Ther impedance would vary by the amount of level difference.
If the amps were totally equal, there would be infinate impedance,
and thats not the way to test an amplifier to promote its
shorcommings.
greg

What you're saying is that the amplifier should be driving a realistic
load for a test to be maningful, and of course you're quite right.

This kind of test has been done before, by potting down the amplifier
output to the same level as the input and using a differencing amplifier
to compare the result with the input signal. You put the "real load"
speaker in another room where you can't hear it, and listen to the
output of the differencing amplifier.

Once HF phase shifts and gain had been balanced out, the result in the
tests I've read about was silence. It was also instructive to
investigate the dramatic sudden noises that would occur in the
difference signal if the amplifier under test was driven to clipping, or
if it was class B and the quiescent current reduced to produce crossover
distortion.

Anahata