Measurement Uncertainty: Why 98.4% Is Not 98.4%

Certificates report purity to one decimal place. The implied precision is almost never justified, and treating 98.4% and 98.9% as meaningfully different is one of the most common errors made when comparing two suppliers.

Where the uncertainty comes from

A purity figure is an area ratio derived from an integrated chromatogram, and uncertainty accumulates at every step before the division.

• Injection. Replicate injections of one sample do not give identical areas. Relative standard deviation of a few tenths of a percent is normal on a well-behaved system, and that is the floor.
• Integration. Where the operator or the software places the baseline under a tailing peak changes the area. Two competent analysts integrating the same trace routinely differ, and the difference grows with peak asymmetry.
• Detection response. An area percentage assumes every species responds equally at the detection wavelength. They do not. An impurity lacking the target’s chromophore is under-counted; one with a stronger chromophore is over-counted. At 214 nm the assumption is reasonable because the peptide bond dominates. At 280 nm it can fail badly.
• Resolution. Anything unresolved is counted as target, which biases the figure upward rather than scattering it. Co-elution covers that mechanism.

What a realistic interval looks like

Combining injection precision, integration variability and response assumptions, a purity figure on a well-run method carries an uncertainty on the order of a few tenths of a percent at best, and around a percent where peaks are asymmetric or partially resolved. An accredited laboratory is required to be able to state its uncertainty; most research-peptide certificates do not.

The practical rule that follows: a difference of a few tenths of a percent between two certificates is noise. A difference of several percent is a signal worth asking about.

Why this matters when comparing suppliers

Ranking vendors by the decimal on a purity figure compares the precision of their integration software, not the quality of their material. Two figures generated by different methods at different wavelengths on different columns are not on the same scale to begin with, before uncertainty is considered.

The more informative comparison is what each document contains: the wavelength, the chromatogram itself, whether net peptide content is reported, and whether the batch on the certificate matches the vial. Why certificates of analysis disagree covers the mechanisms, net peptide content explained covers the figure that is usually missing, and choosing a research peptide supplier in Canada sets out what to ask for.

What uncertainty is not

It is not error, and it is not a defect. A stated uncertainty is a sign of a laboratory that understands its own method; its absence is not evidence of a bad result, only of a result whose precision is unstated. And as with every figure on a purity certificate, it describes composition — not sterility, not endotoxin, and not suitability for use in a person or an animal.

All material is supplied for laboratory research use only. It is not a drug, not a supplement, and not for use in humans or animals.