I would only add one point to this discussion. It is folly to talk about "tube amplifiers" and "ESL"s in general terms, because each and every tube amplifier and each and every ESL in the marketplace has its own idiosyncrasies of design that strongly influence the interaction between the two. In my opinion, tubes and ESLs ought to be an audio match made in heaven. (I have owned nothing but tube/ESL systems since the early 1970s.) However, in the era of solid state hegemony, many ESL companies have designed and marketed products that are tailored to work well with SS amplifiers, which usually has a corresponding negative effect on the capacity of the speaker to be driven by a tube amplifier, and/or there are designs with complex crossovers that impose impedance curves unfriendly to tube amplifiers. Martin Logan seems to have adopted a pro-SS amplifier philosophy many decades ago. Their original ESL, the CLS, had a nominal 16-ohm impedance and was a beautiful match with the Futterman amplifiers that I owned at that time. They followed that up with the "CLS II", which I bought before auditioning them, just because I loved the original CLS so much. The CLS II was a gross mismatch for an OTL tube amplifier like the Futterman, because it introduced circuitry that created an impedance as low as 2 ohms at critical midrange or upper midrange frequencies. Zero autoformers did not exist at that time, and the mating of the CLS II with my Futtermans was a disastrous non-starter. The moral of this story is that it behooves any buyer to educate himself as regards the technical aspects of the components he wishes to match together in one system. There are no rules one way or the other.
I am now using Atma-sphere amplifiers to drive Sound Labs 845PX speakers. Frankly, this was not a perfect situation until I made some rather drastic modifications to the Sound Lab input circuit such that the speaker now presents an impedance that is 20 ohms or higher across the audio band up to about 5kHz, where it falls to about ~8 ohms and ends up at ~2 ohms at 20kHz. (I know this, because I made the measurements myself.) As someone else pointed out, an ESL is in effect a giant capacitor, and there is no avoiding a drop in impedance at very high frequencies. However, it is my opinion that so long as the impedance remains reasonable at least up to 10kHz, the low Z above that frequency is nothing to worry about; there is very little energy demand on the amplifier for frequencies above 10kHz, and the capacitative nature of the load should not bother a good tube design, whereas it can play havoc with SS amplifiers. I might add that, thus modified, my 845PXs can be driven by much less than 50W of good tube power. The ~100W Atma-sphere amplifiers are loafing even during loud passages.
