I assume the 2.5V he got at the sensor connector was measured with the sensor disconnected. If that's the case we already know the problem can't be with the sensor itself and it can't be anything to do with the temperature of the fuel...since the sensor wasn't connected.