This turned edge resulted from the fact that the pre-polishing tool was cast from the mirror after curve generating and before much grinding had been done. Since the grinding had been done with the mirror mostly on top, the curvature increased, so by the time we got to pre-polishing, the pre-polish tool was too flat and wore the edge too much. This was noticable in the pre-polishing phase (the surface cleared in the center but not at the edge) but we hoped it would clear up in the polishing phase. No such luck. Although "anti-turned edge" polishing methods were used, this is a difficult defect to overcome, and in fact the turned edge seemed to only get worse. Later measurements of the error showed that it had increased to about 5 microns off from the desired shape in that region. This is a dramatic error. Fortunately we found some extremely fine 3 micron grinding powder (much finer than the original 9 micron powder). Given the severe error and the scratches, it was decided to take this mirror back to the pre-polish phase with a newly made prepolishing tool. After all the initial polishing the edge had finally polished smooth, but it was turned down, so when we went back to pre-polishing with a new correct-cuvature tool, it roughened everywhere except at the edge. After about 45 minutes, the mirror looked like this...
The left mirror didn't seem to have much of a problem with a turned edge, so we went on to figuring...