Speaker priority: high or low???

I too am going to weigh in on the speaker end, with the provisos that (1) you do have to reach a certain level of amplifier and source performance, which varies from one individual to the next depending on what you can tolerate, and (2) really disparate levels of spending on source/amplification/speakers are bound to lead to a worse result than an intelligent distribution.

Having said that, there is absolutely no question that speakers have by far the biggest deviations and distortions from perfection, that is, they color the original signal far more than any other component, usually by several orders of magnitude! And the worst types of offenses are just "passed over" by the reviewers, e.g., a total lack of time or phase coherence, massive impedance and phase anomalies which put tremendous demands on the amplifier, highly inconsistent diffraction and dispersion patterns, etc. It's as if the reviewers take the attitude that "everyone does it, so it must be okay"! It's not even close to okay, it's just that very few speaker designers have tried to seriously address these issues.

I will make an even bolder statement: If you assembled a system of decent components including a typical 3-way rectangular-box speaker with a typical high-order multiway crossover, and then took a short sample of a complex musical signal and studied it all the way through the chain, it would remain largely intact until it reached the speakers, at which point it would instantly transform into a waveform that is literally nearly unrecognizable compared to the original. Of course, I could be proved wrong, but I doubt it. If anyone has done this, I would love to see a reference to it.

Please do not accuse me of suggesting that measurements are all that matter. I am merely suggesting that screwing up the waveform to such a degree that it hardly bears any resemblance to the original, cannot possibly be a good thing. There is a reason that full-range electrostatics sound so good, and it is that they avoid most of the problems I just mentioned. Of course, they have a whole new set of problems, most of which revolve around unsolvable room interactions, which is why most people don't put up with them.

So while I'm not saying that one should spend more money on speakers than anything else, I am saying that they have the most variability and thus deserve the most effort in searching for one that satisfies.

No system can ever sound better than the limitations of the "upstream" components. Since speakers are at the very end of the chain, though they may serve as tone controls, they can NEVER improve the signal passed to them.

Yes, David, you are right. The speaker cannot "improve" the signal presented to it. As I understand it, what Karls, Unsound and others have said (more ably than I) is that the speaker is the component that is least linear and most apt to impart its "sonic signature." There are clearly demonstrable, relatively gross (objectively and subjectively) differences between speakers. There are not such gross, subjectively and objectively demonstrable differences, for example, between a $500 CD player and a $2,000 CD player. A pair of AR3s is going to sound like a pair of AR3s with $20,000 worth of electronics and will sound like a pair of AR3s with $1,000 worth of electronics. --- The identifiable "quadness" of a pair of ELS 57s will remain regardless of the electronics. A bad amp may muddy the midbass, and present unacceptable colorations to the revealing midrange of the quad. And to be a bit hyperbolic: a deaf man would not mistake the AR3s for the Quads regardless of the electronics used with either or both.

Speakers are much more than "tone controls". They are the imperfect acoustic interface for all the "upstream" components. They are electrical-mechanical transducers with high mass and relatively large nonlinearities and other distortions compared to the electronic components in the system. Even with relatively "low-fi" electronic components, speakers are usually the most limiting factor in an audio system. Consequently, only with better speakers can you take fuller advantage of upgrades made elsewhere in the system.

Now, I am just a fat little boy who works for a cheese company, but in my experience, I can live (and sometimes quite well) with lesser speakers, if they are driven with better electronics. The reverse is not true. My main system is posted. I have swapped out my main speakers (5 driver tri-wired) for Warfdale Diamond 6's (2 way, single run, $149 a pair)(I admit, they have been modified). Once they were setup properly, the Warfdale speaker gave me most of what the main speakers do, with the exception of the lower registers.
If the truth be known, I fall into the Beemer camp of balanced systems, but I am willing to play. I would be much more inclined to live with speakers that were designed with some trade-offs and no glaring errors, as opposed to something better that revealed more of the trade-offs up-stream.
Again, this is just me.