Volume matching in Tolman models

Since the equations of general relativity are nonlinear, it is not strictly correct to obtain average values by integrating over spatial volumes. Yet this is really what is done in the attempt to fit our rather lumpy universe to a standard cosmological model of uniform density. Consequently, the fitting problem, raised last year by Ellis and Stoeger [1], asks how accurate the average values derived from observational cosmology can be, even without measurement uncertainties. Do they really describe the best-fit Robertson-Walker model to our universe? One of the alternatives to averaging they suggested was that of volume matching. We try to provide a first estimate of the error due to averaging by fitting a Robertson-Walker model to an inhomogeneous Tolman model using realistic density profiles. Comparing the results from volume matching and from averaging, we find that errors are of the order of 10% or more.