The spatial correlation of log permeability can be expressed as a combination of the bivariate statistics of both indicator and continuous variables.  Analyses based on this expression can illuminate relationships between the spatial correlation of permeability and hierarchical stratal architecture.  Relevant length scales of correlation can be identified and related to the “statistical geometry” (mean and variance in lengths) and the juxtapositioning of stratal unit types within the hierarchy. This approach can provide a rational basis for modeling the spatial correlation of permeability.  It can aid in identifying sources of bias in computing univariate and spatial bivariate statistics.  It can also provide a framework for using more abundant lithologic data to supplement limited permeability data in defining spatial correlation models.