An alternative, and much more intuitive, approach is proposed to the rigidity of (relatively) strain-free oxide glass networks, based on effectively rigid basic structural units and the true degrees of freedom that allow the formation of such netwokrs; viz. bond torsion angles plus the bond angle at the bridging oxygen atoms. These ideas are extended to borate networks that include rigid super-structural units with no internal degrees of freedom in the form of variable bond and torsion angles, and it is shown that, for an isostatic network, the average super-structural unit connectivity is equal to 4. The role of network rigidity in determining glass formation is discussed, together with the effects of steric hindrance, and a comparison with conventional constraints theory is presented for vitreous SiO2 and B2O3. Finally, it is argued that the so-called intermediate phase is merely an extended rigidity transition range, due to the chemical nano-heterogeneity that characterises the structure of glasses having more than one component, and that, in the case of Ge-Se glasses, the bonding in the interfacial regions between the Se and GeSe2 regions exhibits significant metalloid character.
