>Why not use experimental cmavo clearly related to the non-experimental ones. I could not even imagine trying to do this.