So you're suggesting... What? That the density of the gasoline changes enough to make the fuel sender float neutrally boyant? Or just that it changes the level in the tank?
I think your solution is way more complex than it needs to be. The variable resistor used for fuel level senders is far from a precision instrument. Combined with the odd body contours that fuel tanks generally follow, you get strange patterns of needle movement relative to actual fuel volume.

yea but its not 3 gallons worth. realistically how much fuel will ever be in the fuel lines?? how much fuel will ever get above 100degrees?

I don't know for sure about 92-95, but for 89-91 Honda's, they have a couple of characteristics which are common. I have a close association with 4 different cars and they all are pretty similar when it comes to the gauges. The first 50 miles is free, the gas gauge doesn't even see the change. The second 50 is generally good for about 1/8 of a tank. The second 100 generally drops like a rock, to about 1/2 a tank. The last 1/2 of the tank holds for about 50 and then it drops quickly to about a 1/4 tank.

All 4 cars are Honda Wagon's and the tanks hold about 12 1/2 gallons of gas, from stone empty. The gauges don't seem to see the last 2 1/2 gallons of gas, but I expect that much is required to submerge the fuel pickup.

The sensor in the tank is not linear because of it's design. I suspect that the readings are considered more of an aproximation, with don't let it run on empty for to long, being the primary consideration.

The shape of the fuel tank is a significant contributor to the non-linear characteristics of the fuel gauge.