Thanks for the tip, Farling. I just tried that, and it did not solve the problem. I will look into getting the relevant files publicly available so that others can look at it. The last time I tried zipping them up, it was just above the margin for attachments here, but I still have a fair amount of cruft from the skeleton files and my experiments. I do have it available via GitHub
here if you'd like to take a look at the files themselves.