4 resistors and 3 switches actually
Excuse terrible picture - and the mess of blue silicon that you have to cut and peel away.
View attachment 66622
In case anyone is interested, there are 3 reed switches giving 4 possible positions of the sensor (none, 1,2 and 3). They are activated by the magnet in the float passing by and the awkward part is they latch as the magnet passes by. This has been achieved by using a miniscule magnet stuck at one end of the reed so that it "sticks" as the magnet passes in one direction and unlocks as it goes the other.
The problems seems to be twofold. Firstly the plastic (on the old model sensor) tube splits, lets in water and the water shorts out the circuit causing the resistance to be unexpected.
Secondly the reeds themselves eventually get magnetised so that the latch/unlatch operation doesn't work cleanly.
Thirdly the reeds are fragile - one of mine was broken but impossible to say if that was shattered before or after I took it apart. The stainless tube I would wager is a better bet for at least two reasons.
The resistors are in parallel so as the float slides it brings more than 1 resistor into the circuit.
I am pondering getting a bunch of latching reeds (i.e. ones from the factory designed to latch as the magnet goes past as opposed to creating my own with a cheap reed and gluing a magnet to the end in the hopefully correct point) to see if I can fix it, but I them have the problem of putting it in a waterproof tube so it is becoming more of a bodge every day