Abstract In this third paper of this series we consider two alternative static-lattice models to the defective Sillén structure, namely a <111> superlattice of oxygen vacancies, including <110> defects, oxygen vacancies and pseudo-interstitials (that is oxygen ions occupying vacant anion sites in the transient <111> superlattice). These models are chosen to reflect long-range disorder in δ-Bi 2O 3. The first is the Gattow model in which every oxygen lattice site is occupied for 75% of the time. In a static-lattice simulation this is achieved by complete occupancy of the oxygen sites by “ 3 4 of an oxygen ion”, a concept that presents both conceptual and computational difficulties, especially when the lattice contains defects. The second alternative is a Sillén region embedded in a Gattow lattice. The Gattow model has the advantage that, in contrast to the Sillén model, no restrictions on polarizability prove to be necessary. All three models agree inpredicting that pseudo-Frenkel defects are more stable than conventional Frenkel defects and that δ-Bi 2O 3 contains a high-concentration of charge carriers. However, the Gattow and mixed Sillén-Gattow models are difficult to apply to charge migration, specifically because of the absence of a vacancy superlattice. Some problems associated with the simulation of highly defective materials like δ-Bi 2O 3 are discussed in an appendix.