atmospheric carbon dioxide ͉ electronic structure calculations ͉ paleosols ͉ goethite ͉ gibbsite T he paleosol carbon-isotope method is one of the major techniques for reconstructing the composition of the ancient atmosphere (1, 2). The depth-dependent mixing of respirative CO 2 from soil ( 13 C-depleted) with atmospheric CO 2 ( 13 Cenriched) can be preserved in soil minerals and has been used to infer atmospheric PCO 2 and respiration rates in paleosols (3). The idea was directed originally at soil carbonate minerals (4) and later applied to the CO 2 component of pedogenic goethite (␣-FeOOH) (5-7) and gibbsite (Al(OH) 3 ) (8, 9). The fractionation between mineral-dissolved CO 2 [CO 2 (m)] and CO 2 (g) is a key factor in the technique because it determines how the respired CO 2 vs. atmospheric CO 2 mixing profile is imprinted onto the soil mineralogy. Direct measurements of the CO 2 (m)-CO 2 (g) fractionation factors for (oxy)hydroxide soil minerals have thus far been elusive. The small values (0 to ϩ2.5 per mil) that have been provisionally implied are surprising because, by analogy with carbonate minerals, a 13 C-enriched fractionation factor (close to ϩ10 per mil) might be expected (10,11).In carbonate minerals with the calcite and aragonite structure, the overwhelming majority of carbonate ions are in a single crystallographic site that governs the equilibrium isotope fractionation. In contrast, occlusion of CO 2 in oxide soil minerals at trace concentrations could involve multiple sites with varying populations, possibly depending on the kinetics and conditions of soil formation. For goethite, it has commonly been assumed that the open pseudochannels parallel to the c crystallographic axis are the major host for CO 2 (m) (12), but this has not been definitively demonstrated, and it is possible that other types of sites could be involved in the uptake of CO 2 (m). An important question is then whether the various kinds of CO 2 (m) sites would have large differences in their equilibrium 12/13 C isotope fractionation factors. Little attention has been given to the possibility of heterogeneous chemical speciation of CO 2 (m) within the host mineral. Site heterogeneity would complicate the interpretation of the carbon isotope records preserved in these minerals. At the same time, a better understanding of the crystal chemistry of CO 2 (m) defects in oxide/oxhydroxide minerals and their isotopic fractionations could serve as a foundation for a much more sensitive paleoclimate indicator than would be possible with knowledge of only the overall isotopic composition. Knowledge of site-specific isotopic signatures in minerals could have much wider applicability than the CO 2 -Fe,Al(oxy)hydroxide system.Although the experimental difficulties of sampling these individual sites might be daunting, accurate theoretical calculation of the potential fractionations is now feasible by using methods of quantum chemistry. During the last several years, there has been notable success in first-principles calculation of equilibriu...