• absolute reference frames;
  • hot spots;
  • mantle plumes;
  • plate motion;
  • true polar wander

[1] We defined a new global moving hot spot reference frame (GMHRF), using a comprehensive set of radiometric dates from arguably the best-studied hot spot tracks, refined plate circuit reconstructions, a new plate polygon model, and an iterative approach for estimating hot spot motions from numerical models of whole mantle convection and advection of plume conduits in the mantle flow that ensures their consistency with surface plate motions. Our results show that with the appropriate choice of a chain of relative motion linking the Pacific plate to the plates of the Indo-Atlantic hemisphere, the observed geometries and ages of the Pacific and Indo-Atlantic hot spot tracks were accurately reproduced by a combination of absolute plate motion and hot spot drift back to the Late Cretaceous (∼80 Ma). Similarly good fits were observed for Indo-Atlantic tracks for earlier time (to ∼130 Ma). In contrast, attempts to define a fixed hot spot frame resulted in unacceptable misfits for the Late Cretaceous to Paleogene (80–50 Ma), highlighting the significance of relative motion between the Pacific and Indo-Atlantic hot spots during this period. A comparison of absolute reconstructions using the GMHRF and the most recent global paleomagnetic frame reveals substantial amounts of true polar wander at rates varying between ∼0.1°/Ma and 1°/Ma. Two intriguing, nearly equal and antipodal rotations of the Earth relative to its spin axis are suggested for the 90–60 Ma and 60–40 Ma intervals (∼9° at a 0.3–0.5°/Ma rate); these predictions have yet to be tested by geodynamic models.