The problems you're encountering are the result of trying to
oversimplify electromagnetic theory by reducing it to conceptual models
which are far from adequate. Your basic conceptual models are faulty, so
any conclusions you draw from using these models will eventually lead to
contradictions. That same problem has, in fact, been the cause of a huge
number of argumentative postings on this newsgroup.

You have a great deal of curiosity about electromagnetic phenomena, so
you'd benefit a great deal from a bit of education. To fully understand
electromagnetics, you need a solid background in mathematics. Without
the solid background of math and electromagnetic theory, you'll always
find it necessary to use oversimplified models, and those will always
lead to contradictions. It takes considerable time and effort to gain
the necessary background, but if you truly want the answers to your
questions, it's the only way.

Roy Lewallen, W7EL

David wrote:
I find that there are many anomalies in RF theory. For example, Maxwells
Equations show reflection at a boundary where current flows in the metal
surface down to the skin depth. Books then state that this boundary
condition applies to waves guided by two wires. The wires therefore sort of
reflect the wave and guide it. Power flows as the Poynting vector.

Current can be a flow of electrons or holes. If the current in a P type
semiconductor is holes, it is a flow of emptiness or nothing.

Current is normally said to be the flow of electrons. The electrons actually
move very slowly with a drift velocity of a few mm per second. The signal
part of the current that flows near speed of light is the electromagnetic
wave that flows in the area outside the conductor.

There seem to be a number of disputes in RF e.g. about the effectiveness of
conjugate matching with Walter Maxwell refuting articles in his book
"Reflections".

RF seems to be a black art e.g. when it comes to S parameters, network
analysers, phase matching and batch matching of cables. Little seems to be
written down.