How does calibration turn a sensor response into a useful measurement?

A sensor responds to a physical quantity, but its raw output is not automatically a reliable value in the units we want. Calibration compares responses with known reference values under specified conditions. This establishes a relationship that can be used to interpret later readings.

Suppose an invented temperature sensor outputs 1 volt at 0 degrees Celsius and 3 volts at 100 degrees Celsius. Under a linear model between those points, temperature T = 50 × (V − 1). A reading of 2 volts would then correspond to 50 degrees Celsius.

That calculation depends on the model and the reference measurements. Two points cannot establish that the response is truly linear throughout the interval. Repeated comparisons, additional reference points and tests across relevant conditions can reveal offset, changes in sensitivity or other deviations.

Calibration is also distinct from physically adjusting an instrument, although adjustment may follow it. A corrected reading still has uncertainty from references, noise, repeatability, the model and changing conditions. A useful measurement reports what was compared, which relationship was used and how well the result is supported.