Peak Integration Is a Decision, Not a Measurement

The baseline is an argument

The baseline is the line the software subtracts from the signal before it measures area. On a clean separation it may look obvious. On a real run it rarely is. Gradient drift, a rising solvent front, a broad hump under the main peak, or a small negative dip near the end of the run can all change where that line belongs. Automatic routines make a first proposal; an analyst accepts it, adjusts it, or redraws it by hand. Each of those actions is an argument about what part of the trace is chemistry and what part is background. If the report does not show the baseline, the reader cannot see which argument was made. Two files can carry the same peak list while resting on different baselines, and the difference will sit in the areas rather than in the picture.

Thresholds decide what exists

Integration settings decide what the software is willing to call a peak. Slope sensitivity, minimum height, minimum area, peak width, and smoothing all act as filters. Tighten a threshold and small shoulders disappear from the table; loosen it and noise begins to look like composition. Neither setting is a neutral description of the trace. The choice should follow the question and the method, and it should be recorded. Software version matters too, because peak-detection routines change between releases. A table with no settings and no version is not reproducible from the page alone. It may still be correct, but the reader cannot check how the list of peaks came to exist.

Area percent has a denominator

Area percent is a ratio, and the denominator is a decision. The software adds the areas of the peaks it has included, then divides each peak by that sum. Leave out a late-eluting peak, a solvent front, or a peak below the threshold, and every remaining percentage rises or falls. Include a peak from the sample diluent, and the main component's share can fall even when its absolute area has not changed. The result is a statement about the chromatogram as processed, not a complete statement about the material. Compounds that do not respond at the detection wavelength, counterions, residual water, and material retained on the column are outside the sum. A separate note on how a purity figure is framed is useful here, because a ratio can travel only as far as its denominator and detection method. The number is real; the question is what it is a percentage of.

Retention time is not identity

Retention time is a position on a trace, not an identity. A peak that arrives a few seconds early may be the same compound under a slightly different temperature, flow rate, or column history. A peak that arrives on time may be something else that happens to co-elute. Drift becomes a reading problem when the software tracks peaks by expected time windows: if the window is too narrow, a peak is missed; if it is too wide, two peaks can be merged. The report may show only a final table, which hides both possibilities. A standard run, a spiked sample, or a second detection channel can support an assignment, but the report has to say that such evidence exists. Without it, the retention time is a clue, not a conclusion.

Two tables can both be defensible

Two integration reports can both be defensible. One analyst may include a small impurity to be conservative; another may exclude it because it falls below the method's reporting threshold. One may draw a valley-to-valley baseline; another may use a skim. If both record their choices, the disagreement is visible and can be discussed. If only the percentages survive, the discussion collapses into a contest between numbers that were never produced under the same rules.

Before comparing two area percentages, ask:

• Which peaks were included in the denominator?
• Where was the baseline drawn, and is an overlay available?
• Which threshold settings and software version were used?
• How were peaks tracked across the run, and was identity supported?
• Was the same detection wavelength and sample preparation used?

Those questions do not make integration subjective in the sense of arbitrary. They make it explicit. A decision that travels with its reasons is more useful than a measurement that arrives without them.