I would think that would alter with the speakers and their efficiency/impedance, etc. . . .

Thank you Dan!  Now when my pals tell me a 6 Watt tube amp can not possibly have enough power to be more than barely audible, I can tell them they have been enjoying their own equipment at less than that power.

Thank you Steve & Lin.
Steve, A Lot of good information in those Decware Audio Papers. I have not read most of them since they were new. I have gone back this spring and started rereading the entire lot. That is what has raised many of the questions I have recently been asking.

Lin, Interesting points you make. I like the Nelson Pass quote. I have been a little worried about that myself. If I go to a too large amp and make power I don't use, am I then getting the best sound.  Somewhere I once read that tube amps sound their best at half throttle. If the amp is a small one and the speakers are low efficiency, the amp is run full blast and lacks headroom; therefore the music lacks dynamics.  If the amp is large and the speakers are high efficiency, the amp is barely turned on and for some reason this makes the music sound lifeless or uninteresting.  
That, I think was in one of these Decware papers, if I am remembering correctly.

On the other hand, this new 40W Mystery amp is being well loved by everybody! It is all a bit confusing.

If you want to know how many watts are available at what volume pot setting, you really need to know the gain structure of the amp in question.  Then you need to know what drive voltage will swing enough volts for the max unclipped output swing.  While measuring the output swing then you could see where the pot is set to give you your position.  So, for example, I have a HK DC450 that puts out about 40w per channel.  At 1-oclock on the dial it's pushing almost 40w, due to its gain structure and the voltage going in from the source.  Also, if you don't know the voltage of the source it's even harder to calculate.

An easier way, would  be to see what the sensitivity of speaker is rated for, say 90db at 1watt (we presume 8ohms), then with a portable SPL meter you could measure the sound level coming out of the speakers.  So in that example, at about 1 meter from one of the speakers you'd be at about 1 watt continuous. Then you could do the math to figure out how many watts it takes to produce the sound level being listened to.  Your friends would probably be surprised to see how little watts are needed for adequate listening levels.  That's why a 5 watt amp can work so well in a large number of systems.  Also if you bi-amp and use something more powerful for bass you can use your 5W amp in even more systems.

Also, lots of global feedback, that is typically used in today's commercial amps can give the impression of not a lot of power at lower volume levels.  A lot of SS amps actually sound better when the volume is cranked.   A tube amp, on the other hand, with no or little global feedback actually "sounds"  8-) louder than is really is.  Also when you push the tube amp into clipping it's more of a "soft" clip that is harder to hear.  When the SS amp clips it's much easier to hear, so you'll want the extra power to try to avoid any clipping under any volume condition.  YMMV

The accuracy of most power meters is not too great since the amplifier doesn't know the constantly changing impedance of the speakers and they are too slow for true peak readings.

Similarly an spl meter can be useful, but not terribly accurate in determining actual wattage from the amp.

From Tom Danley on the What's Best Forum talking about high power amps.
"For me this happened when I had a chance to be in an official ABX listening test of a number of amplifiers. I brought in my trusty Threshold stasis as my known reference and used speakers I had designed at work and had at home as well.

There was plenty of time and I went through a number of my reference CD’s to find a few passages that brought out the differences best and then went to “without knowledge” listening. At this point one could still hear some differences between them but they generally fell into two groups which I could reliably hear. What I heard were differences in the decay part of the sound which I have no explanation for (I used to build amplifiers too).

A funny and unexpected thing also happened which is why I am writing, in comparing the various pro-amps to my Threshold, I found that one group and it were indistinguishable except the Threshold began to sound “less dynamic” as I increased the level. AS the Thresholds VU led scale showed the peaks were -20dB or so down, I was puzzled. Connecting an oscilloscope to the output revealed the truth.

At a point WAY below one could hear it as a “flaw”, at a point that indicator showed peaks as 1/100th rated power, there was instantaneous voltage clipping.
To be very clear, this WAS NOT audible at all as a flaw or heaven forbid the familiar “clipping” the only clue was that compared to a much larger amplifier, the Threshold was somehow less dynamic above a modest level.

At that point, I switched to a larger pro-sound amplifier at home too. Currently my upper speakers (SH-50’s) are about 100dB 1w1m and I use an 800W/ch power amp on these above 80Hz.
With any commercial recording or movie soundtrack, this is more than enough headroom to never breathe hard but I can still reach instantaneous limiting with the fireworks recording.

Anyway, the big and unrealized advantage of having more power is that the short voltage clipping events reduce the dynamics long before it’s audible as a flaw. This can be seen (when present) by using an oscilloscope to examine the Voltage waveform going to the loudspeaker.

To be clear, just as a subwoofers job is to extend the response and that is not the same as “turning up the bass”, this IS NOT the same as “cranking it up” although a larger amplifier lets you do that too if your speakers are up to it.. This is about reproducing the peaks (in the “good” recordings) which are often clipped off much more often than people realize."

The argument continues, some think most people have 10 times more power than they need, other think 10 times more power is needed.
I might agree with one point of view today and change my mind tomorrow depending on what music I'm listening to, what pieces I have connected together, which room it is in, how close I'm sitting, and how loud it is playing (never real loud -  I want to keep my hearing as long as I can).

We all hear and listen differently, one person may not notice something that is not allowing another person to enjoy the music.

Another thing I don't think very many people put much if any thought to is "how loud is loud?" What I mean is if a person regularly listens at @ 80dBs, 90 dBs sounds really loud, whereas a person listening regularly at 90 dBs thinks 80 dBs is barely on.

I've posted to Steve that there are forum issues, and I'm going to call him this morning about the Mini-Decfest, so I'll bring it up there as well.

Hah!  Sorry you deleted. I thought those were good points you made.

Tom said it was voltage clipping, according to the VU leds peaks were at 1/100th rated power.

There is a lot more involved than what you are speculating (although I don't know what every variable is myself).

A lot depends on the load the speaker presents. I have used 12 and 15 watt tube amps that drive my low impedance 89 dB 1w/1m speakers very well and I had a 50 watt p/p u/l tube amp with more micro-farads of capacitance than Decware's mono amps that was unable to drive the bass section properly. This same amp sounds phenomenal on 95% of the speakers out there (i.e. speakers that don't require a welder for current).

Danley's speakers are 100dB ~ 4 ohms, so one would think any amp rated/designed for 4 ohm loads regardless of the power will work fine and yet he uses an 800 watt amp.