A 96-sample sequence fails at injection 41. Inlet pressure drops from 12.4 bar to 8.1 bar within a single run, and only one of the last twelve vials produces a clean chromatogram. The supplier’s certificate says the lot passed a 0.8 mm septum puncture test. The lab’s contract says the supplier warrants that goods are “free from defects in material and workmanship.” Both statements are true. Neither tells you who pays for the 96 samples, the instrument downtime, or the re-run. That gap is what analytical testing supply contract terms and liability actually have to close, and most purchase orders leave it open.

The clauses that decide outcomes are not the ones procurement negotiates hardest. Price, lead time, and payment terms get attention. Acceptance criteria, latent defect windows, lot recall obligations, and failure-cost allocation get a paragraph of boilerplate copied from a different industry. This article walks through the four clause families that decide who pays when a run fails, with the specific numbers and test methods that make each one enforceable. It assumes you already know what the consumables do. The question here is what the contract says when they do not do it.

“Conforms to specification” is not an acceptance criterion. It is a deferral. A criterion you can enforce names the measurement, the instrument class, the sample size, and the pass threshold. If the clause cannot be executed by a third party who has never spoken to either party, it will not survive a dispute.

Take septa as the worked case. A usable acceptance clause reads: puncture resistance tested per the supplier’s stated method on 30 pieces from the lot, no piece to require more than 4.5 N penetration force at 0.8 mm needle gauge, with a coefficient of variation across the 30 pieces below 15%. That clause does four things a warranty does not. It fixes the sample size, so the supplier cannot test three pieces and call it a lot. It fixes the needle gauge, because penetration force depends on it. It sets a dispersion limit, because a lot with a good mean and a bad tail is the lot that jams autosamplers. And it names a method, so the result is reproducible.

For vials, the acceptance criteria that matter are dimensional and are usually already on the drawing. Neck finish must conform to the GPI 13-425 or GPI 8-425 thread standard depending on the format, with thread pitch and root diameter measured on a go/no-go gauge rather than calipers. Crimp-neck formats should be checked against the closure drawing at the crimp dimension, not the glass body. The HPLC vial specification sheet is the right place to start, but the contract needs to say which of those numbers are acceptance-tested per lot and which are design targets.

For columns and tubing, acceptance criteria should reference the certificate of analysis and a defined performance test. A typical enforceable clause for an HPLC column is: retention time and asymmetry factor within 10% of the reference chromatogram supplied at order, tested on the buyer’s system within 14 days of receipt. That gives both sides a fixed reference and a fixed window.

A decision checklist for writing the clause

The last row is the one procurement tends to leave vague. “May be rejected” gives the supplier an argument. “Reject and replace within 10 working days at supplier cost” gives the buyer a remedy with a clock on it.

Acceptance criteria also need a defined incoming test on the buyer’s side. If the contract says the buyer must inspect within a stated period, that period is the buyer’s only chance to catch a lot-level problem before it becomes a latent defect argument. Ten working days is common. For consumables with a long shelf life and low incoming test throughput, 30 days is more realistic and worth negotiating.

The latent defect window is a clock, and it starts at a defined event

A latent defect is a defect that a reasonable incoming inspection would not have found. The contract question is not whether latent defects exist. It is how long the supplier remains liable for them, and what event starts the clock.

Three start events appear in supply contracts, and they are not equivalent:

• Date of shipment. Worst for the buyer. A lot that sits in a warehouse for six months before first use has burned half its window before it is opened.
• Date of receipt. Better, but still penalizes slow-moving stock. A septum lot used at 200 pieces a month may not reveal a molding problem until month four.
• Date of first use, or date the defect could reasonably have been discovered. Best for the buyer, and the version worth pushing for on consumables with shelf lives of two years or more.

The window length should track the failure mode, not a single number applied to everything. A reasonable structure:

• Dimensional nonconformance: 30 days from receipt, because a go/no-go gauge catches it immediately.
• Material and molding defects (septum coring, closure cracking, liner delamination): 12 months from first use or 24 months from shipment, whichever is later, capped at the lot’s stated shelf life.
• Performance drift in a column or a lot of tubing that only shows up after equilibration: 90 days from first use.

The “whichever is later” construction matters. Without it, a lot with a 24-month shelf life received on day 700 has a 30-day window. With it, the buyer has a full 12 months of use. Suppliers will resist open-ended windows, and they are right to. Capping at shelf life and requiring the buyer to preserve retained samples is the fair compromise.

Preserved samples are the evidence the clause depends on

A latent defect claim with no retained sample is a conversation. A claim with a sealed, labeled retention sample from the same lot, stored under the conditions on the certificate, is a case. The contract should require the buyer to retain at least 30 pieces or 1% of the lot, whichever is greater, for the full latent defect window. It should also require the supplier to retain its own lot retention for the same period and to make it available on request. If the supplier’s retention is gone, the burden of proof shifts in the buyer’s favor, and the contract should say so.

Storage conditions belong in the clause too. A septum lot stored next to a solvent cabinet is not the same product as one stored at 20 °C in a sealed bag. The certificate of analysis should state storage conditions, and the latent defect clause should be void if the buyer’s storage deviates from them. That is a fair term and it makes the claim defensible.

Lot recall obligations: who owns the downstream mess

A recall clause answers one question: if a lot turns out to be nonconforming after it has been distributed into the lab’s workflow, who pays to find and quarantine the affected material, and who pays for the analytical work that has to be repeated?

The trigger for a recall should be defined, not left to “materially defective.” A workable trigger is a lot-level failure of any acceptance criterion confirmed on retained samples, or a supplier-initiated field action. Once triggered, the clause should specify:

• Notification within 24 hours of the supplier’s determination, in writing, identifying lot numbers and ship dates.
• Buyer’s quarantine obligation within a stated period, typically 5 working days, with the supplier covering the labor cost at a pre-agreed rate or as a credit.
• Replacement or credit at the buyer’s election, with freight paid by the supplier.
• Traceability data the supplier must provide: lot number, manufacturing date, molding cavity if available, and the ship-to list for the affected lot.

The cavity number is the detail most buyers never ask for and most suppliers can provide. If a lot fails because one cavity in a multi-cavity mold is out of tolerance, the cavity number tells you whether the rest of the lot is at risk. The chromatography consumables supply chain usually has this data at the molder, even when it does not reach the certificate. Ask for it in the contract, and ask for it on the certificate.

What a recall clause usually does not cover, and should, is the analytical consequence. If a failed septum lot caused 40 rejected injections, the recall clause should address the cost of the re-run. Most suppliers will not accept full consequential damages. A middle position is a pre-agreed re-run credit: a fixed credit per affected injection or per affected sample, stated in the contract, capped at some multiple of the lot value. That is easier for a supplier to sign than open-ended damages and easier for the buyer to collect than a consequential loss claim.

The clause that decides who pays when a run fails: failure-cost allocation

This is the clause procurement skips and the one that matters most. When a run fails, four costs appear: the consumable, the samples, the instrument time, and the labor. The contract should allocate each one to a party and a trigger.

The default in most supply contracts is that the supplier replaces the consumable and nothing else. That leaves the buyer holding the sample cost, the instrument time, and the labor, which are usually an order of magnitude larger than the consumable. A run of 96 samples on an HPLC with a 20-minute cycle is roughly 32 hours of instrument time. At a fully loaded cost of $150 per instrument-hour, that is $4,800 before anyone touches the data. The vial that caused it cost $0.40.

Allocation table

The key word in the trigger column is “traced.” The contract must define what tracing requires, or every claim becomes a dispute about causation. A workable definition: a failure is traced to a consumable when the failure reproduces with a retained sample from the same lot and does not reproduce with a control lot under the same method and instrument conditions. That is a test the buyer can run and the supplier can verify. It also protects the supplier from claims caused by method drift or instrument problems, which is a legitimate concern.

The cap matters as much as the allocation. A supplier will not sign unlimited consequential liability on a $0.40 vial, and asking for it ends the negotiation. A cap of 10 to 20 times the lot value, or a fixed dollar ceiling per incident, is a realistic ask. So is a schedule of pre-agreed credits that avoids the need to prove damages at all.

Warranty, indemnity, and the exclusions that swallow them

The warranty clause and the failure-cost clause do different work. The warranty says the goods conform. The failure-cost clause says what happens when they do not. A contract with a strong warranty and no failure-cost allocation produces replacement parts, not compensation.

Read the exclusions before you read the warranty. The standard exclusions in analytical consumables contracts are:

• Improper storage or handling by the buyer. Fair, but it should be tied to the certificate’s stated conditions, not left to the supplier’s judgment after the fact.
• Use outside the stated application, for example a septum rated for one puncture used in a multi-injection workflow. Define the rated use in the contract, in numbers.
• Normal wear and degradation within shelf life. This is the exclusion that swallows the warranty if it is not bounded. “Normal” needs a definition, and the shelf life and storage conditions are the definition.
• Consequential damages, excluded in most supplier templates. This is the exclusion that makes the failure-cost clause necessary. If the supplier will not lift the consequential exclusion, negotiate the pre-agreed credit schedule instead.

Indemnity is a separate track and usually covers third-party claims, not the buyer’s own re-run costs. Do not expect indemnity language to solve a failed run. It will not.

Two diagnostic examples

These are hypothetical examples, constructed to show how the clauses interact. They are not reports of actual cases.

Example 1: the septum lot that cored

Symptom. A lab running 96 samples per sequence sees intermittent needle blockage and a rising baseline in the last third of the run. The autosampler reports three failed injections in one sequence.

Useful measurement. Penetration force on 30 septa from the suspect lot, measured at 0.8 mm needle gauge, against 30 from the previous lot. The suspect lot shows a mean of 5.8 N with a CV of 22%; the control lot shows 3.9 N with a CV of 9%. Retained samples from the suspect lot reproduce the coring under the same method.

Corrective action. Quarantine the lot, invoke the latent defect clause with the retained samples as evidence, and file the failure-cost claim for the affected sequence. The contract’s pre-agreed credit per affected run covers the instrument time. The replacement lot is accepted only after the incoming penetration test passes. Without a defined acceptance test and a retained-sample requirement, the claim would rest on a description of the symptom and would likely settle at the cost of the vials.

Example 2: the column that drifted inside the window

Symptom. A new HPLC column passes the supplier’s certificate but shows a retention time shift of 8% against the reference chromatogram by day 20, with asymmetry factor moving from 1.05 to 1.4.

Useful measurement. Retention time and asymmetry on the buyer’s system at day 0, day 10, and day 20, using the reference chromatogram supplied at order. The drift is reproducible across two mobile-phase lots, which rules out the mobile phase.

Corrective action. The 90-day performance-drift window covers the claim. The buyer submits the three time points and the reference chromatogram, and the supplier replaces the column under the performance clause. If the contract had used a 30-day dimensional window only, the claim would have expired before the drift was visible.

Operational precautions before the contract is signed

Clauses only work if the lab can execute the tests they depend on. Before signing, confirm four things:

• The incoming test is defined and resourced. If the contract requires a 30-piece penetration test per lot, someone has to run it. If the lab cannot, negotiate a supplier certificate plus a reduced buyer spot check, and say so in the clause.
• Retained samples have a home. A labeled, sealed retention sample stored under the certificate’s conditions is the evidence the latent defect clause depends on. Assign the storage location and the retention period in the contract, not in a lab notebook.
• Traceability data is specified. Lot number, manufacturing date, and cavity number where available. Ask for it on the certificate. A lab vial supplier that cannot provide cavity-level traceability is telling you something about its quality system.
• The method reference is current. If the acceptance test cites a method version, the version belongs in the contract. Methods get revised, and a clause that cites a superseded version is a clause the supplier can argue around. The U.S. EPA SW-846 series is a useful reference point for how method versions are controlled and why the version number belongs in writing.

It also helps to know what the supplier’s own reference data looks like. Technical resources from column and consumable manufacturers, such as the Agilent chromatography consumables and columns documentation and the Restek technical resources, show the format a reference chromatogram and a certificate of analysis should take. A contract that requires the supplier to supply data in that format is easier to enforce than one that requires “a certificate.” Sample handling and consumable specifications from Thermo Fisher chromatography and sample handling are a reasonable benchmark for what a specification sheet should contain. If the supplier’s sheet is thinner than the benchmark, the acceptance clause is doing more work than the supplier’s documentation supports.

What the contract should say, in one paragraph

Acceptance criteria should name a test method, a sample size, a threshold, and a dispersion limit. The latent defect window should start at first use or the date the defect could reasonably be discovered, whichever is later, and run for 12 months, capped at shelf life, with a retained-sample requirement on both sides. The recall clause should trigger on a confirmed lot-level failure, require notification within 24 hours, and include a pre-agreed re-run credit. The failure-cost clause should allocate the consumable, the samples, the instrument time, and the labor, define “traced” as a reproducible failure on retained samples that does not reproduce on a control lot, and cap the total at 10 to 20 times lot value or a fixed dollar ceiling. Read the exclusions before the warranty, and treat “normal wear” as a term that needs a number attached to it.

The decision rule is simple. If a clause cannot be executed by a third party who has never spoken to either party, it is not a clause. It is a sentence that will be read aloud in a meeting after the run has already failed. Write the test, the sample size, the window, and the cap. Then the question of who pays has an answer before the sequence starts, not after injection 41.

For the broader picture of how these supply terms sit inside an analytical testing program, the analytical testing resource pages are the place to start. The contract is the part of the program that decides what happens when the method, the instrument, and the consumable disagree. Get the four clauses right and the disagreement has a resolution path. Leave them as boilerplate and the resolution path is a phone call.

References & further reading

Related resources from our lab consumables network:

• Analytical Testing at testinganalytical.com
• Hplc Vial at vialforhplc.com
• Chromatography Consumables at chromatographyconsumables.com
• Lab Vials at labvials.com

External references and standards:

• Agilent chromatography consumables and columns
• Thermo Fisher chromatography and sample handling
• U.S. EPA SW-846 test methods for waste and materials
• Restek technical resources for chromatography