Abstract The structures and chemistry of the oxide and oxysalt minerals are impressive in both their complexity and diversity, and it can be reasonably asserted that our understanding of these aspects lags far behind our experimental capabilities for their characterization. Nevertheless, there are some empirical rules that (sometimes weakly) govern the constitution of these minerals, rules that date back to early work on the modern electronic theory of valence (Lewis, 1923) and the structure of crystals. The most rigorous rule is that of electroneutrality: the sum of the formal charges of all the ions in a crystal is zero. Other rules grew out of observations on a few mineral and inorganic structures, and various ideas emerged during the 1920s: that atoms have a specific size, tables of atomic and ionic radii, the idea of coordination number, considering structures as polymerizations of coordination polyhedra. These ideas were refined by Pauling (1929, 1960) who synthesized them into his well-known rules for the behaviour of ‘complex ionic crystals.’ Some aspects of these ideas have been extensively developed up to the present time. There are now available tables of accurate empirical ‘ionic’ radii (Shannon, 1976) whereby mean interatomic distances for specific coordinations can be predicted typically to within ∼0.01 Å. Individual bond lengths can be predicted via various developments of Pauling’s second rule ( e.g ., Baur, 1970, 1971), and the relative strengths of bonds can be calculated (Brown & Shannon, 1973) given the observed bond lengths in a structure. Thus we can currently predict various geometrical