We have performed an experiment in writing an aspect language for replication. This experiment has shown that although at rst sight replication could easily be described as an aspect, a number of problems crop up making full separation of concers hard, if not impossible.