When two laboratories report different results for the same bulk tea lot, hold the affected lot and compare the evidence before ordering another test. First establish whether the laboratories tested equivalent portions from the same representative sample, measured the same quantity with comparable methods and reported results on the same basis. Only then decide whether the difference is expected analytical variation, a sampling problem, an unresolved result or evidence of non-conformity.
A passing result does not automatically cancel a failing result. A third result from a convenient new scoop can make the file less defensible, not more.
Start with a seven-gate comparability record
| Gate | Evidence to align | Stop condition |
|---|---|---|
| 1. Decision | Legal, buyer or contract limit; parameter; product; release point | The reports answer different compliance questions |
| 2. Lot and sample | Lot, package population, sampling plan, composite, split IDs and seals | Either result cannot be linked to the same representative source |
| 3. Custody and condition | Sampler, dates, storage, transport, receipt condition and sample integrity | A split may have changed, leaked, contaminated or lost identity |
| 4. Measurand and method | Analyte definition, matrix, preparation, technique, method edition and validation | The laboratories did not measure an equivalent quantity |
| 5. Reporting basis | Units, as-received or dry basis, recovery correction, LOQ, rounding and significant figures | Numbers are being compared without a common basis |
| 6. Technical validity | Accredited scope, QC, blanks, recoveries, standards, repeat data and uncertainty | A result remains technically unsupported or out of control |
| 7. Decision rule | Limit, uncertainty treatment, acceptance zone, retest route and final authority | No agreed rule says what the numbers mean commercially |
Resolve the conflict in nine controlled steps
1. Freeze the lot and preserve both evidence chains
Block shipment, processing, blending, repacking or sale of the affected quantity unless a competent authority directs otherwise. Save each original report, laboratory correspondence, sampling record, seal record, chain of custody and retained portion. Record the time, owner and reason for the hold. Verify report provenance and revision status with the laboratory-report authenticity checklist.
2. Separate official control from a private buyer dispute
An authority result, import rejection or detention follows the destination authority's law and review route. A supplier COA disagreement is a private quality matter unless a regulator becomes involved. Codex CXG 70-2009 is written for disputes between importing- and exporting-country food-control authorities, excludes sampling questions and microbiological results, and assumes designated laboratories and controlled reserve samples. It is a useful reference model, not an automatic contract clause.
3. Reconstruct sampling before comparing decimals
Confirm that both test portions came from the same correctly mixed aggregate or equivalent controlled splits. Compare lot boundaries, packages selected, increments, mixing, reduction, sample mass, containers and seals. If one laboratory received a production composite and the other a single top-of-bag grab, the dispute begins with sampling. Use the representative sampling workflow; analytical precision cannot repair a biased field sample.
4. Put the results on one technical basis
Create a line-by-line comparison for the exact analyte or property. Align the matrix and product form, method and edition, preparation, unit, dry-matter or as-received basis, recovery treatment, reporting limit and rounding. For multiresidue work, match the enforcement residue definition and component calculation. For moisture or ash, do not compare results expressed on different bases. Ask each laboratory to explain conversions rather than recalculating silently.
5. Ask each laboratory to review technical validity
Request confirmation of sample receipt condition, method validation, accreditation scope, calibration or standards, blank and spike or recovery performance, internal QC, repeat analysis, relevant proficiency performance and the applicable measurement uncertainty. Codex's authority-dispute guideline lists these categories because exchanging the underlying technical information may resolve a disagreement without consuming a reserve sample. The buyer should request only records needed for the decision and handle confidential data appropriately.
6. Interpret uncertainty without erasing sampling risk
Every quantitative result has measurement uncertainty. ILAC G8 and the Eurachem/CITAC compliance guide explain that a decision rule defines how uncertainty is considered when stating conformity. A value near a limit may support a different decision from a value whose uncertainty interval lies clearly on one side.
For its specific government-to-government procedure, Codex CXG 70 gives a critical-difference comparison based on the expanded uncertainties of two results. Do not apply that formula blindly to a private tea claim: it does not resolve different samples, within-lot heterogeneity, biased sampling, incompatible methods or a missing contract rule. Ask the qualified laboratories or a competent statistician to interpret comparability for the actual data.
7. Decide whether any retest can answer the question
A retest is useful only when the material remains representative, identifiable, stable and sufficient. Prefer an unopened sealed reserve split created during the original sampling. Check its seal, custody, storage and remaining mass before opening. Do not repeatedly test new portions until one passes, average selected results after seeing them, or destroy the last agreed referee sample. The retained-sample plan should identify who may authorise opening and what evidence remains afterward.
8. Write the referee instruction before dispatch
Buyer and seller should agree the independent laboratory, sample ID, seal opening, witnesses if needed, method, preparation, unit, reporting basis, uncertainty, decision rule, report recipients, deadline, cost allocation and whether the result is contractually final. Verify that the chosen laboratory's current accredited scope covers the matrix and test using the laboratory selection checklist. If authorities control the case, obtain their approval rather than substituting a private referee result.
9. Record one disposition, not just another number
Close the review as one of five outcomes: results reconciled and pass; results reconciled and fail; not comparable because sampling or methods differ; technically unresolved pending an agreed action; or authority-controlled escalation. Link the decision to the affected quantity, approver, evidence and remedy. Confirmed or unresolved non-conformity should follow the nonconforming-lot disposition plan; commercial notice and remedies belong in the quality-claims workflow.
Keep five requirement types separate
- Legal requirement: the destination authority's applicable limit, official method, sampling procedure and review decision control within their scope.
- Voluntary guidance or standard: Codex, ILAC, Eurachem or ISO material provides a technical framework unless law or agreement gives it binding effect.
- Trade reference: a supplier COA, historical average or customary retest practice helps comparison but does not decide today's lot.
- Buyer specification: fixes the parameter, limit, method, sample link, report basis and private release rule.
- Contract requirement: allocates notification, reserve samples, referee testing, cost, finality and remedies between parties; it cannot overturn an authority decision.
Common buyer mistakes
- Choosing the more favourable number without explaining the difference.
- Calling two reports comparable because the analyte names look similar.
- Ignoring different lots, composites, dry-basis conversions, LOQs or recovery treatment.
- Using a new warehouse grab instead of the sealed original reserve.
- Ordering repeated tests until one passes.
- Treating measurement uncertainty as permission to waive a legal limit.
- Using the authority-focused Codex dispute route as if it automatically binds a seller.
- Consuming every retained split before an independent referee step.
- Releasing goods while the scope, sample identity or decision rule is unresolved.
Practical conclusion
Use the sequence hold - classify the dispute - reconstruct sampling - align methods and reporting - review validity - interpret uncertainty - protect the reserve - instruct the referee - authorise disposition. The goal is not to produce a third number; it is to explain why the evidence differs and make one controlled decision for the identified lot.
Before testing a proposed shipment, align Yunjing Tea's sample-to-shipment quality controls with the buyer's specification, laboratory and retained-sample plan. Then send the tea, destination, lot definition, test panel and required decision route early enough to agree the evidence chain before sampling.
Sources checked 11 October 2026: Codex Alimentarius CXG 70-2009, Guidelines for Settling Disputes over Analytical (Test) Results, adopted 2009; Codex CXG 83-2013, Principles for the Use of Sampling and Testing in International Food Trade, revised 2015; ILAC's current guidance index for G8:09/2019 on decision rules and G17:01/2021 on measurement uncertainty in testing; and the Eurachem/CITAC Use of Uncertainty Information in Compliance Assessment, second edition (2021). Recheck the destination authority, contract, laboratory scopes, methods and sample integrity for the actual lot.
