Paul,

I think that my earlier statement is correct. Getting cluster size reliably is problematical at best. Check the thread#341872. Read Ed Rauh's comments on the subject specifically. He goes into great detail on this subject.