bobbygeeWOW, everything that you said in your first two paragraphs is consonant with what I'm trying to say here, including the part about adapting to real-world circumstances. (I feel that no apology is ever necessary for doing that, by the way; it's a positive accomplishment when it works.)
To me, the graph which you copied from DPA's Web site looks very much as if it was derived from Michael Williams' work, and his book "Microphone Arrays for Stereo and Multichannel Sound Recording" is my primary point of reference, so we're definitely talking the same language. Another good source is the Web site
http://www.hauptmikrofon.de/ (which is in English despite its German domain).
poorlyconditioned, if what you're asking about is the single-digit numbers in the circles along the curved lines in the DPA graph--those are indicators of another aspect of a mike setup which can influence one's choice of angle, directional pattern and possibly spacing: the maximum value of "angular distortion" which will occur for that particular setup.
"Angular distortion" is the discrepancy, in degrees, between where a sound was actually coming from when the recording was made versus where it will seem to be coming from when the recording is played back over loudspeakers in a hypothetical "typical" listening configuration (normally an equilateral triangle with the speakers and the listener in its corners).
"Maximum" angular distortion simply means "the worst-case value." All other things being equal (as they say), it's nice when this number is lower rather than higher. I'm being vague on purpose because some listeners are far more sensitive to this than others, and vary greatly in the esthetic value (if any) which they place on it.
I know some musicians--string players and singers especially--who listen almost exclusively for tone (timbre); as long as there's any sense of spatiality in the recording at all, they're satisfied with the stereo aspect, and they truly don't care whether particular sound sources can be localized or not. To some people localization is a meaningless parlor trick and to others it's a bedrock virtue. Go figure.
By the way, I hope it's clear that any one graph like the one in bobbygeeWOW's message above can be valid only for one directional pattern--in this case, cardioid. The same angles and distances between a pair of wide cardioid or supercardioid or figure-8 microphones would produce very different sets of relationships and very different graphs as a result. If you want the lowest angular distortion, consider Blumlein stereo recording but be sure to watch out for that setup's narrow SRA of only ±45°!
--best regards