I recall something I read back in the '80s or '90s that drag, or was it power needed, increased by the cube of the speed. I looked around for drag vrs velocity today and it at Wikipedia (note that air is considered a fluid in physics) "Note that the power needed to push an object through a fluid increases as the cube of the velocity. A car cruising on a highway at 50 mph (80 km/h) may require only 10 horsepower (7.5 kW) to overcome air drag, but that same car at 100 mph (160 km/h) requires 80 hp (60 kW). With a doubling of speed the drag (force) quadruples per the formula. Exerting four times the force over a fixed distance produces four times as much work. At twice the speed the work (resulting in displacement over a fixed distance) is done twice as fast. Since power is the rate of doing work, four times the work done in half the time requires eight times the power. It's important to value the rolling resistance in relation to the drag force." (See http://en.wikipedia.org/wiki/Drag_%28physics%29
if you're interested.)
Essentially, slow down, but not too slow. Obviously, if you're idling, you're getting zero miles to the gallon so there is a better speed to get the best MPG. This is partially because overall the engine doesn't work as efficiently if you go too slow and with the aerodynamics of the cars back then (I forget when I read it) that the best speed is around 45mph. I haven't tested any speed/mpg as most of my driving isn't where I can maintain a speed.
The '89 Cherokee (not grand) does at best 16mpg. The J2K grand cherokee did around 20+ at 60-80mph with AC on and 87 octane 10%ethanol, Chicago to Munising MI. (I love driving with the windows open, but listening to stories on the radio/CD/MP3 is better with them closed.)