A few weeks ago I dragged my Vandersteen Quatros out of the closet where they've been for a few years (original cloth model; bought these well before the the wood or "carbon" models became available).
The basic design principles of these don't seem to have changed much from the 70s. In fact to this day in their promotional material, Vandersteen says that the most important measurements are: Impedance, step response, and waterfall. Always good to have a benign impedance curve, but the other two would seem to be irrelavent according to the Toole/Olive research.
Atkinson's measurements are kind of hard to interpret, but they don't look all that great either on or off axis:
Sidebar 3: Measurements
www.stereophile.com
These are 4-way speakers with first-order crossovers. So they have a ridiculously narrow vertical sweet spot. Get the vertical tilt wrong and they sound muffled (a common complaint about Vandersteens).
To add to the fussiness, they require an expensive 6 dB/octave @ 100 Hz high-pass filter box that goes before the amp. This filter box requires a 9V battery that's soldered in. The idea is that the built-in subwoofer amp "undoes" the highpass filter and this takes a load off the amp. Which is good because they have a sensitivity in the low 80 dBs. I do the highpass in the digital domain now.
However... if I set them up well away from side walls and in an 8 foot equilateral arrangement (you don't want to get much closer because of the first-order design) so that the reflected sound is minimized, they are quite satisfying. Part of this must be the built-in subwoofers which give much better integrated bass than I was able to get with a subwoofer and 2-way bookshelf speakers. Also, I think the midrange drivers must be very good. And maybe there's something to optimizing the step response at the listening position.
They also take much better to full-frequency room correction (with Acourate) than I would have assumed.