Abstract The fact that the rupture strength of solids is, in general, only an ex­ceedingly small fraction—of the order of one-thousandth—of the theoretical strength has been explained on the hypothesis of submicroscopic flaws existing in the material. In a fundamental paper A. A. Griffith (1920) accounted for the behaviour of glass under rupture tests on the supposition that fine cracks were inevitably present, and showed experimentally that freshly drawn glass threads were much stronger than old ones. His hypo­thetical cracks were in the body of the material, but considerations similar to those given by him can be equally well applied to surface cracks, and such cracks have been assumed by other workers (Joffé 1928) to account for the anomalous mechanical strength of both amorphous and crystalline solids. Experimental evidence has been brought forward for the existence of surface cracks on glass, vitreous quartz and diamond (Andrade and Martindale 1935), but the matter is clearly one where further confirmation is welcome. In the course of experiments on single crystals of sodium it was noticed that the walls of the vessel, of pyrex glass, in which sodium was being dis­tilled presented a curious iridescent appearance in sunlight. This was caused by a large number of very fine lines, with which the surface was covered. Fig. 3, Plate 5, shows a small area of the inside of the shoulder of the glass distilling bulb, at a magnification of × 14. There seemed to be no way in which the surface could have been scratched, so the question arose as to whether these could be submicroscopic surface cracks, developed by the attack of hot sodium vapour. A series of experiments was therefore under­taken with different types of glass and different preliminary treatment of the glass. Hydrofluoric acid was used as well as sodium vapour in an attempt to develop the cracks, but proved ineffective.