Different authors make different choices. In this case there can be an isomorphic transformation between watts and dB so the choice of one over the other is in some sense arbitrary.
Also, since I've got you on the horn, you as well as everyone else who bothered, ignored the diode effect I had mentioned. Would you mind sharing your views on that as well?
There are several diode effects related to wire. In the end, if they were significant they could at least be measured.
Commercial speaker and interconnect wire is usually carefully made so that diode effects are pretty much avoided.
Copper and other materials used in fabricating wires and connectors are themselves highly linear stuff.
Over a period of time corrosion by oxygen, sulfur and other contaminants can lead to the creation of chemical compounds on the surface that are nonlinear. In my experiments highly corroded copper wire remains highly linear, even if finely stranded and heavily corroded, provided you make a good electrical connection to the actual copper core.
In some of my experiments I took heavily corroded (literally green-blue) stranded copper wire, cleaned up the ends with fine sandpaper, soldered the wire to electrical contacts, and tested it for linearity. The wire remained linear to well below the -110 dB level. That is something like my local measurement limit.
I've seen tests that claim to have found non linearity in copper wire somewhat below the -140 dB level but those experiments were questionable simply because audio measurements at that level are inherently flaky. At that level the thermal air currents in your living room are magnified so that they are more like a hurricane.
The threshold of human hearing for nonlinear distortion under very ideal conditions is more like -66 dB. Under typical conditions it may be has poor as -20 dB or even poorer.
So whether tests show nonlinearity 100 dB down, 110 dB down or 140 dB down its all sonically moot.
OTOH make a inadequate connection with corroded cables and you can have large amounts of measured distortion: 30% or more. Corrosion can simply mute the signal.
Edited by arnyk - 12/19/12 at 7:41am