More speed better mileage?
 

Had to really get on the throttle 2 days this tank. Thought for sure I was going to take a MPG penalty. Wound up getting slightly better MPG ?!?! I thought I had read on another forum that the 2.3l Rangers liked rpm's. It's almost making me rethink the rear gear change. I think I'll have to try a tank in 4th gear at 55 and only use 5th for freeway speeds. Should be about 2300 rpm for both gear ranges.

We'll see.

My stick shift Saturn L has factory 4.45 final drive ratio. It runs 2700 rpm at 60 miles per hour.

A better test would be to use your FE gauge and cruise in each gear along the same stretch of road repeatedly in one session. Hold steady throttle, set up a camera to hands-free record an instant FE readout. I did this once with my Civic to explore this exact question and found that the higher gears were much better, even when they gave much taller gearing than my stock set up (I swapped my transmission). My 4th gear was equivalent to my old, stock 5th gear set up. But my new taller 5th gear was much better anyway.

Test it. See what you get. Post it here. I'd love to see.

The second trip with gunning it more may have had better temperature and weather conditions, a tailwind, etc., or better traffic. You would need to test this under exact conditions to know for sure.

OBD gauges can't be relied on for this sort of testing.

An MPGuino, maybe.

That depends on how precise a difference you need to measure. Properly calibrated, an OBD gauge certainly can tell the difference between 45 mpg and 50 mpg. My UltraGauge is consistenly within a single MPG, sometimes spot on, when compared with the pump. It is never wildly off by even 2 or 3 percent. For our purposes, OBD will work fine I think.

When I had my Saturn, I first shifted at 2k. Everyone knows shifting early gets better mpgs... I did some research and found my engines peak tq was around 3500rpms. So I started shifting a 4k, which keeps the engine in its "power band". Still got 40 mpg and it didn't take forever to reach cruising speed. I quit the lugging, that's for diesels.

In my Saturn I found gains to be had for lowering the RPM at which I shift... I started out shifting at 2500 or so and ended up shifting at something like 1800 RPM(if I shifted through each individual gear)... now I’m back up to 2500 or so, but I skip shift at every opportunity(3rd synchro doesn’t like the 1-3 shift very well though)...

When you calibrate an OBD gauge, you essentially tell it "this was my fuel consumption for my average engine speed/load last tank.

Change your average engine speed/ load (even just turn on the A/C) and your calibration goes out the window. This makes it difficult to compare driving styles.

If you take your city calibration and go on a highway run, you might be out by 10%, which is not uncommon on the cars I've got. Accelerating with fuel enrichment, likewise won't register on a system that only measures air.

I drive my Renault all city, but super consistently, last 3000 miles worth of fill ups all within 1 mpg, which gives the SGII an easy life - even then I'm lucky if SGII is within 10% (+/-5%). I know that on a long highway trip, the SG becomes little more than a guess o meter.

By comparison, a Guino that measures fuel, rather than air was able to find the BSFC sweet spot on an older vehicle that I had, a nice dip in consumption right at 80km/h, which of course became my cruise speed.

I don't really wanna argue, but my experience just does not fit your scenario. I don't believe anyone sees 10% random variation on averages measured on a set route in a single session with a calibrated OBD gauge like an UltraGauge or a ScanGauge. If that were true we would all get wildly different readings whenever our tank averages had more highway or city or hills or altitude than when calibrated. I just recently drove my car to the high desert (4000+ ft elevation and hot and dry) from the beach climate I calibrated within (sea level and humid and cooler). My tank average as shown by the pump and the OBD gauge was totally regular, which is to say it was within about one or two percent as usual.

The only thing necessary is to design a test that minimizes the possible sources of error and then taking the results nonetheless with a grain of salt, so to speak.