.

6. Simulated case: repair before measurement

A pipette arrives with a reported leak. The provider replaces its seal before taking gravimetric measurements and then performs the agreed calibration with favourable results. The report includes final results and the repair description but no quantitative measurements in the earlier state.

The laboratory cannot use these results to claim that the pipette conformed before replacement. The reported leak suggests a problem but does not alone quantify error in previously delivered volumes. The laboratory requests the actual sequence, preserves records and evaluates affected uses with available evidence.

Return to use requires review of final tests against intended uses and authorisation under the procedure. Retrospective assessment may remain open. For the next service, a contact point is agreed before interventions that erase the initial state, where that state can be measured safely.

Read about retrospective impact assessment and calibration interval review. Return to the Calibration and Metrology hub.

7. Sources and applicability limits

Checked on 1 October 2026. The chronology and case are original; they do not establish a universal requirement for two calibrations.

1. VIM, third edition, §2.39 and §3.11: official entries consulted, calibration and adjustment; not formal definitions of the two service labels.
2. NISTIR 7383, 2019 edition: full PDF available; relevant SOP 26 sections consulted on maintenance, initial measurements and separation of results. Specific scope: volumetric calibrations.
3. Accredia, accredited calibrations and certificates, 10 November 2023: complete institutional article; before/after adjustment or repair data, when available.
4. ISO/IEC 17025:2017: official catalogue, confirmed in 2023; full standard not consulted. Check applicable requirements in the controlled copy and service agreements.