It seems rolly is a more temporary state though, such as the rolly non-poly depicted in the original post, which could later be non-rolly. Perhaps we need a measure of average roliness, or typical percentage of time spent in a rolly state (under normal circumstances in a typical habitat)