Recently, experiments on expanded fluid mercury, and simulations of alkali metals in liquid ammonia and in molten alkali halides, have motivated interest in the possibility of a Frenkel excitonic insulator (EI) phase. In this paper, the authors discuss several aspects of a model for describing such a phase. They choose a model Hamiltonian which is the asymptotically exact low-density form for a system of atoms each possessing an sp3 basis, and present two methods of analysis. The pairing theory centres on the broadening of the S to P atomic transition into exciton bands, which, as is known, may lead to the formation of a Frenkel EI phase. They analyse two corrections to the traditional exciton picture: (i) double-excitation processes, which are responsible for the van der Waals stabilization energy, are shown to halve the density predicted for the transition to an EI phase; (ii) correct incorporation of non-boson statistics for the exciton operators is argued to drive the transition from first order to second order. The authors also analyse the model Hamiltonian via a Hartree approximation, which proves to be the more tractable method, and further allows an explicit description of the EI phase itself. The validity of the Hartree approximation is justified by comparison with the pairing theory.