I push the phifunctions further, to the point where we are creating a tree ordinal whose ordinal height is equal to the BachmannHoward ordinal. Along the way I mention why we don't fall into some fixedpoint traps that you might think we would.