As early as 1912, it was pointed out by Rosenhain and Ewen (1) that the behavior of metals at high temperatures could be explained by the combined action of the crystals and the so-called grain boundaries. This theory was also made use of later to explain problems in connection with creep phenomena (2). The author discusses subsequent studies in the United States and Germany, relating to the occurrence of brittle fractures in metals after long periods of time. With the aid of the Rosenhain-Ewen conceptions, he attempts to reconcile discrepancies between recent observations on the occurrence of inter-crystalline fractures after long test periods. His evaluation of the theories cited offers an explanation of various phenomena which previously could not be interpreted, and also furnishes a basis for determining the risk of failure with creep in a three-dimensional system of stress. He concludes that the problem of calculating the strength of a material subjected to creep is actually solved only when the metallurgist and the steel manufacturer are in a position to furnish the designer with complete experimental data enabling him to predict failure for all systems of stress occurring in practice.