For the last 2 days, (since about 0230 6/15), Wetness.pond has been low. Upon inspection, there was nothing obviously wrong (though some of the plastics on the probe are cracked). Reseating connectors and cycling power didn't help. With water on the probe, the signal went from ~0.18V to ~0.0, which of course is the wrong sign.
I've brought the probe back to the base for further evaluation. We do not appear to have a spare here.
P.S. Steve S. may have noted this elsewhere, but he told me that he opened up this sensor and found no bad connections. He assumed that the internal circuitry in the the rubber mould died. I note that the rubber had been deformed by the cable ties used to mount it. Thus, we might want to be more careful with mounting these, in the future.