In the long run the chain continued to form an enclosure for the catalyst, while its links kept disintegrating and being replaced. In this way the membrane-like chain became the boundary of a network of transformations while at the same time participating in that network of processes. In other words, an autopoietic net work was simulated.Whether or not a sequence of this simulation will generate an autopoietic pattern depends crucially on the disintegration proba bility-that is, on how often links will disintegrate. Since the deli cate balance of disintegration and "repair" is based on random motion of substrate elements through the membrane, random pro duction of new links, and random motion of those new links to the repair site, the membrane will remain stable only if all those processes are likely to be completed before further disintegrations occur. The authors showed that with very small disintegration probabilities viable autopoietic patterns can indeed be achieved.l 3