Linear aggregation revisited: Rods, rings and worms