Not detected is the most reassuring phrase on a certificate of analysis and the least informative. It is a statement about the instrument, not about the vial. Read without the number that belongs beside it, it cannot be compared between suppliers, between lots, or against any specification — and the difference between a method that can see 0.5% and one that can see 0.01% is a hundredfold difference in what that phrase is worth.
Quick answer
What does not detected mean on a peptide certificate of analysis? It means the analyte produced no signal distinguishable from baseline noise by the method used, at the sensitivity that method achieves. It does not mean zero. ICH Q2(R2) defines two separate thresholds: the detection limit, generally estimated at a signal-to-noise ratio of 3:1, and the quantitation limit, requiring at least 10:1. An analyte can therefore be detected and still not be quantifiable. Not detected without a stated detection limit is uninterpretable; less than 0.05% is a usable result. When a certificate gives the first form, the number to ask for is the limit the method achieved.
Three statements that look alike and are not
| Reported as | What it actually asserts | Usable? |
|---|---|---|
| Not detected | No signal above noise, at an unstated sensitivity | No. Could mean anything from below 1% to below 0.001%. |
| Not detected (DL = 0.02%) | No signal above noise, by a method that would have seen 0.02% | Yes. Equivalent in information to a less-than statement. |
| Less than 0.05% | Below the quantitation limit of a method whose QL is 0.05% | Yes, and it is the clearest of the three. |
| 0.00% | Nothing. No measurement yields exact zero. | No, and it is a reason to question the document. |
The last row is worth dwelling on. An analytical result of exactly zero is not a measurement outcome; every real method has a floor. A certificate reporting 0.00% for an impurity is either rounding a less-than result without saying so, or has been written rather than measured.
What the guideline actually specifies
ICH Q2(R2), the current international guideline on validation of analytical procedures, uses the terms detection limit (DL) and quantitation limit (QL), and defines them as the minimum concentrations at which the analyte can be reliably detected or quantitated respectively. It gives several ways to establish them.
The signal-to-noise approach
This approach applies only to procedures that exhibit baseline noise — chromatographic methods, principally. The ratio is determined by comparing measured signals from samples containing known low concentrations of analyte against those of blank samples, or against signals in an appropriate baseline region. The guideline states that a signal-to-noise ratio of 3:1 is generally considered acceptable for estimating the DL, and that for the QL a ratio of at least 10:1 is considered acceptable. It adds a detail that is easy to overlook: the ratio should be determined within a defined region, situated where possible equally around the place where the peak of interest would be found.
That last requirement exists because baseline noise is not uniform across a chromatogram. Measuring noise in a quiet stretch of the run and applying it to a region where the baseline is drifting produces a flattering detection limit that the method cannot actually deliver where the peak would appear.
The calibration approach
Where a linear response is available, Q2(R2) expresses the limits arithmetically:
- DL = 3.3σ / S
- QL = 10σ / S
σ is the standard deviation of the response and S is the slope of the calibration curve. The guideline allows σ to be estimated either from replicate measurements of blank samples, or from the calibration curve itself — using the residual standard deviation of the regression line, or the standard deviation of the y-intercepts of regression lines.
Both approaches are legitimate and they do not always agree, which is one reason two suppliers running comparable instruments can report different detection limits for the same impurity in good faith. Q2(R2) also allows determination by visual evaluation, and by demonstrating acceptable accuracy and precision at the lower range limit — the last of which is the most demanding, because it requires showing the method actually performs there rather than merely producing a signal.
Why detection and quantitation are different numbers
The gap between 3:1 and 10:1 is not pedantry. It describes a real region where a peak is visibly present and its area cannot be trusted. Inside that band, integration is dominated by where the analyst places the baseline, and two competent analysts will disagree.
The consequence for a certificate is that an impurity reported at a level near the quantitation limit carries far more uncertainty than the number of decimal places implies. A result of 0.11% from a method with a QL of 0.10% is a different kind of claim from 1.10% on the same method, even though both are reported to two decimals. Q2(R2) makes this explicit by requiring that the QL be a level at which accuracy and precision are demonstrated, not merely a level at which a signal appears.
Where this appears on a peptide certificate
| Certificate line | Typical method | What a limit statement should accompany it |
|---|---|---|
| Individual impurity peaks | Reversed-phase HPLC with UV detection | The integration threshold or reporting limit, and the DL if impurities are called not detected |
| Specific deletion or truncation sequences | LC-MS | The DL for that species, which can differ markedly between impurities |
| Bacterial endotoxin | LAL assay | The assay sensitivity and the sample dilution, which together set the reportable limit |
| Residual solvents | Headspace GC | The DL per solvent. These differ by solvent and are not interchangeable |
| Water content | Karl Fischer titration | Not a noise-limited method in the same sense; reported as a value, and a not-detected result here would be unusual |
Endotoxin is the clearest case of a limit that is arithmetically constructed rather than intrinsic. An LAL result is bounded by the labelled sensitivity of the assay multiplied by the dilution factor applied to the sample, so a heavily diluted sample yields a numerically high reportable limit no matter how sensitive the reagent. A result of less than 10 EU/mg and one of less than 0.1 EU/mg can come from the same reagent lot, differing only in how the sample was prepared. We cover the assay in endotoxin testing and the LAL assay.
The threshold hidden inside the purity figure
There is a related cutoff that almost never appears on a certificate and that affects the headline number directly. A chromatographic purity figure is calculated from integrated peak areas, and integration software applies a rejection threshold below which peaks are not integrated at all. Peaks below that threshold do not become impurities; they become baseline.
This means a purity percentage is not a fully specified quantity without the threshold that produced it. The same injection, integrated with a lower rejection threshold, will resolve additional small peaks and return a slightly lower purity. The effect is small for well-made material and larger for material with many minor impurities — which is precisely the material where the distinction matters.
None of this makes purity figures unreliable. It makes them method-dependent, which is the same reason purity values from different suppliers are not directly comparable even when both are honest. The practical defence is to compare lots within a supplier, where the method is held constant, rather than comparing headline numbers across suppliers. Our notes on reading an HPLC chromatogram and deletion and truncation impurities cover what the chromatogram itself shows.
How to read a certificate with this in mind
- Treat every not-detected as an incomplete result and look for the accompanying limit. If there is none, the line is a statement of intent rather than data.
- Prefer less-than formulations. They carry the same information in a form that cannot be misread as zero.
- For endotoxin, check whether the dilution factor is stated. Without it the reportable limit cannot be reconstructed.
- Treat any value within roughly a factor of two of the quantitation limit as approximate regardless of the decimals shown.
- Be sceptical of 0.00% and of results reported to more decimal places than the limit of the method could support.
- Compare lots within one supplier rather than headline figures between suppliers, because detection limits and integration thresholds travel with the method. This is part of the broader case for batch-level documentation in why batch-specific testing matters.
Asking a supplier for a detection limit is a routine question, not a confrontational one. A supplier running real analysis has the number; the request simply distinguishes documentation that reflects measurement from documentation that reflects a template.
Key takeaways
- Not detected describes the sensitivity of the method, not the composition of the sample. Without a stated limit it cannot be compared to anything.
- ICH Q2(R2) sets the detection limit at a signal-to-noise ratio of about 3:1 and the quantitation limit at a ratio of at least 10:1.
- The arithmetic forms are DL = 3.3σ/S and QL = 10σ/S, with σ from blank replicates or from the regression residuals.
- The signal-to-noise approach applies only to methods with baseline noise, and noise must be assessed in the region where the peak would appear.
- Between the two limits lies a band where a peak is visible and its area is not trustworthy. Results near the QL are approximate however they are formatted.
- An endotoxin limit is the assay sensitivity multiplied by the dilution, so the dilution factor is part of the result.
- Purity figures depend on an integration rejection threshold that certificates rarely state, which is a further reason to compare lots rather than suppliers.
- A reported 0.00% is not a measurement outcome.
Frequently asked questions
Is not detected a sign of a poor certificate?
Not on its own. It is standard phrasing and often appears on entirely legitimate documentation. What distinguishes a strong certificate is that the limit accompanies the phrase, so the reader can tell what was ruled out. The problem is incompleteness, not dishonesty.
Can something be detected but not quantified?
Yes, and that is exactly the region between the two limits. A peak at a signal-to-noise ratio of 5:1 is detectable by the 3:1 criterion and falls short of the 10:1 the guideline asks for before a number is reported. The correct report for such a peak is that it is present below the quantitation limit.
Why do two suppliers report different detection limits for the same impurity?
Because the limit is a property of the method, not of the impurity. Column, detector, wavelength, injection volume, sample concentration and which of the permitted estimation approaches was used all move the number. Different limits are expected; unstated limits are the issue.
Does a lower detection limit mean better material?
No. It means a more sensitive method was used to examine it. A sensitive method finding nothing is stronger evidence than an insensitive one finding nothing, but the sensitivity says nothing about the peptide. Conflating the two is the most common error in reading these documents.
What should I keep in my own records?
Record the limit alongside the result, not just the result. A file that says endotoxin not detected for three lots is not comparable across those lots; one that says not detected, limit 0.5 EU/mg for each is. The same applies to impurities and residual solvents.
Sources
- International Council for Harmonisation. ICH Q2(R2): Validation of Analytical Procedures. Sections 3.2.3.1–3.2.3.5, Validation of Lower Range Limits. ICH guideline
- International Council for Harmonisation. ICH Q3C(R8): Impurities — Guideline for Residual Solvents. ICH guideline
This article describes analytical documentation only. It contains no usage instructions, no dosing information and no protocols of any kind.
For laboratory and in-vitro research use only. Not for human or veterinary use.
For laboratory and in-vitro research use only. Not for human or veterinary use. See our research-use disclaimer.
