Section 03 · The Bench E-13

The Calibration Curve

Known amounts measured first, so an unknown has something to be compared against.

Row of sealed glass vials with metal caps lined up on a lab counter — FIG. 01The unknown is read against knowns run the same day.Photo: Jess Loiterton / Pexels

What the curve does

Before a single sample is analysed, the instrument is given a series of known concentrations — standards — and asked to respond. The detector produces a signal for each one: a peak area, a peak height, a voltage. Plot signal against concentration and the result is the calibration curve, usually a straight line across the working range, occasionally a gentle arc where the detector begins to saturate at higher loads.

Every unknown sample is then read against that relationship. The instrument sees a peak area; the curve converts it into an amount. Without standards run on the same instrument, on the same day, under the same conditions, that conversion is guesswork.

An analytical laboratory bench, stainless and glass, hard light — FIG. 02Known amounts are run first; the unknown is only ever read against them.Photo: Nishant Aneja / Pexels

What can go wrong

The standards themselves have to be traceable — prepared from certified reference materials with known uncertainties, not weighed out informally and assumed correct. A poorly made standard shifts the entire curve, and every result derived from it carries that shift silently.

The range matters too. A calibration built across five points from 1 to 10 µg cannot reliably report a sample at 50 µg; extrapolating beyond the highest standard is not interpolation, it is speculation. Laboratories set a working range deliberately, and samples that fall outside it are either diluted or re-run. The integration step that produces the peak area fed into the curve carries its own operator-dependent uncertainty, which propagates through directly.

Linearity is checked statistically — a correlation coefficient close to 1 is necessary but not sufficient; the residuals at each point should be examined too. A single poorly prepared standard that drags the fit can be easy to miss if only the headline r² is reported.

What to look for in a report

A well-documented result will name the calibration range, the number of points used, the r² value and, ideally, the identity and source of the reference materials. If those details are absent, the number in the result column rests on foundations the report does not show you.

A chromatogram trace on an instrument screen — FIG. 03The curve holds between its lowest and highest standard — outside that span the instrument is extrapolating.Photo: Jess Loiterton / Pexels

Filed inSection 03 — The Bench

Calibration curve	a plot of detector signal versus known concentration; the relationship used to convert every unknown into an amount
Working range	the span of concentrations across which the curve is valid; extrapolation beyond it is not supported
Traceable standards	reference materials with certified concentrations and documented uncertainty chains
r² (correlation coefficient)	a statistical measure of how well the data fit a straight line; necessary to report but not sufficient on its own
Residuals	the differences between measured and fitted values at each calibration point; examined to catch a poorly prepared standard