You are telling us that you have evaluated the loss of a MFJ-989C
antenna tuner + open wire line vs coax by on air A/B tests using the
S-meter as your indicator.

Sure.

Presumably you used exactly the same antenna, how are you sure that
the feedline wasn't an active part of the antenna, was the measurement
made over an ionospheric path, was the A/B comparison made at
different times when other things may have changed?

The A/B testing is instant. All comparisons were done in the
usual up/down NVIS envionment. Mostly close in stations, and the
noise floor as another opinion. Just comparing any difference
in noise floor is a fairly accurate indication of efficiency in a
case like that. I didn't worry about the feedline being part
of the antenna, as I never would in the real world. I'm
running ladder line the whole way from the tuner to the antenna.
If the line is radiating, there would not be too much to do about
it at that point, without going to further measures. . I noticed
no obvious problems in tuning, or a "hot" line, rig, etc..

What would be your confidence limits on such a experiment?

Very high overall.

If you recorded your results, the variance would be revealing.

No. Not that time. It was just for my personal amusement.
In fact, I was out camping when I did that test.

I suggest that the noise in your experiment would probably swamp the
true difference between your A and B configuration, and therefore your
conclusion might not be sound.

What noise? I'm just using a switch to compare both antennas.
In that particular case, when switching to the coax fed, the noise
floor rose a bit, and strong signals that were way over S 9 rose
about 5 db worth on that particular meter. Easily repeatable
over and over. I'm not giving any numerical value to this
increase, I just know the coax fed system was the most
efficient. All signals increased when using coax. Noise floor,
desired signals, the whole deal. Switching to the tuner/LL fed
was like adding a small bit of attenuation. The way I see it,
the loss had to be at the tuner. And I was very careful in
tuning, using the least inductance to get a usable match.

Your observation is (as you note) different to "on paper"
expectations, that might be reason to re examine the experimental
setup in search of an explanation.

Why is it different? Actually, I found about what I was expecting.
I was expecting the coax fed to win. And it did.
Now, I realize this is a fairly sloppy test, and could be
improved greatly for accuracy, but there was no doubt
which system was better at that particular time.
The "on paper" expectations of the ladder line winning
would depend greatly on the losses of the tuner involved.
Sure, ladder line on it's own is very low loss, but add a
tuner, and things can change quite a bit as far as
system loss.
Now, granted the difference is small in general, but
I could see it. And if I can see it on a meter, it's not as
small as it's oft cracked up to be, or I wouldn't be able to
see it.
MK