>Since actual error distributions on crystals are bimodal, not Gaussian, you would >normally get just about that error. This sounds interesting. Any ideas why? -Charles