If all entities were completely identical, they could not be distinguished from each other so there would be only one entity. If all things were completely unique, they could not be identified so there would be no knowledge. Since neither of these extremes is the case, we conclude that entities contain sufficient similarity to be classified and sufficient differences to be identified.
Classification calls out similarities among entities and groups them together. Identification calls out differences among entities and separates them from each other. Entities in some classes will have much in common and differences will be few or low-level. Entities in other classes will have some properties in common but differences will be many or high-level. Call the former kind of class homogenous classes and call the latter kind of class heterogeneous classes. Broadly speaking, histories are about heterogeneous classes and sciences are about homogeneous classes.
The greater the heterogeneity of a class, the greater the difficulty is in making a prediction about members of the class that are yet to be identified. Predictability in this case comes down to the question of whether a class is homogeneous or heterogeneous. It may be doubted whether there are any absolutely heterogeneous classes because in order to be a class, there must be some commonality. However, one could also say that there are no absolutely homogeneous classes because in order for the class to have identifiable members, there must be some heterogeneity.
Lawlike propositions concerning homogeneous classes that make successful predictions about members of the class not yet identified constitute scientific theories. There are two basic kinds of such theories, depending on whether or not the predictions are about individual members of a class or about aggregate properties of members of a class. The former are natural sciences and the latter are statistical sciences.
“Storylike” propositions are storylines or the like that connect members of heterogeneous classes. These differ from lawlike propositions in that they focus on differences rather than similarities. The criterion for a successful storylike proposition is meaningfulness. They connect diverse entities in a meaningful way. The more meaningful a proposition is, the more successful it is. One way to determine meaningfulness is via how many and how varied are the entities that are connected. The greatest storylike proposition would connect the most diverse entities; indeed it would connect the whole universe in one storyline.
There are two kinds of histories depending on whether individual members of classes are connected or whether aggregate groups of members are connected. The former are individual histories and the latter are aggregate histories. Social, political, and economic histories are types of aggregate histories and biographical histories are types of individual histories.
Are histories and sciences dichotomous or is there a way to combine them together? They are essentially different and so should not be expected to be integrated into one. However, they may be compatible so that the lawlike propositions of sciences do not interfere with the storylike propositions of histories and vice versa. In fact, it should be a criterion of success that lawlike propositions are compatible with histories and storylike propositions are compatible with sciences. However, if there is incompatibility, one should not be allowed to dominate the other, as has happened with sciences dominating histories. This puts too much of a premium on similarity at the expense of diversity.