Sampling Methods and Their Limits in Commodity Quality Inspection
Quote from chief_editor on June 7, 2026, 5:30 pmHow primary and composite sampling works in bulk commodity inspection, which ISO standards apply, and where sampling breaks down for heterogeneous or blended cargoes.
Sampling is the step that determines whether quality analysis in commodity trade reflects reality or artifact. A laboratory can produce results accurate to three decimal places, but if the sample delivered to that laboratory is not representative of the full cargo, the result is precise but meaningless. In bulk commodity trade, where quality parameters determine price, settlement, and contract performance, the sampling protocol specified in the contract is as consequential as the test methods.
Primary and Composite Sampling in Bulk Cargo
Bulk commodities present a specific challenge: the physical quantity to be characterized—often tens of thousands of tonnes—must be represented by a sample weighing a few kilograms. The statistical validity of any result depends entirely on how that sample was collected.
Primary sampling refers to the collection of individual increments from the cargo stream. For grain loaded by conveyor belt, a cross-belt or cross-cut mechanical sampler intercepts the full width of the belt stream at regular intervals, collecting increments that together represent all portions of the cargo loaded. For grain loaded by pneumatic system or ship's crane, mechanical sampling is not always practical, and manual probing of stockpiles or holds may be required—a method with lower statistical reliability.
ISO standards define sampling procedures for most major bulk commodities. ISO 13909 covers mechanical and manual sampling of hard coal; ISO 24333 governs cereals and cereal products; ISO 3082 applies to iron ore sampling. When a contract specifies one of these standards, both parties have a defined basis for comparing results if a dispute arises. Without a named standard, different inspectors may use different procedures and legitimately reach different results from equivalent cargoes.
The composite sample is assembled from all primary increments collected during loading. Depending on the cargo volume and commodity, a full composite may contain dozens or hundreds of increments. The composite is then divided—through a process called sample reduction—to produce a final laboratory sample of manageable size. This division must be done using a riffle divider or rotary sample divider that preserves the statistical properties of the full composite. Pouring off a portion of the composite by hand introduces bias that is difficult to detect but can shift results toward or away from specification limits.
Where Sampling Breaks Down
Several conditions make representative sampling difficult or impossible, and buyers should recognize them before signing contracts.
Heterogeneous cargoes resist representative sampling. A cargo of mixed metal scrap, industrial waste, or coal blended from multiple pits cannot be adequately characterized by a sample drawn from any one location or time point. The physical properties of the material vary from piece to piece in ways that a sample cannot fully capture. For such cargoes, buyers should specify multiple independent incremental sampling sequences across the full loading operation, with results reported separately by increment.
Small sample fractions create statistical uncertainty. A single 30-kilogram increment from a 50,000-tonne cargo represents 0.00006 percent of the total. Even with 50 increments across the full loading period, the aggregate sample mass is a tiny fraction of the shipment. This is unavoidable in bulk trade, but it means quality results carry inherent uncertainty ranges that are often not stated on the inspection certificate. GAFTA arbitration panels consider this when disputes involve parameters measured close to specification limits.
Sample integrity can be compromised between collection and analysis. A sample transferred through multiple hands without documented custody may be substituted, contaminated, or allowed to change condition—particularly moisture content—during transit to the laboratory. For high-value commodities or potentially disputed cargoes, sealed containers with tamper-evident markings and a documented chain of custody reduce this risk materially.
Split samples provide the strongest protection against result manipulation. When the composite is divided, one portion goes to the primary laboratory, a second is retained by the buyer's representative for independent analysis, and a third is sealed as a referee sample to be used if the first two results differ beyond the permitted tolerance. This three-sample structure is required in some GAFTA contract forms and should be specified in any contract where quality is a material pricing or acceptance factor.
Finally, sampling results reflect the condition of the cargo at the time of sampling. For commodities that change rapidly—fresh oilseed with high moisture, thermally reactive coal, or green coffee—a sample drawn at the start of loading may not represent the cargo loaded several hours later if conditions at the stockpile shifted during operations. Continuous incremental sampling across the full loading period is the only way to capture this variability, and contracts should require it rather than allowing a single-point composite.
How primary and composite sampling works in bulk commodity inspection, which ISO standards apply, and where sampling breaks down for heterogeneous or blended cargoes.
Sampling is the step that determines whether quality analysis in commodity trade reflects reality or artifact. A laboratory can produce results accurate to three decimal places, but if the sample delivered to that laboratory is not representative of the full cargo, the result is precise but meaningless. In bulk commodity trade, where quality parameters determine price, settlement, and contract performance, the sampling protocol specified in the contract is as consequential as the test methods.
Primary and Composite Sampling in Bulk Cargo
Bulk commodities present a specific challenge: the physical quantity to be characterized—often tens of thousands of tonnes—must be represented by a sample weighing a few kilograms. The statistical validity of any result depends entirely on how that sample was collected.
Primary sampling refers to the collection of individual increments from the cargo stream. For grain loaded by conveyor belt, a cross-belt or cross-cut mechanical sampler intercepts the full width of the belt stream at regular intervals, collecting increments that together represent all portions of the cargo loaded. For grain loaded by pneumatic system or ship's crane, mechanical sampling is not always practical, and manual probing of stockpiles or holds may be required—a method with lower statistical reliability.
ISO standards define sampling procedures for most major bulk commodities. ISO 13909 covers mechanical and manual sampling of hard coal; ISO 24333 governs cereals and cereal products; ISO 3082 applies to iron ore sampling. When a contract specifies one of these standards, both parties have a defined basis for comparing results if a dispute arises. Without a named standard, different inspectors may use different procedures and legitimately reach different results from equivalent cargoes.
The composite sample is assembled from all primary increments collected during loading. Depending on the cargo volume and commodity, a full composite may contain dozens or hundreds of increments. The composite is then divided—through a process called sample reduction—to produce a final laboratory sample of manageable size. This division must be done using a riffle divider or rotary sample divider that preserves the statistical properties of the full composite. Pouring off a portion of the composite by hand introduces bias that is difficult to detect but can shift results toward or away from specification limits.
Where Sampling Breaks Down
Several conditions make representative sampling difficult or impossible, and buyers should recognize them before signing contracts.
Heterogeneous cargoes resist representative sampling. A cargo of mixed metal scrap, industrial waste, or coal blended from multiple pits cannot be adequately characterized by a sample drawn from any one location or time point. The physical properties of the material vary from piece to piece in ways that a sample cannot fully capture. For such cargoes, buyers should specify multiple independent incremental sampling sequences across the full loading operation, with results reported separately by increment.
Small sample fractions create statistical uncertainty. A single 30-kilogram increment from a 50,000-tonne cargo represents 0.00006 percent of the total. Even with 50 increments across the full loading period, the aggregate sample mass is a tiny fraction of the shipment. This is unavoidable in bulk trade, but it means quality results carry inherent uncertainty ranges that are often not stated on the inspection certificate. GAFTA arbitration panels consider this when disputes involve parameters measured close to specification limits.
Sample integrity can be compromised between collection and analysis. A sample transferred through multiple hands without documented custody may be substituted, contaminated, or allowed to change condition—particularly moisture content—during transit to the laboratory. For high-value commodities or potentially disputed cargoes, sealed containers with tamper-evident markings and a documented chain of custody reduce this risk materially.
Split samples provide the strongest protection against result manipulation. When the composite is divided, one portion goes to the primary laboratory, a second is retained by the buyer's representative for independent analysis, and a third is sealed as a referee sample to be used if the first two results differ beyond the permitted tolerance. This three-sample structure is required in some GAFTA contract forms and should be specified in any contract where quality is a material pricing or acceptance factor.
Finally, sampling results reflect the condition of the cargo at the time of sampling. For commodities that change rapidly—fresh oilseed with high moisture, thermally reactive coal, or green coffee—a sample drawn at the start of loading may not represent the cargo loaded several hours later if conditions at the stockpile shifted during operations. Continuous incremental sampling across the full loading period is the only way to capture this variability, and contracts should require it rather than allowing a single-point composite.