Ok, ok, I'm hear!
Bdgr gives a good reference. Here's a little elaboration.
Speaker drivers have a thing call their "acoustic center". This is the point in space where the sound waves first emerge. It's approximately where the voice coil is. A woofer voice rests further back than a tweeter voice coil, so likewise, their acoustic centers are misaligned.
For about one octave centered around the crossover frequency the woofer and tweeter work simultaneously. If we draw a line perpendicular to their acoustic centers we find the direction the speaker effectively radiates at the crossover frequency. And since the drivers are also vertically separated they create an interference pattern like that illustrated in the Tannoy manual. The result of this interference is an equal loudness "lobe" pattern. Within a small window near the center line of the lobe pattern all is fine. But as you change angles vertically relative to this center line the response can change dramatically. You can wind up at a null resulting in a big frequency response dip around the crossover frequency.
A properly designed monitor will use some method to align the center of the lobe with the center line of the cabinet. This can be done by physically moving the tweeter back as with the horn loaded Mackie and Genelec tweeters. Other monitors do this electronically using some sort of circuit that adjusts the relative phase alignment of the woofer and tweeter. New digital designs can add a time delay to the tweeter. Aligning the lobe with the cabinet center is definitely desirable, but the drivers are still vertically separated, so you can't get rid of the lobe itself.
Ok, now we get into some heavy duty physics! There's this obscure little force known as "gravity". It's the thing that tends to keep our feet planted on the floor and our asses planted in our seats.

In other words, we move around a lot less vertically than we do horizontally. And since the window of good response from the lobe pattern is pretty narrow, it's preferable to align it vertically with our ears so we get less variation. And since gravity dictates that he live in a primarily horizontal world, our ears evolved with a horizontal separation making us much more sensitive to horizontal spatial information than they are to vertical. Hence another reason why we don't want a narrow response window in the horizontal plane.
The End.
barefoot