A Precise Test Cannot Rescue an Unrepresentative Sample

A laboratory can use a highly precise method and still produce a result that does not represent the lot. The instrument may measure its test portion correctly; the larger problem may be that the test portion never captured the variability of the material from which it came.

That is why a certificate of analysis should be read as a chain of evidence: lot → sample → preparation → test portion → result. If the sample is not connected clearly to the lot, later analytical precision cannot repair the missing link.

Five terms that should not be treated as synonyms

These terms describe different stages. A result obtained from one aliquot is not automatically evidence that every unit, vial or location in a lot is identical. The sampling plan is what connects the measured portion to the claim being made about the larger lot.

What makes a sampling plan informative?

For drug products, 21 CFR 211.160 requires scientifically sound sampling plans and says samples must be representative and properly identified. Section 211.165 adds that written procedures should specify both the sampling method and the number of units per batch to be tested.

Those provisions point to practical questions that are useful well beyond a regulatory checklist:

• Population: What exact lot or batch is the result intended to describe?
• Selection: Were units or locations selected by a predefined procedure?
• Coverage: How many units, containers or positions were sampled?
• Identity: Were the sample, lot and product labels preserved through testing?
• Handling: Could transport, storage, mixing or preparation change the measured attribute?
• Purpose: Was the plan designed for identity, assay, purity, microbiology or another specific question?

There is no universal sample count that answers every research question. A sensible plan depends on the material, expected variability, lot size, container configuration, analytical purpose and the consequence of making an incorrect conclusion.

Homogeneous is a conclusion that needs support

A material described as homogeneous is expected to be sufficiently uniform for the intended test. But uniformity should not be assumed merely because the powder, liquid or finished unit looks consistent.

Variation can arise among containers, within one container, during filling, after settling, through adsorption to surfaces or because of environmental exposure. A composite sample may provide a useful lot-level average, but it can also conceal unit-to-unit variation. Testing individual units may reveal distribution, yet provide less coverage if too few are selected. The correct design depends on the question.

For procurement review, it is reasonable to ask whether a COA result came from a single unit, a composite or multiple independent units. The answer changes what the reported number can support.

Retesting and resampling are different

Retesting generally analyzes another portion of the original homogeneous sample. Resampling examines additional units collected under the original sampling procedure or, when justified, a newly collected sample from the batch.

FDA’s OOS guidance explains that resampling should follow predetermined procedures and sampling strategies. If an investigation shows that the original sample was improperly prepared or the sampling method was inherently inadequate, that weakness should be documented and corrected. A new sample should not simply be used to make an inconvenient first result disappear.

For a deeper discussion of unexpected results, see One Result Outside Specification Is a Signal, Not a Diagnosis.

A procurement checklist for sampling context

When comparing suppliers or reviewing research peptides for sale, the following details help determine what a reported result actually represents:

• exact product name, strength or format, and lot identifier;
• date and location of sampling;
• who collected and sealed the sample;
• number of units or locations sampled;
• whether the tested material was a single unit, composite or pooled preparation;
• sample quantity and any mixing, dilution, extraction or reconstitution step;
• storage and transport conditions before testing;
• chain-of-custody or submission identifiers linking the sample to the report;
• whether retained material remains available if a discrepancy appears.

A laboratory report may not display every one of these details on its front page. The important point is whether the information exists in the underlying record and whether the final result can be traced back to the sampled lot without guesswork.

Read the result at the level the evidence supports

A precise number can create a false sense of certainty. Decimal places describe how a result is reported; they do not prove that the tested portion represents the whole lot. Before relying on a COA, separate two questions:

1. Was the test method suitable and performed correctly?
2. Was the tested sample representative of the population named on the report?

Both questions matter. Strong analytical work and strong sampling design are complementary, not interchangeable.

Primary sources

• Electronic Code of Federal Regulations, 21 CFR 211.160, General requirements (page current through October 15, 2026; accessed October 17, 2026).
• Electronic Code of Federal Regulations, 21 CFR 211.165, Testing and release for distribution (page current through October 15, 2026; accessed October 17, 2026).
• U.S. Food and Drug Administration, Investigating Out-of-Specification (OOS) Test Results for Pharmaceutical Production (October 2006; accessed October 17, 2026).

Editorial method: Published by bwell. AI tools assisted with organization and copyediting; the article was checked against the linked primary sources and the current bwell site. This educational content does not provide medical, clinical or legal advice.