An influential view in soil microbiology is that the microbial diversity of the soil is essentially unstructured spatially. This has largely been driven by research that has set out to map the spatial variability of microbial processes in soil. However, it is well-established that sampling schemes designed to map variation are inefficient at estimating the spatial structure of that variation, and the general conclusion that has been drawn, that "everything is everywhere", is consequently unsound. Spatially nested sampling of the soil is being done with our basic sampling units, adjacent peds (aggregates of 0.1-0.2g soil). These are nested in a sampling scheme that allows us to estimate components of variance and covariance for different spatial scales and our design will consider scales up to a few hundreds of metres. Each sample will be extracted and analysed using DGGE and PCR, which allows the spatial structure of both general diversity measures and abundances for particular groups to be analysed. The results of this work will not only be of general interest in soil ecology, but of considerable practical significance. They will allow us to give confident answers to the important question of how to sample the microbial diversity of fields or experimental plots in order to evaluate the effects of treatments or management practices. In the longer (10-year plan) the nested sampling approach will be ideally suited to research on the much more complex grassland system of the Park Grass Experiment.