A purity percentage is not the whole experiment

Ask what a peptide sample is, and someone will eventually quote you a number. Ninety-five percent. Ninety-eight. The figure arrives on a certificate of analysis, travels into a methods section, and from there into the quiet assumptions of everyone downstream: that the material in the vial is, for practical purposes, the material on the label. The number is not meaningless. But it is a summary of a measurement, and like all summaries it conceals as much as it conveys — starting with what, exactly, was measured, and by which instrument, under which conditions, against which definition of the word purity.

Analytical chemists tend to wince at how the figure is used, and their wincing is instructive. A purity percentage is typically a statement about area under peaks in a chromatogram at a particular detection wavelength. It is not a statement about identity. It is not a statement about what the remaining few percent contain. And it is emphatically not a statement about biological behavior. Treating it as any of those things is the analytical equivalent of judging a book by the thickness of its spine.

The appeal of a single number

The appeal is obvious. Peptide research involves long chains of dependency: a synthesis is characterized, a stock is prepared, an assay is run, a conclusion is drawn. Every step inherits the uncertainty of the last, and nobody has the time to personally re-derive the analytical foundations of their own materials. A single number offers compression. It converts a complicated physical reality into something that fits in a cell of a spreadsheet.

The problem is that compression discards information asymmetrically. What is lost first is everything the measurement was never designed to see. A chromatogram reports what absorbs at the chosen wavelength and elutes within the chosen window, separated by the chosen chemistry. Truncated sequences, deletion products, and protecting-group remnants may or may not register, depending on decisions made long before the trace reached anyone's desk. The number that survives this filtration looks exact. Its exactness is a property of formatting, not of knowledge.

What chromatography actually measures

High-performance liquid chromatography separates components of a mixture by their differential interaction with a stationary phase and a moving solvent. The detector — most often ultraviolet absorbance — records a signal as each component passes. Software then draws baselines, integrates peaks, and reports each peak's share of total area. That share, for the main peak, is what usually becomes the purity percentage.

Every clause of that description contains a judgment call. Where the baseline is drawn changes the area. How shoulder peaks are split changes the area. The gradient's slope, the column's age, the buffer's pH, the sample's solvent — each moves peaks and reshapes them. Two competent analysts integrating the same raw trace can produce figures that differ by more than the reported decimal places suggest. None of this is scandal; it is simply measurement. The scandal, such as it is, lies in downstream readers treating the output as if no human decisions were involved.

Purity versus identity

Here is the distinction that matters most and is blurred most often: a chromatogram can tell you how much of the UV-absorbing material elutes at the main peak. It cannot, by itself, tell you that the main peak is the peptide you intended. A sample of the wrong sequence — one that happens to behave similarly on that column, under that gradient — can produce a beautiful trace and a reassuring percentage. Purity and identity are separate questions, and they require separate methods.

Identity is typically established by mass spectrometry, which reports the mass of the molecular ion and can be extended to fragmentation patterns that confirm sequence. A sample can be highly pure and entirely wrong; it can also be the right sequence contaminated with closely related variants that co-elute and masquerade as a single peak. Only the pairing of methods — separation plus mass confirmation — addresses both questions at once. Laboratories that run both as routine tend to describe the combination as unremarkable. Laboratories that run only one tend to discover the omission at inconvenient times.

The importance of analytical context

A purity figure without its method is like a temperature without a scale. The same material run under a shallower gradient may show the main peak resolve into two; run at a different wavelength, minor components with strong chromophores may suddenly loom large. None of these versions is the true one. Each is a view, defined by the instrument's geometry and the analyst's choices, and the honest way to report any of them is with the conditions attached.

This is why experienced reviewers read the method block of a certificate before the number. Column type, gradient, flow rate, wavelength, injection solvent: these are not boilerplate. They are the lens prescription. A certificate that omits them has not simplified anything; it has merely declined to disclose.

“Precision in measurement does not automatically produce precision in interpretation.”

Verdant Signal · Analytical desk

Why different methods produce different answers

Ask two methods the same question and you will get two answers, both correct. Reversed-phase HPLC sorts by hydrophobic interaction; ion-exchange sorts by charge; size-exclusion sorts by hydrodynamic volume. A dimer invisible to one method may be the most prominent feature in another. Mass spectrometry sees what ionizes; what ionizes depends on the source conditions and the sample matrix. The peptide does not change between instruments. The question changes.

Inter-laboratory comparisons make the point concretely. When the same reference material is characterized independently by different groups, reported values commonly diverge by more than each individual measurement's quoted precision — a reminder that precision describes repeatability within a method, not agreement between methods. Researchers who treat a percentage as an intrinsic property of the material, rather than a property of the measurement, are mistaking the map for the territory in the most literal way analytical science offers.

Interpreting certificates of analysis

A certificate of analysis is a claim about a batch, made by whoever produced it, using methods they selected. Read generously, it is a useful starting point. Read carefully, it is also a document with an evidentiary structure worth checking: Was identity confirmed by mass, or only inferred from retention time? Is the purity figure area percent at a stated wavelength, with the method described? Are related substances individually reported or silently integrated away? Was the batch actually tested, or the certificate inherited from an earlier lot?

None of these questions imply bad faith. They are the ordinary diligence of a field in which the certificate is often the only analytical document most researchers will ever see. The habit worth cultivating is treating the certificate as the beginning of characterization rather than its conclusion — particularly when the material is headed for an assay whose interpretation will lean on it.

The limitations of analytical certainty

It is worth saying plainly what no combination of methods delivers: certainty. Every technique has a detection limit below which components disappear from view, and the space below that limit is not empty — it is merely unobserved. Water content, counterions, residual solvents, and sequence variants at fractional percentages can all ride along invisibly while the reported number gleams. Good analytical practice shrinks the unobserved region. It does not eliminate it.

There is also a subtler limitation. Analytical characterization describes the material as it was, in the vial, at the moment of measurement. What reaches the assay is that material plus its history: storage temperature, freeze–thaw cycles, container surfaces, time. A certificate is a photograph, not a guarantee, and peptide materials in particular are known to change with handling. The distance between the certified material and the experimental one is itself an uncontrolled variable unless someone chooses to control it.

Better questions for researchers

None of this argues for despair, or for turning every experiment into an analytical odyssey. It argues for better questions — the kind that fit in an email and change what a result can mean. What method produced this figure, and at what wavelength? Was identity confirmed independently of purity? What is known about the principal impurities, rather than their sum? How was the material stored and handled between characterization and use? Which of these answers would change the interpretation of the assay?

The single number will survive, because compression is useful and no laboratory can re-derive everything. But it can be demoted from verdict to summary — one figure among several, quoted with its method attached and its limits in view. The instruments behind the number tell a more complicated story. The story is more useful, too.