Experiments carried out by one of the present authors with a slab of glass under flexure have indicated that artificial double refraction in glass strained beyond the elastic limit is probably proportional to the stress rather than to the strain ( cf . L. N. G. Filon, ‘Phil. Trans.,’ A, vol. 207, pp. 303-305). More recently the experiments of Prof. E. G. Coker and Mr. K. C. Chakko (“The Stress-Strain Properties of Nitro Cellulose and the Law of its Optical Behaviour,” ‘Phil. Trans.,’ A, vol. 221, pp. 139-162) have suggested that in celluloid or xylonite, whilst for the highest loads the double refraction is no longer proportional to either the stress or the strain, it is more nearly proportional to the former than to the latter. Further, casual observations in the past had shown one of us that, when a piece of transparent material, whether glass or celluloid, which had been overstrained was released, a certain amount of residual illumination was visible between crossed nicols in certain cases. This illumination gradually died out, showing that the artificial double refraction exhibited some of the characteristics of permanent set, with a slow recovery. No precise measurements of this effect, however, seem ever to have been made.