My memory is that Bob used a transistor amplifier. (Indeed, part of the point
of the experiment was his claim that he could make a transistor amp sound like
a tube amp by approximating the latter's frequency response.) An amplifier
with an output transformer would be less "fungible".
Bob reported that at least some amplifiers' transfer characteristics varied
with line voltage. Given that transistor amplifiers generally have follower
outputs and overall feedback, this is somewhat difficult to believe, but I'm
not going to gainsay Bob.
* * Where an audible difference REALLY exists, there will be an easily
measured difference too. This what Bob found and figured he could
emulate with a few passives added to one of his designs. It is not
necessary to match amps so precisely to make them sound
indistinguishable in normal listening circumstances.
Although the final sentence of this statement is almost certainly true, the
belief that audible differences necessarily correlate with "easy" measurements
is wishful thinking, for which there is no proof or contradiction. The
more-general statement that, if something is audible, it /must/ be measurable
is -- almost by definition -- true.
I agree strongly. The human hearing system is a relatively
inaccurate tool for hearing fine details which show up glaringly
in some very simple tests and measurements.
And your proof or evidence for this is...?
Perhaps the fundamental problem with this dispute (which has been going on 60
years) is that there are no hard data about whether amplifiers really do (or
do not) "sound different". Without such data, it is impossible to even begin
to establish meaningful correlations between what is (or is not perceive), and
measurements.
Both "sides" are wrong, and hold unscientific beliefs based on what they want
to think is true.
PS: If you'd like to hear an amplifier that "measures good" but "sounds
awful", try one of the Crown K series, which were discontinued about a year
after they were first manufactured. They were designed by Gerry Stanley, who
had an outstanding 40-year track record for amplifier design. He apparently
botched this one.