valid Temp Sending unit test?

[Scott52]

I have tested all the gauges in the 1952 Crestline using the variable resistance method in the Shop manual. In essence using a resistor instead of the actual sending unit. Adjust the 50 ohm resistor until a reading of 1.5 volts is measured between the Volt input and the sending unit terminal should give a gauge reading of near the center of pointer travel. It does on Gas, Oil and Temp gauges. I connected the actual temp sender to the gauge and suspending it in a can of water that I heated to boiling. The gauge with no voltage sits at the high end of the range. as soon as the voltage is turned on to 12V with a Runtz regulator and the water room temperature the gauge drops to a couple pointer of widths below highest reading. As the water is heated up, the gauge steadily drops to max cold reading. Obviously the opposite of what I would expect a good setup to do. see in flaws to my test setup? I was thinking of buying an aftermarket sender unit. My thought was to use the dash gauge for one side of the engine and a mechanical gauge for the other. What readings should I expect with a good sending unit? Where does the gauge sit at startup and does it increase proportionately at the engine temp increases and where on the gauge is normal operation temperature. I get the impression that there is little difference between normal operation temp and boiling on the gauge. Thanks