TABLE OF CONTENTS The first steps in any sampling investigation are audit and assessment: find out what is going on and whether the current sampling variation is acceptable. If not, then some way must be found to reduce it. This would be easier if the total variation could be broken down and the component parts addressed separately. Pierre Gy's theory does this. Gy (1992) decomposes the total variation into seven major components (sources). He calls them errors because sampling is an error-generating process, and these errors contribute to the nonrepresentativeness of the sample. The seven errors are as follows:
fundamental error (FE)
grouping and segregation error (GSE)
long-range nonperiodic heterogeneity fluctuation error (shifts and trends)
long-range periodic heterogeneity fluctuation error (cycles)
delimitation error (DE)
extraction error (EE)
preparation error (handling) (PE)
These errors are introduced below, with descriptors matching those in the rest of this book.
The word homogeneous is defined in Merriam Webster's Collegiate Dictionary, Tenth Edition, [ ] as "consisting of uniform structure or composition throughout." To appreciate the variation that poor sampling can generate, we must realize and accept the fact that no material is homogeneous. Everything is heterogeneous, even if only at the molecular level. The constitution (or makeup) of the material causes it to be heterogeneous. Gy calls this the constitution heterogeneity (CH). It represents the differences between particles or molecules. The CH of solids is influenced by particle size, shape, density, chemical composition, and...
