With a mixed population of large and small units selecting only large units would give a sample that is typical of the large units but not representative of the whole population. To obtain a representative sample, the population should first be divided into two separate groups (strata) of large and small items. These groups should then be sampled separately. In sampling from a bulk material, homogenization of the material prior to sampling is likely to be required. This means that tanks must be vigorously stirred and powders must be well mixed.

For solutions in tanks we must select a representative sample, measure the concentration of nuclear material in that sample, and then measure (or know) the volume of the tank.

Any non-homogeneous item or mixture may be subject to matrix effects. In this context matrix refers to the non-nuclear part of nuclear material and items. These matrix materials can, in some cases, influence the measurement result. For example, the presence of hydrogen or fluorine in a matrix can have an impact on results obtained through the use of a neutron coincidence counting techniques. Another example, the presence of strong photon absorbers like steels or lead in scrap could influence gamma-ray measurement techniques. These matrix effects must be understood and accounted for in the measurement. This is most often done via calibration and calculation.

Page 16 / 17