If speakers were stacled vertically, wouldn't all the drivers, on a relative basis, be similarly oriented to the listener?
Because different size drivers produce diferent sized wave lengths and dispersion patterns the comb filtering occurs as those wave lengths (energy causing the air to move) interact when they encounter each other. It might be likened to small waves and large waves in a swimming pool.
The energy from both the small and large waves has an effect on each other. It is all very complex and my example is "oversimplification".
Like wave lengths like those from the tweeters if the speakers are stacked tweeter to tweeter would comb through each others energies as the dispersion from each intersected the other. The effect of the energy (wave distortion) is what causes the filtering, as they tend to cancel or reduce each other to a degree.
Since I am not an acoustical engineer, I am explaining this very badly. Maybe a dual garden hose example might offer some insight.
Lets say the "tweeters" were like having two garden hoses in your hands and you have each set to "spray". As the spray from one hose spreads out and encounters the spray from the other their energies (sprays) interupt each other and cause distortions not seen in the spray for a single nozzle.
Hope that offers a small amount of a "lay" explanation of the energy interaction.
Next week we'll discuss how this causes what is called the "venetian blind effect" in large single planar drivers like the magneplanar. LOL....