You don't need a Pr Sensor to find the position.
Let's say you place the bouy support (tube) on the container cap. Then the bouy can slide vertically through that tube depending on the liquid voyancy.Then you place a rod with a ferrite section in the bouy shaft and windings in the tube coincident with the center of the windings at normal voyancy. Study LVDT (Linear Variable Differential Transformer). You can convert any displacement of the bouy shaft to electrical signal at the output of the LVDT. You can make your own LVDT by winding a wire on the tube (2 windings). There are specifics on how to do that of course. You need to study that. The advantage is the bouy will be a passive element with no wires and the electrical system will be on the support attached to the container cap and all that can be external if you make the bouy shaft long enough.
To make the sensor make to identical windings on the tube. One for excitation where you apply a high frecuency low voltage signal (let's say 10 KHz). The other winding is the pick up coil where you receive the signal. When the rod (ferrite position changes) the amplitude of the picked up signal changes. Then you do AM demodulation to recover the envelope with a simple diode and a low pass filter (a capacitor). The envelope is your voltage changing with the rod position and therefore voyancy. If you want you can go FM by using the inductance as a component to change the frecuency of your oscillator (study Clapp Oscillator). The ferrite movement will affect the inductance of the coil and then you can demodulate the signal by using a phase detector (XOR gate) or a PLL unit like the CD4076. You need to study that too. It will give you a voltage output dpending on the frecuency changes. You can amplify (using LM324 if required) and feed that to Arduino. The FM version will be more complicated but you will get better inmunity to noise.
The main problem I see is you will have to be Zeroing the stuff regularly as the position of the bouy will change not only with voyancy; but with the liquid level inside as it evaporates, etc... maybe others have an idea of how to solve that problem. If they don't come up with a nice solution you may end up with an evaporation meter instead.