Abstract
Previous criteria of narrative coherence have failed to come to terms with narrative intelligibility. The principle of chronology is only a negative criterion. The one entity-one story criterion, which requires every episode to refer to one and the same entity, fails both in its positive and negative forms. The Aristotelian concept of necessary connection is useless for historians because there are no natural beginnings or endings in history. Yet genetic relationships in narrative, though they cannot be reduced to causal or probabilistic relationships, do give coherences. History becomes science when historians transform stories into histories by seeking the mechanisms which underlie the genetic relationships between historical incidents. The construction of history in narrative form does not, therefore, cut the historian off from science, as positivists contend.