For the events "e" and "f" to be identical, They must have the same subject and spatio-Temporal location, And their (participial) property-Descriptions must belong to the same "modification set" (e.G., Reddening, Reddening slowly, Reddening in july). The same criterion applies to actions, Which are here treated strictly as a proper subclass of events (john's closing the door = the door's being closed by john = the door's becoming closed). Actions related by goldman's "causal generation" are therefore distinct, But those related (...) by his other three types of act-Generation are not. This conclusion requires abandonment of the view--Questionable on other grounds--That causal contexts are thoroughly extensional. (shrink)