There are so many things that will give different readings for anything when you use 2 different methods. In this case the probe measures how well electricity is conducted by the fluid between 2electrodes and the refractometer is measuring how different the speed of light is in 2 different mediums.
The calibration fluids for these are not sea water. In both cases you are using something that mimics sea water in 1 specific way for 1 specific test. For the probe the calibration fluid conducts electricity the same as sea water and for the refractometer the fluid bends light the same as sea water. Did you use separate calibration fluids designed for each item? Did you calibrate both around the value you are trying to read? Some people calibrate refractometers to 0 with RO/DI water then try to read 35ppm, calibrate around 35 ppm.
Both of these things read a property and convert that to a number that represents the total amount of stuff in water. Each item dissolved in water affects how things differently.
A solution of 35 ppm sodium and chlorine (salt and water) conducts differently than sea water that contains 35 ppm sodium, chlorine, calcium, carbides, strontium, etc. Salt from Red Sea will conduct differently from salt from Instant Ocean. The same is true for the bending of light.
If it is calibrated well and everything else is fine the best you can get is a solution that conducts electricity or bends light the same as the calibration solution.
A single tiny air bubble stuck against one of the electrodes will completely change the reading. The voltage is so low and the changes are so small that the cable will act as an antenna and pick up stuff from other wires in next to it.
A bump can ruin a refractometer
The refractometer must be one designed to measure sea water. You can get refractometers that measure sodium chloride, sugar, or less commonly a whole bunch of other stuff. Is yours for sea water or salt? It is a precision instrument, have you, and the manufacturer, been real careful and avoided bumps? Do you calibrate it every time you use it?
How good are the calibration solutions? Things can go wrong with manufacturing. If they are sloppy then the stuff they use to make the stuff may need calibration.
Enough of this stuff. To answer your question. The easiest thing to do is get a tiny bubble trapped either during calibration or in the tank. A tiny micro bubble will cause a huge difference in the reading. Another problem is that magnetic fields, from electrical stuff, can be picked up by the cable acting as an antenna. When you calibrate and when you put it into your tank keep the cable away from other wires as much as possible.
In the end you are probably not going to get the same readings from both. The best you can do is pick one you want to trust and stick to that one. Calibrate both, carefully. Run the probe wires away from other stuff as much as possible. Read the salinity with both and use one you are going to believe to determine the level. The reading on the other should not change much over time. Occasionally check with the other.
A steady value is better for the living stuff than a fluctuating one. Even if the steady one is not the "ideal"
