W5DXP wrote:

Let's say we have two sources each equipped with a circulator and load
resistor. We'll call such a source an SGCR (Signal generator equipped
with a circulator and load). We set up the following experiment.

100W SGCL#1------------50 ohm lossless coax-----------50W SGCL#2

Seems to me that the 100W from SGCL#1 will flow unopposed and
dissipate in SGCL#2's circulator load resistor. Seems to me that
the 50W from SGCL#2 will flow unopposed and dissipate in SGCL#1's
circulator load resistor.

May I modify your experiment slightly? Change SGCL#2 to produce 100W.
Restating your explanation for the new experiment:
"Seems to me that the 100W from SGCL#1 will flow unopposed and
dissipate in SGCL#2's circulator load resistor. Seems to me that
the 100W from SGCL#2 will flow unopposed and dissipate in SGCL#1's
circulator load resistor."

Let's also make the 50 ohm lossless coax long enough that we can
find a voltage maximum. At this voltage maximum the current is 0,
always.
Assuming that P = V x I, the power is 0, always.
This seems to be at odds with explanation that SGCL#1's power is
dissipated in SGCL#2's circulator load resistor since there is no
energy flowing at the voltage maximum.

The obvious (and probably controversial) answer to this
inconsistency is that the power dissipated in CL#1 originates
in SG#1. Similarly for SGCL#2.

To further the experiment slightly, whenever a conductor in a circuit
has 0 current, the conductor can be cut without changing a thing.

Let's cut the coax at the voltage maximum (where the current is 0).
Nothing changes. The voltage and currents remain the same every
where. The powers remain the same everywhere, and yet there is no
longer a path from SG#1 to CL#2. The power being dissipated in CL#1
must be originating in SG#1.

Comments and corrections invited.

....Keith