Know what the number can support

Calibration — open definition establishes a relationship between an instrument’s output and reference values under documented conditions. It can include offsets, sensitivity and nonlinear behavior. It is not a promise that a device stays perfect forever. Temperature changes, aging and a new operating mode can alter the response.

A check against a reference helps reveal whether a change belongs to the target or the instrument. Repeated measurements of the same target alone may not reveal a shared systematic error — open definition. Keep reference measurements and their own uncertainties in the evidence chain.

Precision — open definition concerns how closely repeated results agree under specified conditions. Accuracy concerns closeness to the relevant true or reference value. A device can give tightly grouped measurements that are consistently offset. Averaging can reduce some random variation, but does not automatically remove a common bias.

Uncertainty communicates the range or distribution of values reasonably supported by the measurement and model, with its meaning stated. Extra decimal places do not create new knowledge. A reported uncertainty needs a method and context, not merely a plus-or-minus symbol copied from another instrument.

In a fictional linear instrument, a zero-input reference produces 8 counts and the sensitivity is 4 counts per unit. A target reading of 48 counts gives (48 − 8) ÷ 4 = 10 units. Dividing by sensitivity before accounting for the offset would report the wrong result. These exact numbers illustrate the arithmetic, not a real calibration certificate.

If the reference drifts to 12 counts, the old correction becomes inappropriate unless the drift is investigated and accounted for. A stable-looking target can appear to change when the instrument changes. Reference observations help distinguish these possibilities but still have finite precision.

Subtract the offset, then apply sensitivity

A nondetection means that a signal was not established under the observation’s sensitivity and decision criteria. It does not generally establish that the physical quantity is exactly zero. Similarly, an over-range value tells you that the instrument’s calibrated range was exceeded, not how far beyond it the target lies.

A useful report gives the quantity, calibration, time, relevant conditions, uncertainty and flags. Then it states the claim those measurements support. This final step links all the instrument lessons: choosing a sensor is choosing both an opportunity to measure and a set of limits you must carry into the conclusion.