@erik_squires 

Hi E
I have used several iterations of that design in my system with little success. The last time I tried using two speakers in a kind of horizontal array with two AR4c.  It was better but still not totally free of distortion. From time to time I was still left wondering what was the actor saying. And my speaker was not in a cabinet by the way. In that quote I referred to, Toole was talking specifically of center channels. I am glad that yours sounds fine.  I thought about designing my own too but got discouraged at the end. Then I realized the solution was in front of my eyes all along. My front speakers are Audience 1+1, they have zero crossover distortion due to the fact that they have no crossover to start with and they are clear as a bell.  They use a single full range driver that can go down to about 100 hertz in my system where they crossover to my four ML subs. So I got one more and now the center channel intelligibility is a solved problem in my system. Here are some more quotes from Toole about a design solution for the MTM design. 
“These designs also show up in vertical arrangements, in which case the acoustical interference is heard after reflection from the floor and ceiling. An intermediate configuration, sometimes called the 2 ½-way, rolls off one of the woofers at a low frequency, allowing the second unit to function as a midrange. The result is a slight improvement in overall performance, but the horizontal-plane interference pattern is then asymmetrical and still not what is needed. The real solution is to add a midrange loudspeaker allowing both woofers to be crossed over at a frequency sufficiently low that the acoustical interference is avoided. The explanation is in the caption.”

I cannot copy the caption but it shows the design with the midrange driver in the middle with the tweeter on top of it. I will find a way to include a picture later.  Hope that helps. 
I understand where Toole is going with this, and in absolute arguments, comparing them to ideal, he’s right, they are not optimum. In practice I’ve been quite successful. Maybe I accidentally created the best center channel ever??? << laughs >> Sorry, you all know I don’t believe that. Let me share the design I used. It sounds great. It blends seamlessly with my sides, which are simple 2-ways. With a little EQ, and decent room treatment the sound is seamless and there is no extra distortion caused by using an MTM array and it is perfectly fine to the sides as well.

https://speakermakersjourney.blogspot.com/2016/03/introducing-lm-1c.html
As I’ve alluded to, but not really said, for a counter point to Toole’s arguments, please see Joseph D’Appolito on his eponymous speaker configuration.

P.S. The LM-1C uses 2nd and 3rd order filters.
Just to be clear, @spenav : You and Toole seem to be discussing two different things.

You keep saying distortion and dialogue clarity. If we’re talking about on-axis, then nothing Tool is describing really should create a lack of dialogue clarity. The only thing I can think of is if you have a bare floor, which is causing cancellation due to the wide dispersion. However I hear NO dialogue distortion at all. Certainly not like you are describing.

Toole is focused on off-axis frequency cancellation, but the floor bounce of the Paradigm CC 208 should be quite similar. On axis, I don’t see how the 3-way inherently resolves distortion, nor can I see how the 2-way would cause it.

I wonder if the problem isn’t just that commercial 2-way center channel speakers are bad? :-) Truth is, I haven’t heard a consumer, off the shelf center channel in 15 years at least. If given the chance to listen to systems it was always for music. About 5 years ago I built my own center and never thought about listening to them in a store.


Best,
E
Sorry for the confusion about my use of the word distortion. I am talking about it in generic terms. You and Toole are using it in a more polished way than I am. I talked about it as whatever changes the original signal. I know you are kidding about creating the perfect center but you just might have. If the dialogue is clear and intelligible then you have accomplished the goal. Problem solved. I went a different way but am very happy with the results. I have always have an aversion to passive crossover. I am ok with active. I am familiar with the D’Appolito design and while it has solved a lot of the problems with TM and MT configurations, it introduces its own difficulties in regards to crossover and drivers distances and sizes. Guess nothing is easy and implementation does count for something 😊.  Glad you pulled it off. Stay safe.