Bjørnerud and Austrheim1 interpret the geological evidence in the rocks of Holsnøy at Lindås nappe, Norway, to be inconsistent with our cold-crust model2, but do not question our new argon isotopic data, on which we base the thermal history of the terrain. A critical flaw underlying their arguments1,3 is the implicit assumption that element diffusion does not occur in dry environments, although there is clear evidence to the contrary2,4,5. Counter to earlier claims3,6 of element and isotope immobility in dry rocks, we have demonstrated the existence of diffusion profiles in phlogopite associated with the uptake of argon during the Caledonian in ‘unreacted’ protolith of the Lindås nappe. Diffusion has taken place in these dry rocks and cannot be ignored.
Addressing their specific objections to our model in turn, we consider first the transport of hot fluids through the crust. Our contention is not that the whole terrain was heated by 300 °C, but rather that regional ambient temperatures were not significantly affected by localized advection of hot fluids. Regarding the source of fluids, it is widely accepted that ample volumes can be provided from dehydration of the subducted rocks and there is nothing to prevent these fluids travelling long distances.
Second, regarding the temporal relationship of fluid access, recrystallization and deformation, Bjørnerud and Austrheim1 suggest that the hydrated eclogite formed before ductile deformation (also at eclogite facies). This implies that the hydrated assemblage was overprinted by a later fluid-absent, high-strain event, confined only to the eclogite, for which there is no field evidence. We believe that the structural and microstructural relationships indicate that deformation and fluid infiltration were coeval. Channelized fluid flow through shear zones7 accompanying fluid-enhanced dynamic recrystallization8 is a widely accepted process.
In further support of our contention that the plagioclase-rich granulites remained cool and were not at 700 °C, as suggested by Bjørnerud and Austrheim1,3, we note that plagioclase deforms plastically at temperatures above about 550 °C (ref. 9): we do not observe this deformation. In fact, the experimental data10 quoted by Bjørnerud and Austrheim1 show that, under dry conditions, plagioclase-poor rocks were stronger than plagioclase-rich rocks, with the rock strength defined by dislocation creep in plagioclase.
Contrary to the assertion of Bjørnerud and Austrheim1, we do not find the occurrence of frictional sliding at great depths to be surprising, as many deep earthquakes have been recorded in subducting slabs and in the mantle. We can reconcile the geological evidence with recent interpretations of seismic data from seismogenic zones11. After an earthquake of magnitude 8 in Chile in 1995, 4,426 aftershocks were recorded over 3 months and interpreted as representing the rapid migration of fluid into the overlying plate. This is analogous to what could have been occurring in the Bergen arcs during the Caledonian, with large earthquakes producing pseudotachylyte and aftershocks resulting from fluid infiltration.
We therefore find that the arguments put forward by Bjørnerud and Austrheim1 do not provide compelling evidence against our model, nor do they provide an alternative scenario that reconciles all the geological evidence. Our model coherently and consistently integrates the geological observations with isotopic data, geophysical constraints and the tectonic setting of the Bergen arcs.