Field test of gibbsite as a proxy for paleo-PCO2 indicates that the current one-dimensional Fickian diffusion model in use does not adequately account for physical and biological processes in the upper portion of the soil (above the B horizon). Cycling of nutrients and non-uniform input of organic material at all depths result in stable isotope compositions of CO2 trapped in pedogenic gibbsite that is not well represented by a simple diffusive mixing model. As a proxy for paleo-PCO2, differences in results between this and previous studies indicate that differences in the conditions of soil formation such as parent material, may also affect the preservation of carbon in ways that are not accounted for by the model.