Following on from my earlier email... This may be unsurprising, but I also notice that when my program aborts and returns me to hugs, re-running the program with a smaller problem size (one that USED to work OK) sometimes also aborts. Just now the same thing happened for ALL problem sizes - right down to drwaing just half a dozen lines. When I quite Hugs and then re-start it problems the 'problem' seems to go away. In other words, I'm seeing non-deterministic behaviour! Tony Field. Imperial College