Abstract
In this paper it is argued that the generality of the Meyer-Neldel rule compels one to adopt a phenomenological approach if a universally valid model is aimed at. It is shown that there exists only one possible phenomenological model. This model is based on an exponential probability distribution of energy barriers. The model predicts a power-law frequency dependence of the AC conductivity with the exponent s given by s=1-T/T0 where T0 is the characteristic temperature of the Meyer-Neldel rule. It is conjectured that the exponential energy barrier distribution derives from a 'glass transition' at T0. The generalisation of the model to account for the compensation effect in other contexts is briefly discussed, using the case of heterogeneous catalysis as an example.