the simulated variability at deep/shallow levels. Density changes responsible for the pseudo-reconstructed FCT are mainly driven by zonal temperature differences; salinity differences oppose but play a minor role. These results thus support the use of the thermal-wind relationship to reconstruct the oceanic circulation past variability, in particular at multidecadal timescales. Yet model-data comparison highlights important differences between the simulated and the proxy-based FCT variability. ECHO-G simulates a prominent weakening in the North Atlantic circulation that contrasts with the reconstructed enhancement. Our model results thus do not support the reconstructed FC minimum during the Little Ice Age. This points to a failure in the reconstruction, misrepresented processes in the model, or an important role of internal ocean dynamics.