The rate of removal of mechanical strain should follow Maxwell's first‐order law, but for glasses it has been found that the rate is proportional to the square of the birefringence. The present paper explains this in terms of the structural picture of glass. During annealing the stress optical constant of the glass, as well as its viscosity, undergoes a continuous change. Birefringence, therefore, is not a correct measure of the stress, and the Adams‐Williamson findings are only apparently in contradiction to the classical derivation of Maxwell.