The composition of the continental crust is one of the central parameters in our understanding of the Earth system and its evolution. While the uppermost crust is available for direct observation, our analysis of even this directly observable crust is not free of biases. The deep crust largely remains deeply enigmatic, as the deepest boreholes reach less than a third of the way through the continental crust, and other samples (xenoliths and exhumed terranes) are modified by the same processes that bring them to the surface. Other information, including gravity and seismic data, do carry direct information about the composition of the deep crust, but this information is often nonunique. Here we consider new Bayesian approaches to estimating the composition of the deep crust, as well as sampling approaches for best understanding the composition of the upper crust, and their implications for our understanding of the operation and evolution of the Earth system.
