Frank Gilliland wrote, among other things, "The point is that the error
is insignificant when the directional coupler is much shorter than the
wavelength."

Certainly "directional couplers" for HF may be built at essentially
zero length, and ideally would have exactly zero length, monitoring the
current and voltage at a single point on a line. Then SWR or
reflection coefficient magnitude or even complex reflection coefficient
may be calculated under the assumption we know the desired reference
impedance. But if the equipment combines the voltage and current
samples in the wrong ratio, you will get the WRONG answer. Even if the
coupler looks like a perfect 50 ohms impedance section of transmission
line (with some attenuation), the error _in_measurement_output_ can be
significant indeed. Just because the coupler looks like a 50 ohm line
to the line it's hooked to doesn't mean it will read zero reflection
when IT's presented with a 50 ohm load.

And by the way, not everyone who measures and cares very much about SWR
(or reflection coefficient) cares a whit about field strength. Not all
loads are antennas.

Indeed, as Reg says, we might do better in amateur applications to
consider the SWR meter as an indicator of the degree to which we're
presenting a transmitter with the desired load. That's really what
we're using it for, most of the time. It may ALSO be interesting to
know the field strength, but please be aware that a transmitter's
distortion products may be significantly higher if it's presented the
wrong load impedance, even though the power output may be increased.
Field strength alone is not acceptable to me as a means to adjust an
antenna load to a transmitter, or as a way to adjust the operating
point of the transmitter.

Cheers,
Tom