[robespierre]

That's really too bad the page is unavailable (and blocked on the Wayback Machine). It had useful information and pictures showing the disassembly process and a few ideas for repairs.

If the problem is cured by pinching the frame, then it seems obvious that there is a connection issue with some of the cables; I believe there is an elaborate flat printed circuit between the two panels, the phototransistor, and the batteries at the sides. Connections between copper and glass are always touchy and there could be possible cures using copper or graphite paint, similar to the products for repairing rubber membrane switches and "zebra strip" LCD connectors in devices like multimeters. But the loss of the web page is really unfortunate because I don't think it will be easy or obvious how to repair these.

addendum:
Here's a thought. Maybe you could contact the webmaster of bio.caltech.edu and ask if they can send you the page and its linked images? It's worth a shot.