These differences of opinion may be explained on the basis of the dual role of stress concentrators [1], namely, that they increase the magnitude of the maximum stresses, contributing to the rise of corrosion cracking, but at the same time they reduce the zone of application of dangerous stresses, thereby decreasing the probability of the occurrence of weak points in this zone, which possess an increased tendency to corrosion cracking. Depending on circumstances, each of these factors may play a predominant role. In the present work we investigate the influence of the simplest stress concentrators encountered in fabrication on the tendency for corrosion cracking of the alloy AMg6, under conditions of pure bending and constant deformation. The specimens were prepared from sheets of AMg6M of 2.5 mm thickness, having the following chemical composition: 6.5% Mg; 0.8% Mn; 0.06% Cu; 0.1% Zn; 0.3% Fe; 0.1% Si; 0.04% Ti; 0.0005% Be. The structure susceptible to corrosion cracking is formed as a result of tempering at 450~ (heating for a period 4 h with cooling in air}, cold roiling with a reduction of 24%, and aging at 175 ~ C for 24 h. The industrial protective coating was removed by pickling in a 10% solution of NaOH for 8 min at 60~ followed by brightening in 25~0 HNO~. The following specimens were tested: plain, flat specimens of dimensions 110 12.5 1.9 mm; specimens measuring 110 15 1.9 mm, with a central hole of 2.5 mm diameter; others having a double-sided, round groove of radius p=2.5 mm; and still others having sharp notches of depth t=0.5, 2.5, and 5 mm (the angle of the notches was 40 ~ and the radius of the apex was p =0.1 mm). The bending stresses were applied to the specimens in a textolite device having four supports. The loading was accomplished on VILS equipment which, with the aid of a timer, made it possible to assign a deformation to the specimens with an accuracy of 0.01 mm. In order to determine the bending deformations corresponding to the assigned loads, a calibration curve of load vs deformation was developed for each type of specimen. The calibration was made on an FM-250 machine, through the use of equipment which simulated the scheme of loading in the corrosion-cracki ng experiments. Under conditions of elastic deformation, the calibration curves for the plain specimens agreed with results calculated from the formula reported in [5]. Tests were carried out at 25 ~=I~ under conditions of natural aeration and complete immersion in a solution containing sodium chloride (0.25 N), acetic acid (0.05 N), and sodium acetate (0.2 N), specially prepared for accelerated corrosion cracking tests for these alloys [6]. The life of the specimen was defined as the interval of time elapsing before the formation of cracks cutting through the whole width of the specimen. At each level of stress, 10-20 notched specimens of each type were tested, and 25-50 plain specimens were tested, thereby permitting a statistical analysis of the results [7].