Maybe someone knows some interesting scientific reason, but the reason that comes to my mind is that a right angle is easy to eyeball or measure with a book or any other makeshift square object. Therefore the pattern is easy to reproduce. All of these close mic patterns seem to have been worked out heuristically. If you measure when you set up, you will get more consistent, predictable results.

[url=(dead link removed)[/url] is a nice summary of Stereo Microphone Techniques.
I don't know of anything magic about 90 degrees.

Usually, I point each microphone in the coincident pair at something I am interested in the sound of. For example the cymbals on the far sides of the drum kit, or the bridge and sound hole of a guitar. I have even used them to cancel out a section of a choir that was too loud. One phase inverted and pointing at the loud sopranos, and the other pointing at the weak bass section. (I also had an AB pair for that recording for the stereo image)

Except with a 180 degree angle, what is directly in front of both microphones is always louder than what is on the sides with a coincident pair of microphones. As the angle increases from 0 to 180 degrees the amount that the center portion is louder decreases. The size of the stereo image widens with the angle. So you go from a center hyped mono, to a even response wide stereo image.

My guess is that 90 degrees is a compromise that creates a stereo image that resembles what you would hear if you were there.

It's to avoid phase cancellation.

The 90 degrees comes from the original Blumlein configuration for ribbon mics, which have a figure-8 pattern. If you sum two figure-8s that are 90 degrees apart, you get a perfect circle, giving equal all-round (omnidirectional) coverage with 2-channel output.

Use any other angle and the result is not omni. This, of course, can be exploited to good effect, but it's not Blumlein. The term "X-Y" is usually reserved for the 90 degree configuration, whether it's of cardioids, hypercardioids or some other pattern.

The Blumlein configuration does not have to be X-Y, that is, with each microphone at 45 degrees to the centre of the sound source. I often use a Blumlein configuration where one Fig-8 microphone points directly at the centre of the sound source and the other is at 90 degrees to it. I call this configuration "MS-Blumlein", as I use M-S decoding to extract stereo information.

Nate-Redline wrote: It's to avoid phase cancellation.

Just to clear things up: Being coincidental will avoid phase cancellation. The 90 degree angle between the two microphones has nothing to do with phase.

Thanks Boswell, that makes sense.

The 90° configuration comes from the cardioid polar pattern that symmetrically overlap center sound images while providing a full width hemispherical like 180° coverage. It would be really great to create a stereo microphone from a single capsule. But that hasn't happened yet. And there are no Cyclops' nor unicorns.

I always wondered why my Neumann KM 86's were never released as some kind of stereo capable microphone? But Audio Technica only recently figured out how to rip off that form factor/design package for a "side address" looking stereo XY microphone.

XY just makes it simple for lousy recording engineers to not screwup too badly. Some of us just make recordings that are MS but sound great anyhow. But I think every good woman needs an XY once in a while regardless of orientation? How else do you know what's better? A man or a vibrator?

I'm not saying
Ms. Remy Ann David

Boswell wrote: The 90 degrees comes from the original Blumlein configuration for ribbon mics, which have a figure-8 pattern. If you sum two figure-8s that are 90 degrees apart, you get a perfect circle, giving equal all-round (omnidirectional) coverage with 2-channel output.

I know this is a nit, but if you simply sum the voltages of two figure-8 mics at 90 degrees, the output will NOT sum up to an omni. However, the power sum of them WILL produce an omni output. Each figure-8 traces out a pattern that is a sinusoid of the angle around it.

X - cos (angle)
Y - sin (angle)

take a signal at 45 degrees from each mic (which, in a Blumlein would be a signal at center stage). The voltages are:

X = cos (45) = .7071
Y = sin (45) = .7071

X + Y = 1.414 / an omni would = 1

However, if you sum the powers:

X^2 = cos (45) ^ 2 = .5
Y^2 = sin (45) ^2 = .5

X^2 + Y^2 = 1 / by definition, sin^2 + cos^2 = 1 (a circle, or omni)

Pick any other angle, 0-360 and the relationship holds. It's POWER that makes the Blumlein what it is.