Skip to content
Technical & commercial enquiries: sales@rssechem.com Certificate of analysis issued per lot · Worldwide shipping
Register Request a quote & Contact us

Two certificates for the same compound: reconciling them field by field before trusting either number

Reference standards & documentation Published By RSSE Technical Team · Reviewed by RSSE Quality Unit

Two documents, one compound, and two numbers that appear to disagree. Most of the time they do not: they are answering different questions. This works through the five fields that have to match first, the arithmetic that removes the gaps which are not real, and the one case where no comparison is possible at all.

On this page

A number is not a value until its measurand is named

Two documents arrive for the same compound. One says 99.4%. The other says 98.1%. The purchasing decision looks like a ranking, and the ranking looks obvious. It is not, because neither document has yet said what its number measures.

Reference material documentation is a structured set of claims, not a description of a powder. ISO 33401 exists precisely to sort those claims into mandatory, recommended and optional, so that a user can confirm suitability from the paperwork alone. Reconciliation is therefore a field-by-field exercise: identity first, then the measurand, then the basis, then the assignment, and only then the arithmetic. Done in that order, most apparent disagreements dissolve. Done in reverse, they become disputes.

The five fields that must match before any subtraction

FieldWhat it fixesWhat goes wrong if it is skipped
IdentityWhich substance, in which formTwo materials are compared as if they were two values of one material
MeasurandWhat the number countsTwo correct numbers are subtracted and the difference is read as quality
BasisAs is, dried, anhydrous or free baseA basis shift of about one point is recorded as a purity difference
Assignment routeMass balance, qNMR, titration, chromatographicA systematic offset between methods is mistaken for a lot-to-lot change
Uncertainty and currencyHow large a difference may be claimed, and as of whenA difference inside the noise is claimed, or a value describing a state that no longer exists is applied

Each field can stop the comparison on its own. None of them is a formality.

Identity: the formula weight settles it in one line

A CAS number usually identifies the parent compound. Salts, hydrates and solvates frequently share that number with the free base, so a matching CAS proves that two suppliers are discussing the same molecule and proves nothing about the form they shipped.

The fastest test is subtraction. Take the formula weight on each document and subtract the smaller from the larger. A difference of 36.46 is one equivalent of hydrogen chloride; 18.02 is one equivalent of water. If either appears, the documents do not describe two values of one material, they describe two materials. No reconciliation is required and no substitution decision is available, because the correct operation is a factor, not a comparison.

Take a free base of 300.00 and its hydrochloride at 336.46. The base fraction of the salt is 300.00 divided by 336.46, or 0.8917. A salt certificate reading 99.0% therefore delivers 88.3% of base. Setting that 99.0 against a free-base certificate at 98.5 and concluding that the salt is the better material is an error of roughly ten percentage points in the wrong direction.

The measurand: four quantities that all get called purity

Reported quantityWhat the number actually countsWhat it cannot see
Chromatographic purity, area per centThe share of integrated peak area belonging to the main peakWater, residual solvents, inorganic residue, non-chromophoric organics, anything retained on the column
Mass-balance content100% minus each measured subtractive termWhatever none of the component assays detected
qNMR contentAnalyte protons against a calibrant of known purityImpurities sharing the same chemical shift; water unless suppressed
Titration assayFunctional groups that react with the titrantEverything inert to that reaction, including closely related analogues

The first two are routinely filed side by side and read as if they disagreed. Take one vial: water 0.15% by Karl Fischer, residue on ignition 0.05%, total chromatographic impurities 0.35%. The chromatographic purity is 100 minus 0.35, or 99.65%. The mass balance is 100 minus 0.35 minus 0.15 minus 0.05, or 99.45%. The gap of 0.20 points is not a discrepancy; it is exactly the two terms the chromatogram never sees.

That arithmetic also works as a diagnostic. When a supplier reports 99.9 area per cent alongside 98.6 mass balance, ask for the subtractive terms. If water, solvents and residue total 1.4%, both numbers are consistent and the material simply carries 1.4% of matter that does not absorb. If the supplier cannot produce that sum, one of the two numbers is being used as a purity claim it was never designed to support.

The basis: the 1.2 points that never existed

Both major pharmacopoeias calculate on an as-is basis unless a monograph states otherwise, and the European leaflet is explicit about it: the assigned content of a reference standard is expressed as is, and a standard dried before use is no longer suitable for its intended purpose. Drying is not a neutral preparation step. It changes the material the number refers to.

Two documents, both internally consistent, both about the same substance:

  • Lot A: 98.6% as is, water 1.2%. On the dried basis: 98.6 divided by 98.8, times 100, gives 99.80%.
  • Lot B: 99.5% on the dried basis, water 0.8%. As is: 99.5 times 0.992 gives 98.70%.

Read raw, the file shows a 0.9-point gap in favour of B. Reconciled on one basis, the gap is 0.10 points, inside almost any uncertainty either document is likely to carry. The conversion is two lines of arithmetic and it reverses the ranking.

The error also has a direction worth naming. A dried-basis number applied as though it were as is inflates the apparent strength of the standard by the water content, here 1.2%, and that inflation propagates into every result calculated from it. Being a bias rather than a scatter, it is invisible to replicate injections and survives every precision check in the file.

Uncertainty: whether a difference is even permitted to exist

An assigned uncertainty is not the producer's repeatability. ISO 17034 requires the budget behind an assigned value to include, at minimum, the characterization step including any difference between the procedures used, between-unit and within-unit inhomogeneity, and changes during storage and transport. The figure on the certificate is therefore the value at release under the stated storage, with a shelf-life model already folded into its uncertainty. A lot held three years at the wrong temperature has spent part of that budget without changing a character on the document.

Where both documents state an expanded uncertainty at the same coverage factor, the comparison is mechanical. Lot A at 99.4% with 0.4, lot B at 98.9% with 0.3, both at k equals 2. The difference is 0.5 points. The combined expanded uncertainty is the square root of 0.4 squared plus 0.3 squared, which is 0.50 points. The difference equals its own uncertainty to the last digit, so the two documents do not support a statement that the lots differ, whatever the chromatograms look like.

The harder case is a document that declines to state an uncertainty at all. The Ph. Eur. leaflet template is explicit that the uncertainty of the assigned value is not stated because it is considered negligible in relation to the defined limits of the method-specific assays for which the standard is used. That is a defensible statement and it is not a number. It cannot be entered into the square root above.

What survives is a different question. Instead of asking whether two values agree, you ask whether each value falls inside the limit the method must enforce. That is weaker, but it is the only comparison the paperwork supports. Supplying a plausible uncertainty to reach the stronger claim produces a conclusion with a precision the file does not have.

What the document is designed not to contain

Some absences are deliberate, and none of them can be reconciled against anything.

USP does not issue certificates of analysis for its reference standards; it issues USP Certificates, which are stated not to describe completely the results obtained during evaluation. The assigned value appears on the label or the certificate, and a value of 100.0% cannot be assumed for any USP reference standard that does not carry one. Where no assigned value is given, the lot was not developed for quantitative use. An absent number is a statement of scope, not an omission, and setting it against a supplier's 99.2% does not create a comparison, it creates a category error.

The European leaflet carries a comparable boundary in its instructions: the standard is used as is and not dried, and once the container has been opened its entire content must be used immediately, further storage and re-use not being warranted. The state the number describes is a sealed container. Where a file compares a certificate issued years ago with a vial that has since been opened, resealed and stored, the second item may not be the material the document describes, and no arithmetic repairs that.

Decision rules

SituationWhat the documents supportAction
Identity differs by salt or hydrateNothing comparableApply the factor, or treat as a different item
Same measurand and basis, difference below combined uncertaintyNo demonstrable differenceInterchangeable for the stated use; record the comparison
Same measurand and basis, difference above combined uncertaintyA real differenceBridge the lots before switching
Different measurandsNo substitution decisionBridge, and state both numbers in the file
Uncertainty not stated on one documentOnly a limit checkBridge in every case

Where the table says bridge, the cheapest design is a ratio rather than a pair of values: both lots at the same nominal concentration, injected alternately in the same run against the same calibration, so that everything common to the two preparations cancels and the comparison no longer depends on either assigned value being correct.

Six questions before the comparison starts

Ask the supplier in writing, and keep the answers with the certificate: what exactly does the reported percentage measure; is it as is, dried, anhydrous or on the free base; if it is a mass balance, what are the individual subtractive terms; what is the expanded uncertainty and at which coverage factor; what is the assignment route and the traceability statement behind it; and what are the storage and handling instructions after opening. Five of the six fall under mandatory or recommended content in ISO 33401. The sixth, the subtractive terms, is the one that lets a reviewer check the arithmetic instead of trusting it.

None of these is answered by a catalogue page, and none requires a laboratory. They decide whether two documents can be compared at all, which is an earlier and different question than which number is higher. The guides below cover reading a single certificate, the uncertainty that follows it into solution, and the conversions that fail when a basis is assumed rather than read.

Reading a certificate of analysis | reference standard or analytical standard | salt and hydrate conversion | uncertainty budget for a standard solution | when the standard is the OOS suspect

Frequently asked

Two certificates for the same compound report different purities. Which one is wrong?
Neither, necessarily. Before the numbers can be compared, the identity, the measurand, the basis and the assignment route have to match. Where they do, the difference can be tested against the combined uncertainty. Where they do not, the two values are answers to different questions and one of them being lower says nothing about quality.
Can a chromatographic purity be compared with a mass-balance content?
No. Area per cent counts only what elutes and absorbs at the detection wavelength, so water, residual solvents and inorganic residue are invisible to it. On a vial carrying 0.15 per cent water, 0.05 per cent residue and 0.35 per cent chromatographic impurities, the two quantities read 99.65 and 99.45 per cent. The 0.20-point gap is the definition, not a disagreement.
How is a dried-basis content converted to an as-is basis?
Multiply by one minus the water fraction expressed as a decimal. A content of 99.5 per cent on the dried basis with 0.8 per cent water is 99.5 times 0.992, or 98.70 per cent as is. Both directions matter: two lots reading 98.6 as is and 99.5 dried are 0.10 points apart once converted, not 0.9.
Is a difference smaller than the combined uncertainty a real difference?
No. With expanded uncertainties of 0.4 and 0.3 at the same coverage factor, the combined expanded uncertainty is the square root of the sum of the squares, or 0.50 points. A measured difference of 0.5 points therefore sits exactly at its own uncertainty, and the documents do not support a claim that the two lots differ.
What if one certificate states no uncertainty for its assigned value?
Then a difference uncertainty cannot be computed, because the missing term cannot be entered into the sum of squares. The Ph. Eur. leaflet explains the omission as negligible against the limits of the method-specific assays. The comparison degrades to checking whether each value falls inside the required limit, and any substitution decision needs a bridging experiment rather than arithmetic.
Why does a USP reference standard certificate carry no purity figure?
USP issues certificates rather than certificates of analysis, and states that these do not describe the evaluation results completely. The assigned value appears on the label or the certificate, and 100.0 per cent may not be assumed where none is given, because that lot was not developed for quantitative use. The absence is a statement of scope rather than a gap to be filled.

Sources and further reading

Every document below is published by the body that owns it; none of them is a summary written by us. Accessed 2026-11-18.

Send us a CAS number

Availability, documentation and lead time — answered by the people who handle the batch, not a ticket queue.

Request a quote sales@rssechem.com

How to read a certificate of analysis Field by field: identity, purity, content basis, the retest date, and the questions worth asking a supplier before you file the document.How to choose a reference standard supplier A practical checklist for evaluating reference standard suppliers on documentation, traceability, lot consistency, logistics and technical support.Stable isotope-labelled standards and when to use them When labelled internal standards earn their place in an LC-MS or GC-MS method, what to specify when ordering, and the traps in quantitative work.Nitrosamine and NDSRI standards: what to test for and what to buy The difference between small nitrosamines and NDSRIs, what a laboratory needs as a standard for each, and how to specify one so the result stands up to review.What is an impurity reference standard? What the term means, how it differs from a pharmacopoeial standard and a working standard, what the documentation has to state, and how to pick the right form and purity basis.ICH M7 in practice: classifying mutagenic impurities How the five-class hazard assessment works, what the 1.5 microgram default really covers, and how the four control options decide which standards you buy.qNMR vs mass balance: how purity is really assigned Two certificates, one compound, two purity values. What mass balance and qNMR each measure, why the figures differ, and how the basis changes the number.Elemental impurities: from a PDE to a limit you can measure How a permitted daily exposure becomes a concentration limit, why the dose moves the number, and what an elemental impurity specification must state.Salt, hydrate and free base: converting content so the numbers reconcile How to move content and impurity values between as-is, anhydrous and free-base bases, in what order, and why the basis can turn a pass into a fail.Forced degradation study design: how far to push each stressor How to set stressor strength, read a kinetics screen, judge mass balance and catch over-degradation before it wastes a method development cycle.Analytical method transfer: acceptance criteria you can defend Four transfer routes, criteria built from the method's own precision, and the standard's assigned content that shifts every result by a fixed amount.Chiral and stereoisomeric impurity standards Enantiomer, diastereomer, epimer, E/Z isomer: four different standards and four different questions. What each has to state before it can control a method.Reference standard, analytical standard, CRM: what the words mean Four labels appear on the same vial in different catalogues. They are not synonyms, and the difference decides what you may claim in a report.Peptide reference standards: what changes when the analyte is a chain Deletion sequences, epimers, pyroglutamate, deamidation, counter-ion and water: why a peptide standard needs a different certificate from a small molecule.Cleaning validation residue limits: where the number comes from How a residue limit is derived from the dose, from 10 ppm and from a health-based PDE, and what the swab method and its standard must be able to reach.Cold chain and stability of reference standards: what transport and receipt checks protect Storage regimes, stability budgets and mean kinetic temperature, plus the receipt checks that decide whether a standard still supports its certificate.Impurity profile comparison: what a reference standard proves, and what it does not How reporting, identification and qualification thresholds set the scope of an impurity profile comparison, and where a standard stops proving anything.Weighing a reference standard: minimum weight and moisture How the 0.10% repeatability rule sets a minimum weight, why a hygroscopic standard is used as is, and which figures belong on the worksheet.Standard solution stability: how long a prepared standard can be used How to set a hold time for a standard solution: two baselines, six failure modes, a defensible study design and criteria fixed before the data arrive.Reporting, identification and qualification thresholds by dose How the 0.05, 0.10 and 0.15 per cent thresholds are banded by dose, where the milligram cap takes over, and what each limit forces you to test.USP 233 in practice: the J value and the four preparation routes How chapter 233 turns a limit into a J value, which of four preparation routes to use, and the acceptance criteria a method must meet before routine use.Building an uncertainty budget for a standard solution A worked budget for a 1 mg/mL standard solution: weighing, assigned value and volume, the correlation trap, and when k = 2 understates coverage.When the reference standard is the suspect in an OOS result How to test whether a reference standard caused an out-of-specification result, and how to separate a changed material from a misapplied assigned value.