A standard way of plotting an S-N (fatigue strength vs no of cycles) graph is to start off by taking a value of Sm (for N=1000) cycles and joining it to the point depicting Se(Endurance Limit). Now we commonly take Sm= 0.9Sut for steels. Now I was wandering, before you actually get to use the component for 1000 cyces at 0.9 Sut the component will yield at around 0.5Sut(before you reach Sm), as Sy(steel)≈.5Sut. SO should yielding be the cause of failure for values of stress rather than fatigue? Thank you in anticipation.