We calculate the changes in the electrical conductivity of crystalline rocks due to the closure of cracks as pressure is applied. We have modeled the system as a random array of pore conductors, crack conductors, and nonconducting bonds on a network. There are several different ways of doing this; the inputs to these theories are the high-pressure behavior of the conductivity (where the cracks are presumed to be closed and only the pores are conducting) and the amount of crack porosity, ϕc(P), which is derived from the dilatancy data. In all cases where there is a clear demarcation between a region of conductivity dominated by crack closure and that due to residual porosity, we have been able to successfully model this dependence with one free parameter. The conclusion is that the pores and cracks are arrayed on a lattice having an average coordination number in the range 2<Z<2.5 and that at random, they each close completely as pressure is increased; our operational definition of coordination numbers is quite different from that deduced by visual inspection of micrographs. The free parameter is the occupancy of nonconducting bonds, expressed as a fraction of the maximum available number of such bonds. Other plausible models based on other ways of distributing crack and pore porosities are strictly ruled out for all values of the relevant parameters.