Short answer: an approved sample protects you only if, months later, both parties can prove which sample it was, that nobody has opened it, and what the contract says it is for. Before production, give each approved sample a unique reference, split it so that buyer and seller each hold an identical jointly sealed counter-sample, and name that reference in the contract. State what the sample governs, set tolerances, agree how arrival samples are drawn, and name in advance the laboratory whose "umpire" analysis of a sealed spare will settle a disagreement. A sample kept loose in a drawer is usually just an opinion.
Three samples doing three different jobs
In a typical food purchase there are at least three samples, and each answers a different question.
| Sample | When it is drawn | What it is for | What it cannot prove |
|---|---|---|---|
| Approval sample (pre-production, reference or "type" sample) | Before the contract or before production | Shows the quality the seller offers; becomes the reference for qualities that are hard to write down, such as colour, appearance or cut | That the shipped lot will match it |
| Pre-shipment sample | From the finished lot, before or during loading | Checks the actual lot against the specification and the approval sample | Condition after loading and transit |
| Arrival (discharge) sample | At destination, on or after unloading | Checks what was delivered; the basis of any claim | What the goods were like when they left |
A counter-sample is the duplicate of any of these, sealed at the same time and held by the other party or a neutral custodian. A retention sample is one kept unopened purely for later reference or dispute. Whatever the name, each sample needs an identity, a seal, a custodian and a stated job in the contract.
Identifying and sealing a sample so it can be relied on months later
Trade-association rules give a benchmark even where they do not apply. Gafta's July 2026 Sampling Rules No. 124, effective for contracts dated from 1 July 2026, define "sealed" as samples jointly sealed by the buyers and sellers or their superintendents, sealed so as to prevent any access to the sample without breaking or removing the seal, and require seals that are tamper-proof and identifiable (Gafta, Sampling Rules No. 124, rule 3.10). Rule 8.1 requires labels showing, among other things, the commodity, the parties, the date and place of sealing, the sample numbers and the seal marks. FOSFA's contractual methods for oils, fats and oilseeds require the label to be affixed and sealed to the sample bottle, jointly where appropriate, and to show the FOSFA contract number (FOSFA, official sampling methods).
For an approval sample outside those rules, a workable minimum is:
- A unique reference on the label, repeated in the contract or purchase order.
- Identical portions. Draw one quantity, mix it and divide it into at least three portions: buyer, seller and an umpire spare. Portions taken separately are not counter-samples of one another.
- Joint, tamper-evident sealing with a numbered seal recorded on the label. A seal applied by one side alone proves little.
- A record: photographs with the label legible, and a signed note of who drew, divided and sealed the portions, where and when.
- Enough material for every analysis you may want, including a repeat. Gafta sets a minimum of 3 kg per sample (rule 9.1.2); ask your laboratory what your product needs.
Counter-samples: who holds what, and for how long
The two trade-association regimes show that custody and retention are written down, not assumed.
| Question | Gafta Rules No. 124 | Suggested approach for an approval sample |
|---|---|---|
| Who owns the samples? | Contractual samples "remain the property of both parties" (rule 9.3.2) | Each sealed portion is joint evidence, whoever holds it |
| Who holds them? | For grains, pulses, seeds and rice, one set consists of three samples—one for analysis and two spares—and all are held by the sellers or their representative (rules 9.5.1.1–9.5.1.2) | One portion each to buyer and seller; the spare with a neutral inspector or laboratory where possible |
| How long? | Retained by the superintendent for 3 months from sealing unless either party instructs otherwise in writing (rule 9.3.1) | Until the claim period and any dispute are over, counted from the last delivery that references the sample |
FOSFA's methods also set a minimum of three months, longer if circumstances so dictate, with storage in cool, dry conditions away from strong light, in a hygienic area free of insects and rodents (FOSFA, official sampling methods, section 5).
Retention is the term most often left blank. An approval sample sealed in March for a contract shipping in September is useless if it was discarded in June. Write the period into the contract, and store your portion as carefully as you expect the seller to store theirs: if your counter-sample has faded on a sunny shelf, the seller will say so.
Referencing the sample in the contract without making it the whole specification
The law may give a sample contractual weight whether or not you planned it, so decide deliberately what it is for.
- England and Wales, Scotland and Northern Ireland. Under section 15 of the Sale of Goods Act 1979, a contract is a sale by sample where there is an express or implied term to that effect, and there is an implied term that the bulk will correspond with the sample in quality and be free from any defect making the quality unsatisfactory which would not be apparent on reasonable examination of the sample (legislation.gov.uk, Sale of Goods Act 1979, s.15). The section does not apply to consumer contracts covered by the Consumer Rights Act 2015.
- International sales under the CISG. Where the UN Convention on Contracts for the International Sale of Goods applies (broadly, sales between parties with places of business in different Contracting States, art. 1(1)), article 35(2)(c) provides that, "except where the parties have agreed otherwise", goods do not conform unless they "possess the qualities of goods which the seller has held out to the buyer as a sample or model". Article 6 lets the parties exclude or vary the Convention (UNCITRAL, CISG text).
Other legal systems differ. Ask a trade lawyer which law governs your contract and what it implies about samples. Three drafting choices matter most:
- Say what the sample governs. For example: "Colour and appearance to be not inferior to sealed sample ref. HB-2026-014; all analytical parameters per the attached specification." Measurable parameters belong in a written specification with methods and limits, not in a sample.
- Say which prevails on conflict. If the sample happens to show 3% broken material and the specification allows 5%, which one binds? Write it down.
- Say which samples are contractual. An informal sample sent by courier before negotiations is not the approval sample unless the contract says so.
Tolerance: natural variation in agricultural products versus a failed delivery
A dried flower, seed or spice never matches a sample exactly, and two analyses of one lot rarely agree to the decimal. Codex's General Guidelines on Sampling, CXG 50-2004 (revised 2023), treat measurement uncertainty at length and note that acceptance procedures should include provisions for disputes (Codex Alimentarius, CXG 50-2004).
A usable tolerance clause separates three bands for each parameter:
- Within tolerance: accept at the contract price.
- Outside tolerance but within a rejection limit: accept with a price allowance calculated by a stated formula.
- Beyond the rejection limit, or any safety finding: the buyer may reject.
For visual qualities, state the comparison conditions: which sample, what lighting, who compares, and whether an instrument reading or a colour reference is used alongside the eye.
Arrival sampling: who draws it, how, and in whose presence
An arrival result is only as strong as its sampling. Agree in the contract:
- Who draws it: an independent inspector or superintendent, named or defined by qualification.
- How: the sampling method or plan, so that the arrival sample is drawn the same way as the pre-shipment sample.
- In whose presence: give the seller notice and the chance to attend or appoint a representative. Under Gafta rule 4.2, if one party is not represented for sampling, the other must call on a superintendent from another company to act for it (Gafta, Sampling Rules No. 124).
- When: promptly. Where the CISG applies, the buyer must examine the goods, or have them examined, "within as short a period as is practicable in the circumstances" (art. 38(1)), and loses the right to rely on a lack of conformity without notice specifying its nature within a reasonable time (art. 39(1)).
Arrival samples should be split and sealed exactly like the approval sample. Record lot, container and seal numbers against every sample and keep them with the shipment file; see the document set behind one shipment. For the difference between a sampling instruction and the scope stated on an inspector's report, see independent inspection certificates.
Disagreement: independent umpire analysis and how to name it in advance
When two results disagree, each side holds a number from its own portion. The usual way through is an analysis of a third sealed portion by a laboratory neither side controls, often called an umpire or referee analysis. Gafta's rules show how precise this can be. For feedingstuffs, biomass, marine and animal products, and malting barley, either party may, within 14 consecutive days of receiving a copy of the certificate, have another sealed sample analysed by a second approved analyst; the mean is final if the two vary by no more than 0.50%, and otherwise a third analysis is available, with the mean of the two closest results final and binding (rule 9.5.3.4). FOSFA's complaints procedure for member analysts requires a complainant alleging an analytical discrepancy to have the samples sent to the Federation's consultant chemist for safekeeping (FOSFA, Member Analysts Scheme).
Outside those rules, write your own version before the first shipment:
- Which sample: the sealed spare, with its reference and seal number, opened only in the umpire laboratory.
- Which laboratory: named, or defined as accredited to ISO/IEC 17025, which sets requirements for the competence, impartiality and consistent operation of laboratories (ISO, ISO/IEC 17025:2017), for the parameter in dispute.
- Which method: the method in the specification.
- Deadlines: for requesting the umpire analysis and for releasing the sample.
- What decides: the umpire result alone, or a stated combination of results.
- Who pays: for example, the party whose result is further from the umpire result.
- What happens next: the price allowance or rejection it triggers, and the arbitration or court route for anything the analysis cannot settle.
If payment is by letter of credit, remember that the bank deals in documents, not samples; any quality certificate the credit calls for should match these terms (see reading an LC before you accept it). If payment is released against pre-shipment inspection, seal that sample the same way; see advance payment: how much is reasonable.
Worked example: hibiscus colour and broken content
Hypothetical teaching example. Names, figures, limits and prices are invented for illustration. They are not real specifications, market prices or test results.
A buyer contracts 40 tonnes of dried whole hibiscus calyces at 2,400 CU per tonne, 96,000 CU in total. The contract sets broken material at a maximum of 5.0%, with an allowance of 1% of the invoice value for each percentage point above 5.0% (pro rata) up to a rejection limit of 10.0%, and colour "not lighter than sealed sample ref. HB-2026-014". On arrival the colour looks lighter, and the buyer's laboratory reports 8.2% broken material.
Scenario A: sealed samples exist. The approval sample was split three ways and jointly sealed. The arrival sample was drawn with the seller's representative present and split the same way. The seller's laboratory reports 4.1% on its portion. The contract names an umpire laboratory whose result decides, with the fee (450 CU) paid by the party whose result is further from it.
| Step | Figure |
|---|---|
| Buyer's claim at 8.2%: (8.2 − 5.0) × 1% × 96,000 CU | 3,072 CU |
| Umpire result on the sealed spare arrival portion | 6.9% |
| Allowance at 6.9%: (6.9 − 5.0) × 1% × 96,000 CU | 1,824 CU |
| Distance from umpire: buyer 8.2 − 6.9 = 1.3; seller 6.9 − 4.1 = 2.8 | Seller pays the 450 CU fee |
| Colour | Compared with sealed HB-2026-014 under the agreed conditions |
The buyer does not get its full claim, but the dispute ends with a 1,824 CU allowance and a colour comparison against a sample neither side could have altered.
Scenario B: no sealed samples. The approval sample sat unsealed in the buyer's office, and the arrival sample was drawn without notice to the seller. The seller says the reference has faded, the arrival sample is unrepresentative, and its own pre-shipment test showed 4.1%. The whole 3,072 CU claim is in dispute, and the colour complaint has no agreed reference. The realistic outcomes are a compromise, a costly arbitration, or nothing.
The goods were the same in both scenarios; the difference was paperwork done before production.
Common mistakes
- Approving a sample by email photograph and treating that as a reference.
- Keeping a single sample, so the other side can say it was altered.
- Discarding samples after three months when the claim period runs longer.
- Writing "quality as per sample" for everything, including parameters a laboratory should measure.
- Drawing arrival samples without notifying the seller.
- Leaving the umpire laboratory to be agreed after the dispute starts.
Limits and what to recheck
Gafta and FOSFA rules apply only where the contract incorporates them, cover particular commodities and are revised; check the current edition. Section 15 of the Sale of Goods Act 1979 applies across the United Kingdom, and the CISG only where its scope rules are met and the parties have not excluded it; other jurisdictions differ. This is general information, not legal or laboratory advice: confirm contract wording with a trade lawyer, sample sizes and storage with your laboratory, and sampling arrangements with your inspection provider.
FAQ
What is a counter sample? A duplicate of a contractual sample, drawn and sealed at the same time and held by the other party or a neutral custodian, so that either side's result can be checked later.
Is a reference sample the same as a counter sample? Not quite. A reference sample is the approved standard, such as for colour or appearance; a counter-sample is a sealed duplicate of any sample. An approval sample should itself be split into counter-samples.
How long should samples be retained? As long as the contract says. Gafta and FOSFA use three months as a baseline, but for approval samples the period should cover the last delivery plus the claim period.
Does a pre-production sample bind the seller? Only as far as the contract and governing law make it do so. State in the contract which samples are contractual and what they govern.
Next step
Before approving the next sample, make a one-page sample record: assign a unique reference, list the qualities it governs, record the seal numbers and custodians, set the retention period, and name the sampling method and dispute laboratory. Attach that record to the purchase order or contract, then have a trade lawyer check the wording and the laboratory confirm the quantity and storage conditions it needs.
Resources
- Gafta — Sampling Rules No. 124 (contracts dated from 1 July 2026) — https://www.gafta.com/media/5qynr511/124_2026.pdf (accessed 15 October 2026)
- FOSFA International — Sampling Methods, FOSFA International Official Methods (PDF, April 2017/2018) — https://www.fosfa.org/wp-content/uploads/2022/04/Contractual-Methods-of-Sampling.pdf (accessed 15 October 2026)
- FOSFA International — Member Analysts Scheme — https://www.fosfa.org/wp-content/uploads/2022/04/Member-Analyst-Scheme.pdf (accessed 15 October 2026)
- Codex Alimentarius (FAO/WHO) — General Guidelines on Sampling, CXG 50-2004 (revised 2023) — https://www.fao.org/fao-who-codexalimentarius/sh-proxy/es/?lnk=1&url=https%3A%2F%2Fworkspace.fao.org%2Fsites%2Fcodex%2FStandards%2FCXG+50-2004%2FCXG_050e.pdf (accessed 15 October 2026)
- UK Government (legislation.gov.uk) — Sale of Goods Act 1979, section 15 — https://www.legislation.gov.uk/ukpga/1979/54/section/15 (accessed 15 October 2026)
- UNCITRAL — United Nations Convention on Contracts for the International Sale of Goods (text and explanatory note) — https://uncitral.un.org/sites/default/files/media-documents/uncitral/en/19-09951_e_ebook.pdf (accessed 15 October 2026)
- ISO — ISO/IEC 17025:2017 General requirements for the competence of testing and calibration laboratories — https://www.iso.org/standard/66912.html (accessed 15 October 2026)
