One error I see is that the holes in the case were sealed with silicon, not silicone. What's worse, the silicon was probably mutating into semiconductor devices. These active devices may well have developed a mind of their own and sabotaged the experiment. I strongly suggest using silicone in future similar experiments.
You may jest, but the local Tektronix oscilloscope factory repair facility had a tub of water / detergent out back. First thing they did with a unit which came in for repair was to wash it in that tub. I admit that was before the days of semiconductors.