A wide range of residual stresses was reported for 850, 900 and 950 F. Variation in the initial stresses was a primary cause for these ranges. This is shown through the replotting of some of the residual stress-time curves, page 28. All the data, when cross plotted as initial stress versus residual stress for the four time periods, page 28, show the influence of the several variables more clearly. 1. The relative strengths over the temperature range and time periods considered increased with initial stress. The influence of initial stress decreased with increased temperature of testing, and to some extent with increased stress and test time. Some tests at very low initial stresses consequently gave very low relaxation strengths, far below the potential relaxation resistance of the steels. Many of the tests at intermediate initial stresses were also below the potential strength, further contributing to an apparent scatter in relaxation strength. 2. In most cases, normalizing and tempering resulted in higher residual stresses than quenching and tempering for a given initial stress. The one annealed material had low relaxation strength for the initial stress in comparison to the hardened conditions. 3. There apparently was no consistent difference between steels with chromium contents of 1.0 and 1.25 per cent. 4. The influence of austenitizing temperature and tempering conditions is not clear from the data. The sparse data suggest that effects of these variables are secondary to quenching versus normalizing. 5. The one set of data for tests with elastic follow-up (elasticity factor greater than 1) shows a substantial increase in residual stress over what would have been expected for an elasticity factor of 1.0. Restressing and rerunning a test also increased residual stress. Elastic follow-up and some increase from restressing contributed to the high residual stresses for the second test on Material 2 at 930 F. 6. There are some variations in comparative properties for which no ready explanation is apparent in the information available. The curves showing variation in relaxation strengths with temperature were based on the correlation of residual stress with initial stress. The range of initial stresses used were from approximately 45,000 psi at 800 F to about 33,000 psi at 1000 F. Many of the test points deviate from these initial stresses and accordingly lie above or below the curves.