Schroeder Reference Arm

Doug,
You're very right about the process--with one addition, perhaps a very personal one: I love accidents, mistakes and wrong turns. In musical composition (which many people don't see as rigorous, but actually is, just as much as scientific explorations, in different ways) as well as in tonearm making, I've often come to theoretical understanding after empirical discoveries resulting from unreasoned decisions (what if? oops! what did I just do? etc...).
Hypotheses sometimes get proven by testing, but more often than not, they just serve as jumping boards to jog the brain into action, to bring oneself into places one hadn't thought about.
If you get a better result than before, you try to reproduce the outcome again and start generalizing... until you find a new reason to move on and discard the theory you just established.
This is fun.
Joel

Lewm,
I'm not advocating setting azimuth by ear as necessarily the best method over any other one, but one thing I've noticed is that using only one frequency (1kHz, or another one) to measure crosstalk isn't terribly reliable. When I managed to establish a good balance at 1kHz, I measured that it was not right at other frequencies. Obviously, a musical signal is extremely complex because of the number of acoustic waves occurring at any given time, but also because of the way they interact with each other (addition, subtraction), to stay with just the frequency domain. I wonder if one could make a recording where, instead of just single frequencies, there would be "blocks" of stacked up sine waves in various groupings across the spectrum... But even that might not be realistic enough to parallel the complexity of musical signals.
Joel

From what I've learned about the Foz and Feikert's software I would expect this. The Foz would get one close, and the software package would get you pretty much dead on the theoretical optimum. If one can do it by ear, it gets to be much faster. I would ask how much deviation from the theoretical optimum can we tolerate?

Dear Joel et al, By choosing 1kHz in my little proposal, I did not mean to take the position that 1kHz is the sine qua non for setting azimuth. The goal of my proposed experiment would simply be to compare the outcome when a really good ear (like Doug's or Paul's or Tbg's) sets azimuth according to his/her preference vs when azimuth is set by a typical audiophile conventional electronic method ((based usually on a 1kHz test tone). I have no preconceived notions (i.e., no hypothesis). It's actually dangerous to design an experiment to prove an hypothesis. It's better to acquire some good data and then make an hypothesis centered on the data. Further experiments then test the strength of that hypothesis. If you can develop a way to use a set of frequencies to set azimuth electronically, rather than just a single frequency, that would be a useful improvement on the electronic method, IMO.

If one could determine what frequency range that crosstalk is most discernable to the ear. Then maybe there would be an ideal test tone to use.