The concept that equates oxidation and pressure has been successfully utilized in explaining the structural changes observed in the M 2 S subnets of M 2 SO x (x = 3, 4) compounds (M = Na, K) when compared with the structures (room-and high-pressure phases) of their parent M 2 S 'alloy' [Martínez-Cruz et al. (1994), J. Solid State Chem. 110, 397-398;Vegas (2000), Crystallogr. Rev. 7, 189-286;, Solid State Sci. 4, 1077-1081]. These structural changes suggest that if M 2 SO 2 would exist, its cation array might well have an antiCaF 2 structure. On the other hand, in an analysis of the existing thermodynamic data for M 2 S, M 2 SO 3 and M 2 SO 4 we have identified, and report, a series of unique linear relationships between the known Á f H o and Á f G o values of the alkali metal (M) sulfide (x = 0) and their oxyanion salts M 2 SO x (x = 3 and 4), and the similarly between M 2 S 2 disulfide (x = 0) and disulfur oxyanion salts M 2 S 2 O x (x = 3, 4, 5, 6 and 7) and the number of O atoms in their anions x. These linear relationships appear to be unique to sulfur compounds and their inherent simplicity permits us to interpolate thermochemical data (Á f H o ) for as yet unprepared compounds, M 2 SO (x = 1) and M 2 SO 2 (x = 2). The excellent linearity indicates the reliability of the interpolated data. Making use of the volumebased thermodynamics, VBT [Jenkins et al. (1999), Inorg. Chem. 38, 3609-3620], the values of the absolute entropies were estimated and from them, the standard Á f S o values, and then the Á f G o values of the salts. A tentative proposal is made for the synthesis of Na 2 SO 2 which involves bubbling SO 2 through a solution of sodium in liquid ammonia. For this attractive thermodynamic route, we estimate ÁG o to be approximately À500 kJ mol
À1. However, examination of the stability of Na 2 SO 2 raises doubts and Na 2 SeO 2 emerges as a more attractive target material. Its synthesis is likely to be easier and it is stable to disproportionation into Na 2 S and Na 2 SeO 4 . Like Na 2 SO 2 , this compound is predicted to have an anti-CaF 2 Na 2 Se subnet.