Water temp gauge calibration

This article refers especially to the 924S/944 series. The calibration principle applies to all models, though.Higher than expected water temps on my gauge led me to this. Initially thought about installing a lower temp thermostat. I prefer to verify than to guess even if it takes a little bit longer.....

Tech background:
The water temp sender for the gauge is switched in line with the gauge. Ignition plus is applied to the gauge. The gauge itself is connected to the sender. The sender is grounded. The sender is a "Heißleiter", less resistance upon heat. The gauge acts as an Amperemeter. The hotter the water, the lower the resistance of the sender, the higher the current, the farther the needle goes on the gauge.
As long as the needle expresses the temperatur range we are interested in, a sender and a gauge that are off specification could work properly, nevertheless.
Ignition plus is battery tension if the car is not running (perhaps 12V) but 13,8V when my car is idling. We usually look at our gauge when it's running, so we need the 13,8V. The battery tension only (12V) will yield in lower currents, thus, lower readings.That's why I let the car idle during the test.

Test realisation:
Bought a new sender and an electronical thermometer. Precision is +-0,1 degree Celsius +-2digits from -100 degree Celsius till 150 degree Celsius,  it shows temperatur in  0,1 degrees steps, so the total is +- 0,3 degrees.
I hung the new sender in a pot with water. Put the pot on a cooking plate. I grounded the sender with battery mass.
I disconnected the sender cable at the sender in the car and made a cable connection to the new sender in the pot.
Principally both cables create an additional resistance that reduce the gauge current and reading. But the cable resistance is low as a percentage of minimal possible sender resistance (at 120 degree Celsius which is supposed to be the red corner) So that does not matter.

Now I slowly start boiling the water. As the temperatur of my pot increases the needle rises. At my predefined marks I check the water temperature with the electronic thermometer.
We have to consider that the water will probably heat faster than the sender inside, so the results would show somehow higher than real temperatures. That's why we want to heat it up slowly.
Water can only reach 100 degrees Celsius max. That's why I reran the test with the pot filled with oil. I heated the oil up to around 145 degrees Celsius and let it cool down.

 Results