Here's a link to the page about linearizing the stock gauge:
I did something like this a few weeks ago on my '92. Using Wallyman's research as a baseline, I cobbled together a test setup with three big variable resistors (two for the adjustment, one to simulate the sensor) and calibrated my stock gauge so that it was almost totally linear in the range from 150-250F, with 195 dead center at the 12 o'clock position.
I used the same basic design as the one shown above, but I settled on 18 ohms series, and something around 79-81 ohms parallel. This is just what it worked out to for the exact gauge dynamics that I want. To get these values out of what was on the shelf at Fry's I used a 20 and a 200 together to make 18, and some other two values to get the other.
A few days later I replaced the stock thermostat with a Stant 180 degree unit, and now under normal cruise the needle sits straight upright, and in slow, hot traffic is wanders over into the 2:00 position until I start moving again. Just like a temp gauge should.
I'm in Phoenix at the moment, but what I get home Thursday I can dig up the pictures I took.