Thanks Michael, that's exactly what I was after! I'm still thinking in terms of poly-modelling too much, that's clear. NURBS is quite different in approach, but that method works just fine.

Many thanks again.

Steve