Skip to content
VTValiTracAI
Uncertainty

Drift as an uncertainty component

Between calibrations an instrument's error moves. Its history tells you how much; that amount enters the budget for any result obtained during the interval.

ValiTrac AI editorialUpdated 2026-09-132 min read

A reference thermometer calibrated in January and used in November has probably changed. If successive certificates show the ice-point resistance moving by an amount equivalent to 0.01 °C per year, then at any time during the year the error is somewhere within about ±0.01 °C of the certificate value — a rectangular distribution with u = 0.01/√3 ≈ 0.006 °C.

Refinements

  • Time-proportional: use the time elapsed since calibration rather than the full interval.
  • Corrected drift: if the change is linear and consistent, apply a correction and carry the uncertainty of the fit.
  • Interim checks: an ice-point or fixed-point check mid-interval can tighten the estimate and catch abnormal behaviour.

A new instrument with no history has unknown drift; use the manufacturer's stability specification as a conservative rectangular limit until history exists.

Frequently asked questions

Can drift be ignored for a freshly calibrated reference?
Not for results obtained through the interval. A budget should be valid for the whole period the certificate is relied on.

References

  1. [1]JCGM 100:2008 — Guide to the expression of uncertainty in measurement (GUM)
  2. [2]EA-4/02 M: Evaluation of the Uncertainty of Measurement in Calibration

General technical guidance written against the cited sources. It is not regulatory or legal advice and does not replace the applicable standard, guideline or a qualified reviewer's judgement.

Related articles