Sampling is the process by which a representative fraction of a lot or material stream is extracted through an equiprobable (equal-probability) probabilistic selection of increments from the lot or stream. The objective is to obtain a sample whose expected value equals that of the lot and whose variance is controlled and quantified, so that the characteristics measured in the sample statistically reflect those of the whole.

Pierre Gy defines sampling as a sequence of selective and non-selective operations that culminates in obtaining a representative portion of the lot.

A good analysis cannot correct poor sampling.
Representativeness is not declared—it is engineered, validated, and audited.

This aligns directly with Gy’s Theory of Sampling (TOS), in which the Fundamental Sampling Error (FSE) is unavoidable, but can be minimized through proper design.

SAMPLING OBJECTIVES

Sampling is the basis of metallurgical and commercial control: a sample must represent the stream or lot so that the measured grade, particle size distribution, moisture, density, and/or composition are reliable for mass balances, process control, metallurgical reconciliation, and commercial settlement (NSR).

In real operations, sampling fails when “convenient” (non-equiprobable) increments are taken, when the cutter does not intercept the entire stream, or when uncontrolled segregation/stratification occurs. In an industrial framework, the goal is to reduce bias and variance through: (i) correct primary sampling, (ii) proper preparation/homogenization, (iii) representative splitting, and (iv) full system traceability.

JKL application: JKL equipment is conceived as a complete system (not isolated pieces): primary sampler + handling/transfer + homogenizer/splitter + secondary sampling (Vezin or equivalent) + collection/labeling + audit and validation interfaces under applicable ISO standards and current sampling procedures.