The relation between description and laws in explanation has been a traditional topic in the analytical philosophy of history. Raymond Martin proposed a new approach to this problem: analyze how historians try to show that their explanation is better than competing explanations. The goal of this article is to develop Martin's account by introducing the concept of colligation to provide a better understanding of the role of description than Martin's account. According to Martin, when historians try to show that one explanation is better than others, there are two relevant factors: the justification and sufficiency of explanation. To change these factors, historians use four kinds of arguments: (1) increasing or (2) decreasing the likelihood of a particular explanans, a sentence to explain other sentences, (3) increasing the likelihood that a particular explanans is partially sufficient, and (4) decreasing the likelihood that a particular explanans is sufficient. In Martin's account, the arguments of kinds (3) and (4) deploy the strategies regarding lawful connections. To complement this account, I argue that historians also deploy a particular kind of description, colligation, in the arguments of kinds (3) and (4). Colligation unifies discrete lower-order descriptions into a single higher-order description whose criteria of justification is different from likelihood. I suggest that colligation plays a crucial role in deciding which law-like generalizations are relevant to explanation, which is why description can play a role in the arguments of kinds (3) and (4). I will demonstrate these claims through the case study of the controversy over the relationship between the Enlightenment and the French Revolution.