Originally Posted by Bill Fitzmaurice
There's nothing theoretical about it. When you measure the sensitivity of a speaker it's done with a constant voltage signal, because while current and power vary with the load impedance, voltage swing does not.
It's all theory sir! The machines that are used for measurement, are running mathematical extrapolations, based on merely two knowns. Namely, Voltage (because its easier than current to measure) and resistance (because it too is easy to measure). The rest is all very much programed theory.
And when actually 'measured', as in at 1m/1w the signal agitation is pink noise, which doesn't output a constant voltage. The sensitivity score is then selected, as being the frequency (or pass band, depending on standard being used) exhibiting the highest SPL @ 1m/1w, or 1w/1m, however, you prefer to express it.
The speaker doesn't have a uniform / linear sensitivity to all frequencies throughout its pass band, that is quite impossible. The spec published is the highest.
They should publish the actual sweep, so we would all have a more informed understanding (but many don't take actual measurements any more, they theorize what the performance is likely to be, and when they do, the scores are always bolstered / higher to the linear tune of 3dB vs real-world measurements), however, one could look to a frequency response sweep (if available), then locate the highest peak, and reasonably assume that is where the Sensitivity spec is based.
Regardless of being measured or purely theoretical, current resides within every volt in a locked relationship, based on load, no exceptions, not even in here!