Non-repeatable observations may contribute to an experiment but they are not the experiment themselves. Whilst they are part of the collected data and a source of information, main or not, such observations are not the experiment. Surely the whole point of a successful experiment is that it must be repeatable.