2.4 Sampling, Assaying & QA/QC

Sampling, assaying and quality assurance/quality control form a measurement chain linking geological material to engineering decisions. Sampling selects material to represent a defined target; preparation reduces it to a form and quantity suitable for testing; assaying measures specified constituents. A reliable result requires evidence about the whole chain, not merely a laboratory number with several decimal places.^[1]

The same chain supports exploration, resource evaluation, grade control and production accounting, but its design changes with the question. Establishing whether an element is present is different from estimating the average grade of a mining block or the composition of a product shipment. The required property, material boundary and acceptable uncertainty should therefore be defined before choosing a sampling and analytical procedure.

From the Decision to the Sampling Target

The sampling target is the material about which a conclusion is required. It may be an identified core interval, a shipment or a material stream over a stated period. The measurand is the quantity intended to be measured, including necessary qualifications such as the element, material, moisture basis and extraction procedure. A laboratory measurement of the submitted sample is not automatically a measurement of the larger target.^[1]

For geological estimation, individual samples also have a spatial support: their size, shape and orientation. Core intervals, channels and production samples may cover different supports and cannot simply be treated as interchangeable observations. CIM's 2019 estimation guidance addresses sample support and compositing separately from laboratory quality control.^[2]

This distinction has practical consequences. An assay pulp is the pulverised laboratory powder, not a mineral-processing slurry. Additional assays on one retained pulp may improve knowledge of that pulp's analytical variability, but they do not add new geological locations. Conversely, extra locations will not correct a consistently biased analytical method. Investigation and testing budgets should address the part of the chain that limits the intended decision.

Representativeness Through Collection and Preparation

Mineral distribution, particle size, density differences and segregation can affect which constituents enter a sample. A collection procedure should not preferentially select convenient fragments or omit a material fraction that belongs to the target. Minnitt's review of South African minerals operations documents how errors can arise at collection, extraction and subsequent handling; its operational examples illustrate mechanisms rather than universal sampling specifications.^[3]

The sample then passes through further selections. Crushing may precede splitting, followed by pulverising and withdrawal of a test portion. Every reduction in mass needs to remain compatible with the particle size and heterogeneity at that stage. Material lost, contaminated or excluded upstream cannot be restored by a more sensitive instrument. CIM's exploration guidance accordingly links representative collection, preparation and retained material with documented responsibility and sample security.^[4]

Physical preparation also requires checks. For example, a particle-size test can establish whether a crushed or pulverised product meets the specified preparation target. It does not, by itself, establish unbiased splitting or absence of contamination. Separate checks are needed for those properties. ALS's preparation quality-control note describes sizing checks and preparation duplicates as different controls, not substitutes for each other.^[5]

Choosing an Appropriate Assay

An assay method includes more than the final instrument. Its result depends on the test portion, decomposition or extraction, separation steps, measurement technique and reporting basis. Validation establishes whether a method is suitable for its intended use; verification examines whether an adopting laboratory can achieve relevant performance. Neither concept is satisfied merely by selecting a familiar instrument name.^[6]

The distinction between total and extractable concentrations is particularly important. ALS describes aqua regia as a partial digestion and four-acid digestion as near-total, with some resistant minerals still incompletely decomposed. A result from one decomposition therefore need not equal a result from another, even when both use the same instrumental finish.^[7]

Analytical requirements should cover expected concentrations, important impurities and the material matrix. Detection capability addresses low concentrations, while the working range also has an upper boundary. Results beyond the validated range require an appropriate additional procedure rather than unqualified extrapolation. A result reported below a limit is qualified information, not proof that the constituent is absent.^[6]

Assurance, Controls and Their Different Roles

Quality assurance establishes confidence through planned responsibilities, suitable procedures, training, records and review. Quality control uses operational checks and corrective responses to fulfil specified quality requirements and monitor continuing performance. In mining, this extends beyond a laboratory's own controls to the way the project collects, handles and accepts its information.^[6][4]

Certified reference materials assess agreement with assigned values for specified properties. Blanks examine contamination or background contributions within the stages they encounter. Duplicate samples assess variability across the stages actually repeated. A coarse duplicate taken after crushing, for example, does not independently repeat the original field collection or shared crushing operation. These controls answer different questions, so success with one cannot establish success with all.^[2][5]

Controls need to cover the materials, concentrations and procedures relevant to the project. Their frequency, acceptance criteria and review arrangements depend on the investigation and the consequences of error. The cited CIM documents are Canadian professional guidance, not a single worldwide prescription for insertion rates or permissible differences. Their useful organizing principle is to integrate controls into ordinary work and document both the findings and the response.^[4]

Traceability and Data Acceptance

A result needs a traceable connection to its location or lot, original sample identifier, preparation history and laboratory certificate. Method codes, units, reporting limits and qualifications belong with the value. Database checks should identify missing or duplicated records, interval errors and inconsistent conversions without silently replacing the source information. CIM's database guidance also calls for explicit selection of the assay used when more than one result exists.^[2]

This documentary trail differs from metrological traceability: the connection of a measurement result to a stated reference through a documented calibration chain, with associated uncertainties. Sample custody and identifiers preserve provenance but do not establish that calibration relationship.^[8]

Acceptance is a decision about fitness for a stated use. A failed control requires investigation of its cause and the affected samples; it does not automatically demonstrate that every result in a campaign is wrong. Equally, an apparently satisfactory batch average can conceal an isolated contamination event or a change in method performance. Reviews should retain sequence information and distinguish individual anomalies from sustained patterns.

Where reanalysis is justified, the repeat should address the suspected stage. Measuring the same solution again cannot test an earlier splitting error. A corrected data release should retain the superseded result and the reason for replacement, so later users can reconstruct the decision rather than see only a preferred number.^[2]

Measurement uncertainty is not an allowance for gross mistakes such as mislabelling, transcription errors or major misuse of a protocol. Those problems require investigation and correction, not merely a wider uncertainty interval.^[1]

Feedback to Geological and Operational Decisions

Consider a hypothetical campaign in which reference materials perform consistently but field observations reveal loss of fine material during collection. The laboratory controls support the analytical part of the process; they do not establish the composition of the lost fraction. The appropriate review concerns collection conditions, affected intervals and whether additional material or independent observations can resolve the uncertainty. Repeating the same pulps alone would not answer that question.

The final interpretation should state both what has been checked and what remains uncertain. Sampling uncertainty, analytical performance and geological coverage are related but separate. Keeping those distinctions visible allows corrections to target the actual failure mechanism and prevents precise-looking results from carrying more engineering confidence than the evidence supports.^[1][2]