Laboratory measurements of the electrical conductivity of mineral assemblages thought to comprise the transition zone in the mid mantle have been performed. The results agree very well with conductivity models of the transition zone derived from magnetotelluric data. This is surprising given that laboratory measurements are performed on small samples, whereas magnetotelluric sounding samples over a huge volume, and implies that the mid mantle is reasonably homogeneous with respect to conductivity. The mantle is a semi-conductor and its conductivity generally increases with depth, as does its temperature.

Magnetotelluric sounding curves often indicate a zone of high conductivity in the mid- to lower-continental crust. This high-conductivity layer might, for example, be modelled as a 5-km-thick, 0.1 S m− 1 layer. However, there is no Earth material with a conductivity of 0.1 S m− 1, except brine, and we don't expect a 5-km-thick liquid layer in the crust. Rather, the modelled conductance represents a mixture comprised of a resistive rock matrix, and a more conductive second component. There is an ongoing debate as to the nature of this second component, which may promote electrolytic conduction in a network of cracks filled with concentrated brines, or electronic conduction in a network of graphite or ores. Mixing laws, scaling techniques and models of crack connectivity lead us to a more quantitative understanding of the conduction mechanism. It seems inappropriate to pose the question ‘fluids or graphite?’