We present a microscopical approach to the electromagnetic vacuum energy of a random medium made of non-polar molecules. We evaluate the contribution of statistical fluctuations to the average total vacuum energy. While the Lamb shift is a function of the electrical susceptibility only, the vacuum energy is generally not, except in the quasicrystalline approximation. We comment on the possibility of testing experimentally our results. We clarify why the effective medium vacuum energy (i.e., that of long-wavelengths) does not account for the total vacuum energy of a molecular dielectric. Consequently, the Lamb shift does not derive from the effective medium vacuum energy except at leading order in the molecular density.
展开▼