> Dunno why Walter had odd readings.
Duh! Walter has a '47 chevy.
GM had the "other" system. And since all my GMs had worse problems than gauge issues, I never delved it.
On a Ford, the needle is held up by heat, goes to "E" when power is off.
The GMs mysteriously hold their last reading with the power off.
I think GM uses two coils at 90 degrees, a fixed resistor, and the vari-resistor on the tank float. This gives first-order compensation for battery voltage: a 20% change of voltage does not cause a full tank to read 8/10 or 12/10ths. Apparently at 200% of design voltage, it still gets the E and F end-points about right, but is a bit funky in between.
It is traditional for fuel gauges to linger around F and then go to E fast. If you are near full, you don't care. If you are getting low, how-low is a major concern. Of course this could have just been rationalization for an intrinsically non-linear gizmo.