Thanks, Alan, for pointing out that the "live-end/dead-end" theory has been unsupported by tests. I stand corrected -- and feel better, too!