Calanctus, Your speakers are a good example of what I was talking about. ESLs in general, regardless of their particular impedance curve, want to see the same power for a given sound pressure regardless of what their impedance is at a particular frequency.
In effect, they are driven by power.
If an amplifier driving speakers like this puts out a different amount of power into different impedances, the effect is a shift in tonality. This is why transistor amplifiers (in general) tend to be very bright on ESLs, as their 'constant voltage' characteristic causes them to put out a lot of power into the low impedance of the ESL at high frequencies. Some speakers (for example B&Ws) are designed to expect a 'constant voltage' amplifier, and so will have flat response in the room. This is much trickier with ESLs!
In essense, there are two paradigms competing in high end audio today- the 'voltage paradigm' and the 'power paradigm'. No-one talks about this!- but we see it all around us:
The Voltage Paradigm is the reigning test and measurement paradigm. It is only concerned about voltage, and so when measuring an amp or speaker, voltage is the only thing considered. 'Constant voltage' output refers to an amplifier that puts out constant voltage regardless of load (contrary to morbious' comments, this does not require power supply regulation to accomplish this- merely enough feedback). Such an amplifier is regarded by these rules to be 'load impervious'.
The problem lies in the fact that negative feedback runs counter to how our ears detect volume (higher odd orders being the key- these are enhanced by negative feedback). Additionally there is the question of whether or not a speaker is driven by voltage or power, and of course the answer is power. So the 'power paradigm' says that the amplifier will have little or no feedback to reduce the odd ordered harmonic that the human ear dislikes, and that the amplifer will produce (or attempt to produce) constant power into all loads. This, BTW, is very nice for ESLs, horns and magnetic planars. Amps that fall into this category are SETs, some OTLs, transformer coupled push-pull tube amps and a very small number of transistor amps with zero feedback. Such amplifiers typically have higher output impedances, and nearly all will be unsuccesful at the goal of constant power (in fact many designers of such gear may not even acknowledge that they even *have* such a goal).
A parallel controversy is the subjectivist/objectivist debate. Roughly, the power paradigm is more subjectivist and vice-versa, but this is *not* by any means cast in concrete!
The power paradigm has its roots much further back than the voltage paradigm (1920s), and only seemed to resurface in the last decade or two. The obvious proliferation of advanced SETs (and tubes in general), horns, single driver full range speakers and the like are an indication that no-one has all the answers.