Abstract
Summary: The assumption that the variation of soil can be characterized by the variation of a set of relatively few soil properties is examined. Different sets of properties describing the same soil profiles, in particular those measured in the field, and those measured in the laboratory, are used to group these profiles by principal components analysis. This numerical method is used so that each set is grouped by exactly the same procedure. Data from the Riverina plains of New South Wales and some Cotswold (Gloucestershire) beechwoods are examined in this way and a close correspondence found between grouping based on field and laboratory data separately. The results cannot be generalized, though the method is suitable for any area. In particular, the results are unlikely to hold if comparisons are made between single properties.

This publication has 3 references indexed in Scilit: