The compatibility of ice-core CO2data with direct estimates of biotic CO2release is tested using a response function analysis that, apart from stationarity, leaves the oceanic uptake of CO2largely unconstrained. The role of CO2-enhanced growth (which had not been considered in an earlier study of this type) is examined explicitly. It is found that allowing for CO2-enhanced growth increases the discrepancy between the source estimates and the ice-core data. The proposed explanation is that some of the early increase in CO2 concentrations reflects a recovery from a perturbation associated with the little ice-age.