The initial error was shocking enough. But the the defective pins were identified at the heat treater and came back in a separate box with a red tag, individually marked with yellow lines, and bearing distinctly different oxide colors. They were somehow mixed back in, and later recorded as passing(!) hardness tests. (I'd also be surprised if the huge difference in hardness wasn't obvious when the parts were stamped with serial numbers, though the report doesn't say exactly how that was done.) Maybe they got confused over whether they were manufacturing helicopter parts or paperweights?