Earlier in the week Geoff and I discussed the need for a "boolean_not_and_operator" predicate. This clearly shows the case that needs it. The boolMM3_internal1 patterns needs it for correctness and the boolMM splitters should have it for clarity. David