Detection of strength-impairing surface flaws in glass

Experiments have shown that the cracks produced by sodium vapour treatment on the surface of glass (Andrade & Tsien 1937) are fractures due to a superficial tensile stress developed on cooling in a surface layer of higher thermal expansion. Tensile surface stresses and corresponding systems of cracks can also be developed, in soda-lime glass, by treatment with a molten lithium salt. The points of origin of the crack systems can be clearly recognized; they must be the sites of Griffith cracks. The structure (linear or star-like) of each Griffith crack can be recognized from the orientation of the network cracks at the point in question. The experimental evidence for the present interpretation of the Andrade—Tsien cracks is as follows: (1) A large amount of sodium is absorbed in the surface layer of the glass. (2) The characteristic crack patterns are not visible at the end of the heating period in sodium vapour, but appear progressively as the sodium-treated glass cools. (3) The cracks disappear when the glass is reheated, and a new pattern appears on subsequent cooling.