Crystalline BeF 2 and SiO 2 have similar structures in which tetrahedrally coordinated cations (M) are bridged by bi-coordinated anions (Y) producing a network of MY 4 tetrahedra with shared corners. Each compound melts to produce a very viscous liquid which no doubt is highly polymeric. When a basic fluoride is added to BeF 2 or a basic oxide is added to SiO 2, the viscosity of the resulting NY-MY 2 mixtures drops rapidly, presumably because bridging anion linkages have been broken, and the degree of polymerization decreased. This plausible qualitative description has been treated more quantitatively on the assumptions that: (1) The many possible polymeric species M aY b which can be formed in such binary mixtures all involve tetrahedrally coordinated M cations and mono- or bi-coordinated Y anions; (2) the stability of each such polymeric species M aY b depends primarily on the number of bridging anions (-Y-) and the number of non-bridging anions (-Y) it contains and, to a lesser extent, on the size of rings formed in the cross-linked polymer structures; and finally (3) the activity of species M aY b in the melt can be derived from volume fractions according to Flory's approximation for the entropy of mixing and from a heat-of-mixing term. Although the evaluation of component activities ( aMY 2 and aNY) from this model requires extensive numerical calculations, the model is inherently a simple one with but three adjustable parameters. It fits observed activities quite well over the full composition range in the LiF-BeF 2 system and in several silicate systems for which accurate data are available.