I followed this thread and decided to run some tests with
my Syncro Westy with a Scan Gauge.  My idea was to run
a piece of highway in third gear and in fourth gear and
compare numbers.  I also ran a test at various speeds in
4th gear to see how speed changed things.

I picked TPS, LOD, MAP and MPG for the Scan Gauge to
report.  The first three, throttle position, engine load,
and manifold pressure are reported from the ECU.  The
last, MPG, is computed by some (unknown) algorithm by
the Scan Gauge.

I could do this because I have a Subaru 2.5 (OBD II).  I
also have, for reference, stock gearing and 215/75-15 tires
(27.8" diameter) and a heavy Westy.

First the speed test.  Resistance goes up by the cube of
the speed, or something like that.  Nothing here contradicts that:
-----------------------------------------------------------------
Varying speed test in Nevada (S of Boulder City) on April 8, 2012

        S       G       RPM     TPS     LOD     MAP     MPG
        75      4       3800    33      53      11.3    13.0
        70      4       3600    23      43      10.7    14.7
        65      4       3400    21      37      9.8     16.0
        60      4       3100    16      31      9.8     16.3

------------------------------------------------------------------

Now the comparison of speed and gear.  Two runs, two directions,
50 mph in 3rd and 4th gear and 65 mph in 3rd and 4th gear.
Three or four readings were recorded and these are the averages:
-------------------------------------------------------------------
        S       G       RPM     TPS     LOD     MAP     MPG
out     50      3       3900    8.7     21.7    4.6     20.5
out2    50      4       2800    9       18.7    6.5     21.7
-----------------------------------------------------------------
back    50      3       3900    14      31.3    6.5     13.9
back2   50      4       2800    16.3    28.3    9.1     15.6
-----------------------------------------------------------------
out     65      3       5000    19.7    39.7    6.3     13.8
out2    65      4       3150    14.5    32.5    8.2     17.7
-----------------------------------------------------------------
back    65      3       5000    27      57      8.5     10.2
back2   65      4       3150    45.5    52.7    10.9    12.4
-----------------------------------------------------------------
Out was slightly downhill and back was slightly uphill.   With
one exception, it takes more throttle to go a speed in 4th
than 3rd, but the load on the engine is less.  Mileage is
better in 4th than in 3rd.

Again, these are averages from one test run.  The raw data
is at: http://jones.colorado.edu/VW/ScanGaugeTest.txt
I wouldn't start a chip tuning company on these--I'd run at
least one more test--or maybe a dozen.  ;-)

Richard