Representative sampling of large kernel lots I. Theory of Sampling and variographic analysis

Official testing and sampling of large kernel lots for impurities [e.g., genetically-modified organisms (GMOs)] is regulated by normative documents and international standards of economic, trade and societal importance. The focus nearly always includes only analytical issues - omitting, with very few exceptions, proper accounting for sampling errors. With total sampling errors for irregularly distributed contaminants and impurities typically 10-100 times larger than analytical errors, this issue is critical for procedures based on general notions of effective material uniformity. When the focus includes sampling, most guidelines recommend sampling plans based on the assumption that kernel-lot impurities, if present, are randomly distributed. The only exceptions are EC Rec. 787/2004 and prCEN/TS 1568 (2006), which suggest sampling strategies suitable for more heterogeneous situations.A recent field project, KeLDA, documented highly significant heterogeneity in 13 out of 15 randomly chosen soybean kernel shiploads arriving in Europe intended for the feed market. The KeLDA study argued strongly that only sampling guidelines taking this into account can be viewed as authoritative for kernel-lot testing. The Theory of Sampling (TOS) is the only fully comprehensive, scientifically documented approach for representative sampling of all types of heterogeneous lots and materials (trace constituents, contaminants), and, in this context, GMO-contaminated lots constitute no special type.In this three-part series, we re-interpret KeLDA data from a proper TOS perspective. Part I introduces the fundamental principles for process sampling, resolves terminology differences between TOS and ISO usages and defines variographic analysis in the full detail necessary for parts II and III