The sender is in the middle of the tank (which makes sense) . . . I took the sender out for some bench testing.
The tank is pretty much empty . . . might be a gallon or so towards the back. The tank looks fine.
The sender does have some issues in a couple of the ranges. . .

.
In the upper (more full) range the resistance is out of sequence and in the lower range (25-30% full) there are a couple of improper resistance values that show up intermittently . . . which is probably the reed switches acting up. It explains why I was seeing the inconsistency & dramatic drop as the tank level was lowered.
The values are not linear, but there does not seem to be a standard for such, other than the endpoints (33 - 240 ohms)
Anyway . . . I'll be getting a new sender.

