OK, I figured out the flaw in my reasoning. There are in fact 2 equally good parities for tetrahedra: "innies" and "outies", or (a) facing any face, having traversed the first face counterclockwise, you go back (behind the plane of that face, or away from you) to the 4th vertex, or (b) having traversed the first face counterclockwise, you come forward (in front of the plane of the face, or closer to you) for the 4th vertex.
Either formulation is fine, AS LONG AS YOU STICK TO IT throughout the whole connections list. Mixing parity is not permitted (well, it is sort of, in that the file is importable and renderable, but the second parity encountered will be inside out as far as the renderer and normals calculations are concerned). Thus, the mystery of the flexible tets is solved: DX reads the first tet's parity in the connection list and locks that one down as "good". If any after that are the opposite parity, they are by definition "bad". So, if some bozo reorders the list such that the opposite parity now appears first, all formerly "bad" are now redeemed, and vice versa. Another chapter in my book "DX for Dummies (by Dummies)" written!! Whoo-hoo! Chris Pelkie Vice President/Scientific Visualization Producer Conceptual Reality Presentations, Inc. 30 West Meadow Drive Ithaca, NY 14850 [EMAIL PROTECTED]