Just a couple of questions.
Presumably those plots were taken from the voltage on the amplifier output, not from the sound from the speaker.
If the speaker's impedance varies at different frequencies then doesn't the power dissipation and thus output vary significantly?
If you plotted the real power dissipated in the speaker, you'll probably find it varies hugely as the frequency is changed. Does that mean the output will be distorted by the same factor?
A1: Yes, the power supplied by the amplifier to the speaker does vary with frequency, a lot. Importantly, it can be reactive power: a capacitive or inductive speaker impedance will simply sling the power right back into the amplifier (which must dissipate it, as heat).
A2: The sound output from the speaker does not vary nearly so much. Speakers are designed so that the sound output (usually measured as sound pressure level, on-axis and 1m from the speaker) is proportional to the
voltage on its input terminals. The transduction efficiency, and hence the current drawn from the amplifier, varies with frequency, but this should ideally not affect the output SPL. Usually the variation with frequency depends on the resonant frequency of the various drive units and enclosures within the speaker. Transduction becomes more efficient close to a resonance - though resonances bring problems of their own. The directivity of the speaker also varies with frequency: this may show up in the impedance curve, but even if it does not, varying directivity with frequency means the off-axis frequency response may be anything but flat!
All these effects are, in principle, linear and so do not constitute harmonic distortion. Speaker distortion is usually much worse than that of the amplifier, maybe around 1%, though the very best electrostatic speakers (like the Quad ESL63s I'm listening to now) are down around 0.1%.
Max