Gary,
I believe that what you are describing is known as a composite pattern. We have them all of the time. Usually with an -A-, -B-, -C- requirement to the holes, and then to -A- by itself to a tighter positional tolerance than the full datum structure. What the engineer actually wants is the hole pattern to be perpendicular with the primary datum, then the pattern to meet the tolerance in a best fit type situation.

Picture a square plate with a hole pattern to the main surface (-A-) and then two sides as -B-, &-C-. Now picture a gauge to evaluate this pattern. Two pads would align the two sides for the secondary and tertiary datum's, pins would go through the holes for a go-gauge. The composite requirement would not have the -B- & -C- pads and the part could rotate to pin up the holes.

The mathematical calculations are not very easy. In our old LK WinCMES programs we would perform a 3sigma calculation to eliminate real bad holes. Average the -X- deviation, and the -Y- deviations, translate the origin and then calculate the angle of rotation. Hopefully in the near future the soft gauging in DMIS will help with this.

Hope that this helps.

Ray Admire
Lead Quality Engineer, CMM Programming
Lockheed Martin Missiles and Fire Control - Dallas
972-603-2074