The symmetrical arraying of dual mid-range driver lines around a line of tweeters does create several potential advantages as pointed out by Rick. You can think of this design as a stack of MTM elemental speakers. The key to successful utilization of this design (as with any horizontal MTM speaker) is to cross over to the tweeter from the mids at a low enough frequency to mitigate lobing issues which would narrow the horizontal plane coverage for off axis listeners. Crossing low necessitates a very close spacing between the mid range drivers and a small width tweeter to assure a wide horizontal lobe in the horizontal plane.

The McIntosh XRT28 symmetrical line array design is an example of a line array design that uses the two mid range driver lines and a line of small domes concept. While the XRT28 does have dome tweeters, the drivers are rather different from common small dome tweeters. For example, the Aura Sound NSW1-205-8A (I suspect that McIntosh uses this driver in their XRT28 as the photos indicate this) is a small 1" diameter full range driver (Fs of 220 Hz) which can be used as tweeter in such a symmetrical array. That small driver enables the mids to create a wide horizontal lobe which tight spacing between drivers. The mid to tweeter crossover can be very low in frequency which assure a broad lobe in the horizontal plane. The NSW1-205-8A is listed by Madisound at:

http://www.madisound.com/aurasound.html

You can gain additional information on the Aura Sound site as well.

Robert Greene tested the XRT28 for the Absolute Sound magazine and adds additional analysis in his on-line site at:

http://www.regonaudio.com/McIntosh%20Supplement.html

Greene points out the advantage of this arrangement as the horizontal lobe somewhat limits side wall reflections within the listening room. Thus you'll experience both minimal ceiling and floor reflections inherent with a line array plus minimal sidewall reflections. The sound that reaches the listener is more direct (vs. a mixture of direct and reverb sound) so any room signature is limited.

One final point is that you really need to shoot for a center to center spacing of a wavelength (or less) spacing between the midrange drivers for an optimal near field array. That is the vertical spacing of the drivers. What happens is that the amplitude response of the midrange line tends to droop as you go beyond one WL spacing.

In the dual midrange symmetrical array you need to shoot for a center to center spacing between the mids in the two lines of a wavelength as well to create a favorable lobing situation.

Jim