Thanks to Joel Bard for pointing out the problem.
Paul.
I recently found that for 2 weeks or so the chi angles code was broken.
Sorry about that (I'd been trying to get conformation sampling to work
better). As of rev 1920, things are back to how they used to be.
