The d13C from CO2 (d13CO2) in ice cores covering the last 150 kyr contains, besides other variabilities, a 0.4permil trend between the Penultimate and the Last Glacial Maximum (PGM and LGM) and a wide and deep minimum around 60 kyr BP, both of which are still not understood. Here we use BICYCLE-SE, a model of the global carbon cycle, and new compilations of marine d13C data for a revised interpretation this d13CO2 time series. We find that in addition to rather well known variability in ocean circulation, and the marine and the terrestrial biology, solid Earth processes are potentially necessary to explain the reconstructed d13C data. In detail, variability in the d13C signature of volcanic CO2 outgassing, or of d13C of weathered carbonate rock could explain at least the long-term trend in d13CO2 between PGM and LGM, while for an explanation of the minimum around 60 kyr BP the necessary assumptions on d13C signatures are outside of ranges supported by data. Model results cleary show, that the global mean surface ocean d13C contains most of the dynamics already found in atmospheric d13CO2. This finding is supported by a data/model comparison of marine d13C. In detail, a non-polar global d13C stack from planktonic foraminifera (here restricted to data based on the species G. ruber) agrees with simulation results, but only if isotopic fractionation during calcite formation as function of surface ocean CO2-concentration, the so-called carbonate ion effect (CIE) which has been determined in laboratory experiments, is considered. The carbonate ion effect needs to be considered for Termination I, but seems — for reasons yet not understood — unimportant for the dynamics leading in and out of the Eemian.