Grain sampling methods to achieve consumer confidence and food safety


Cereals & Oilseeds
Project code:
01 December 2002 - 31 March 2003
AHDB Cereals & Oilseeds.
AHDB sector cost:
£14,797 from HGCA (Project No 2912).
Project leader:
D ARMITAGE Central Science Laboratory, DEFRA, Sand Hutton, York YO41 1LZ



About this project


Grain is sampled for one, possibly, two reasons:

  • to determine the average quality based on a representative sample
  • to detect contaminants.

This information may trigger rejection of a load or determine whether the levels of a contaminant in a bulk exceed a regulatory threshold. 

This review examined published sampling regimes for a number of contaminants that now, or in the future, may affect food safety. These were: genetically modified organisms, mycotoxins, micro-organisms, heavy metals, agrochemical residues and arthropod pests. 

A common, ISO 13690-based approach using 5-11 samples from 15-500t is used for some contaminants. This is reasonable when the contaminant (analyte) is homogeneously distributed, although sample size and number of samples/tonne are often poorly defined, standards used may have no scientific basis and there are rarely validation details. Sampling regimes for insects and in particular, mycotoxins, which are heterogeneously distributed, are very demanding. 

An EU directive recommends taking 100 samples of 100g from 50-1500t for regulatory purposes for mycotoxin determination. Live insect contamination is best estimated using traps but the regulatory mycotoxin sampling may be adaptable for sampling insects and would also apply to dead insects.

It is recommended that the ISO-based sampling method be validated experimentally for a variety of analytes since it is widely cited and convenient to use. However, its effectiveness is unclear. A simplified method of sampling for mycotoxins should be developed that will give results comparable to the regulatory sampling. This would demonstrate 'due diligence' and avoid costs associated with the 'regulatory' sampling protocols. (If such a method proved 'substantially equivalent' to the regulatory sampling, it could be used for regulatory processes if there were cost savings). The pattern of mixing grain during outloading has considerable influence on the distribution of analytes. It therefore affects the sampling regime and merits study.