To Turbo FC3S:
Well your questions are certainly fair.
I get my ideas of driver directivity from a loudspeaker text, from my own measurements (extrapolating from measurements of larger diameter woofers), and from a loudspeaker modelling program that has a solid track record in the industry. Unless your measurements are time-gated to exclude reflections, they aren't giving you a reliable indication of the off-axis response because reverberant energy is being included in the measurement. I use time-gated measurements for that and other reasons.
However, I do see what you mean from the curves on that Peerless woofer! I wouldn't say that the woofer is omnidirectional at 600 Hz based on those curves as they only go out to 60 degrees, but they certainly do seem to contradict what I have posted. Now if Peerless took their measurements on a one-meter wide baffle that would account for some of the discrepancy, but not all of it. Unfortunately I don't have a way to post the curves run by my simulation program (LMP by LinearX).
Loudspeaker cones behave as rigid pistons at some frequencies but then transition to the breakup region. In the breakup region, the radiation pattern can be over twice as wide as rigid piston theory would predict. While that may apply to the Peerless cone, I don't think it applies to the Seas aluminum cones within their operating passband.
I definitely believe that a low-resonance tweeter can go as low as Bob uses them because I understand what collapsing the directivity does for a driver. Now if the waveguide is too small to properly load the driver, that introduces problems in terms of directivity, frequency response and power handling.
Duke
Edit - I just thought of another reason why that Peerless woofer could have a much wider pattern than what I would have expected: The bullet-like pole-piece extension. From what I understand such features are claimed to reduce beaming, and maybe that's true - I haven't really studied much about them.
It looks like the woofer in the Mini has a similar feature, which (if it really does widen the pattern significantly) would mean that you were probably more right than I was regarding the radiation pattern similarity between woofer and tweeter below the frequency where the waveguide loses pattern control.