Bram Vaes , Douwe J.J. van Hinsbergen 

True polar wander (TPW), the rotation of the solid Earth relative to the spin axis, is driven by changes in the Earth's moment of inertia induced by mantle convection and may have influenced past climate and life. Long-term TPW is typically inferred from large polar shifts in paleomagnetic apparent polar wander paths or computed directly by rotating them in a mantle reference frame. However, most apparent polar wander paths do not incorporate uncertainties in paleomagnetic data, which may bias estimates of TPW. Here, we provide new quantitative estimates of TPW since 320 Ma by placing a recent global apparent polar wander path corrected for age bias and with improved uncertainty quantification in existing mantle reference frames. We find large amplitude (>10°) but slow TPW rotations that predominantly occurred about two equatorial axes that are approximately orthogonal. During the Triassic and Jurassic, a ~24° TPW oscillation occurred about an axis at ~15°W, close to the present-day TPW axis at ~10°E. In contrast, the TPW axis was located at ~85°E during a smaller oscillation (~6–10°) over the past ~80 Ma, as well as between 260 and 320 Ma. We propose that these varying TPW axes reflect changes in the distribution and flux of subduction in the Tethyan and Pacific realms. We find no evidence for previously postulated fast (>1°/Ma) TPW oscillations in the Cretaceous and Jurassic. Finally, we propose that calibrating mantle convection models against reconstructed TPW will improve our understanding of mantle dynamics and the drivers of TPW itself.