Frank wrote, "Prove it."

OK, here I am at the track (the bench). I have an SWR meter that I've
verified with my HP8653 to behave like a short section of 50 ohm line
at the frequency of interest. I put a load on its output that I've
also verified to be 50 ohms at the frequency of interest. I've applied
power to the load through the SWR meter. The indicated SWR is 1.23:1.

I took the SWR meter apart, and located a particular resistor. I
changed its value slightly. I re-verified that the meter still looks
like a short section of 50 ohm line. I re-ran the experiment of
applying power through the meter to the load. The indicated SWR is now
1.05:1.

Yes, I really have done that!

This particular meter is built, as very many of them are, to sample
current and voltage at a point of essentially zero length on the line.
The current sample (through a current transformer: line center passes
through a toroid; secondary is several turns, loaded by that
calibration resistor) is converted to a voltage by dropping it through
a resistance, and by changing that resistance, I can change the
relative amount the current contributes to the measurement. In other
words, if the voltage sample is v(samp)=k*v(line), I want to adjust the
current sampling so v(i(samp)) = k*Zo*i(line), where Zo is the
impedance to which the meter is calibrated to measure SWR. In some
meters, there is a means to adjust the voltage sampling ratio easily
with a variable trimmer capacitor. Either way works. The adjustment
DOES have a TINY effect on the impedance the meter presents to the line
it's in, but that is very minor, compared with the range of adjustment
of the impedance calibration value.

Yes, I really have adjusted a meter which uses the variable capacitor,
too.

Cheers,
Tom