I use 29,000,000 for E. I think the consensus is that the clamp setup that I used was pretty poor as far as simulating textbook cantilever conditions. When I set it up, I figured it would be reasonably close, but several thought otherwise. It's much easier to clamp flat things than round, and when I tested a piece of flat stock and got numbers very close to theory, that confirmed to me that the calculation method I was using was valid. The only choice for the wierd results with the round bar at that point was the clamp setup introducing the error. Next time, I'll weld the round to a plate.