ABSTRACT: Twice in the Poetics, Aristotle contrasts poetry with history. Whatever its didactic value, the contrast has not seemed to readers of special philosophical interest. The aim of this paper is to show that this contrast is philosophically significant not just for our understanding of tragedy but also for the light it sheds on Aristotle’s overall methodology. I shall show how he uses the method sketched in the Topics to define tragedy and explain why the same method will not define history. In particular, tragedy admits of definition because its parts constitute a unity, and much of the Poetics aims to show how, despite being defined through six distinct parts, tragedy can be one. In contrast, history, though a proper preliminary to poetics and concerned also with human action, does not admit of scientific treatment because it contains no essential unities. Aristotle’s understanding of ‘science’ is used here to explain why any attempt to create a scientific history would turn history into poetry.

Twice in the Poetics Aristotle contrasts poetry with history. Whatever its didactic value, the contrast has not seemed to readers of special philosophical interest. The aim of this paper is to show that this contrast is philosophically significant not just for our understanding of tragedy but also for the light it sheds on Aristotle's overall methodology. It is often noticed that Aristotle's own philosophical works do not generally follow the logically demonstrative method that he sketches in his Analytics. My concern here is to explore one methodological alternative that he does use. I shall show how he uses the method sketched in the Topics to define a tragedy and explain why the same method will not define a history.


Aristotle claims that the art of dialectic sketched in the Topics contributes to philosophical knowledge because it can be used to find indemonstrable first principles from common opinions: "for, being capable of examining, dialectic has a path to the principles of all disciplines" (B œŸ›) (I.2.101b3-4). Scientific knowledge of a subject consists of grasping its principles and demonstrating its essential attributes from them. How does one come to know the first principles? Obviously, they cannot be demonstrated from prior principles; they are first principles. As such, they are somehow determined by dialectic. Thus, dialectic transforms what we can call, for lack of a better term, a "subject matter" into a science. What is the state of this subject matter before dialectic discovers its principles? It is clear from our Topics text that this examination will look for common opinions, and it is well recognized that Aristotle's actual inquiries often begin from common opinions.(1) So the pre-scientific subject matter must contain common opinions about its facts. Aristotle has a name for such a setting out of facts: in the Prior Analytics, he speaks of deriving the principles of each field from experience and he refers to the account of the phenomena of a field as a "history" (˜) (46a17-27). Evidently, "history" precedes "science," and transition is effected by dialectic. Aristotle has much to say about how knowledge is derived from sensation and experience, but he never explains how (or whether!) his many remarks fit together into a single process. I suggest the following: individual universals come directly from sensation and experience (An. Po. II.19); they are combined by experience and common opinions about them formed; these common opinions are examined by the method of the Topics to determine whether their components are always conjoined; and definitions are finally determined by the method of An. Po. II.13. The middle stage, the formation of common opinions, is the "history." We can see what it can include from the History of Animals. This is a preliminary collection of facts that is to be followed by a treatment of the same subject through causes (H.A. I.6.491a7-14), that is, by the properly scientific treatment in the Parts of Animals (see I.1, e.g. 639b5-10). Along these lines, think of the first books of the Physics and the De Anima and the first two books of the Politics: there is no definition until the second or third books. The first books cannot quite count as science. Each is rather a preliminary sorting-through of common opinions and aporiai—a "history," though quite different from the History of Animals. In general, then, history is a preliminary to science.

What, though, about the discipline that is not the "history of X" but just history, the discipline that the Poetics contrasts with poetry? Does it share anything in common with other histories, aside from its name? Is it, perhaps, a preliminary to a science of history? Aristotle never mentions a science of history; rather, he claims that history concerns the particular rather than the universal (9.1451b6-7), a sure indication that it is not a science, for science is of the universal. History would seem to be well-named: it is indeed an ˜ and unlike other histories it is never transformed into a science. Aristotle sees the discipline of history as a chronicle of human events that are intrinsically particular and, consequently, never the subject of scientific knowledge.(2)

With the evident failure of recent attempts at formulating historical laws, most contemporary readers are not likely to feel uncomfortable with Aristotle's understanding of history. But it should have been problematic for him, for in Metaphysics A he speaks of successive discoveries of first causes as a necessary development and suggests that it will be followed by decline and later rediscoveries along the same lines. That is, when Aristotle does say something that resembles history he advances a necessary and universal cycle. And how, we might ask, could there be particulars that do not admit of universals? So the particularity of history remains puzzling.(3)

If history is peculiar in not being a preliminary for some science, then poetry is peculiar in not having a preliminary treatment of common opinions. The Poetics begins by listing the topics that need to be treated in this "discipline" (œŸ›— 1447a8-14), using the same term that the Topics had used for the result of dialectic (I.2.101b3-4). There is, to be sure, what is typically called a "history" of poetry in chapters 4-5, Aristotle offers this account to distinguish difference of "manner" and it bears no resemblance to the setting out of common opinion that we typically find in the first books of his work.(4) Apparently, poetics is already a discipline without a prior dialectic. How can it forgo "history"?


The Poetics begins by identifying tragedy and other arts as imitations (1447a13-16) and by distinguishing matter, object, and manner of imitation (a16-18). Imitation is the genus and these latter are ways of differentiating it. Aristotle proceeds in the first three chapters to characterize the distinctive matter ("in which"), object, and manner of tragedy by contrasting them with those of other arts. These part of the definition are the principles that dialectic should aim to uncover. Though they have seemingly been uncovered without dialectic, we can still inquire whether there is any hint of the techniques sketched in the Topics.

We need to read between the lines here to seek what went into the construction of what is presented to us as a finished discipline. But there are hints in the text. Aristotle simply announces that imitation is the genus under which tragedy and other arts fall (1447a13-16). Yet, if we consider the text that follows, we see that an alternative genus is, in effect, proposed and that Aristotle defends imitation. Shortly after introducing the genus, he remarks that there is no name common to poetic works that use language alone with or without meter, nor would there be a name common to mimes of Sophron and Xenarchus and Socratic dialogues even if they were all written in a single type of meter (1147a28-b13). However, some people, "attaching the term `poet' to the meter, speak of an `elegiac poet' or an `epic poet,' not terming them poets in respect of imitation but by what is common in respect of the meter" (1147b13-16). The consequence of their thinking is that someone who writes a work of medicine or science in meter counts as a poet; but, Aristotle declares, there is nothing in common between Homer and Empedocles except the meter (1447b16-20). Further, someone who mingled different meter forms would still be called a poet (1447b20-23).

Those who use "poet" to refer to a meter do not use the term to refer to imitations. They have, in effect, proposed "having meter" as an alternative to imitation as the genus of poetry. If having meter makes a work poetry, then Empedocles' physics counts along with the Iliad as poetry because both have meter and a Socratic dialogue would also count as a poetry if it had meter, whereas mimes could not be poetry because they lack meter.

It need scarcely be said that, though we do restrict "poetry" to what has meter, no Greek would accept such a restriction for poesis, any more than we would so restrict the application of the phrase "fine arts." Even so we may well wonder whether Aristotle addresses the objection to imitation as the genus adequately by simply insisting that Homer and Empedocles have nothing in common and that mimes and Socratic dialogues would not have the same name—read, belong to the same genus—even if they were both in meter. And what could be wrong with mixing meter if poetry is characterized by meter?

His objections are, however, readily intelligible if we look to the Topics. The first topic he mentions there under "genus" is to consider whether the things that fall under the genus do not also fall under what is proposed (IV.1.120b15-20); in our case, mimes, recognized as poetic works, do not fall under writings with meter. Similarly, Aristotle's claim that Empedocles is more a physicist than a poet (1447b18-20) amounts to objecting to the proposed genus on the ground that it would include something that does not belong in the genus, a example of second topic of the Topics' "genus" section (120b21-29). Works that combine meters would seem unable to belong to any of the species of a genus of poetry defined by its meter, an instance of still another topic (121a27-39). That is, if the genus of poetry were characterized by meter, then its proper differentiae would have to be types of meter; and works that mix meter would fall under more than one species. If characterizing the genus as meter results in some of its content not falling under any species of the genus, the characterization must be wrong. I could go on: often a single argument can be seen as an instance of more than one topic. But the point is clear: Aristotle relies upon distinctions and topics like those he expounds in the Topics to reject implicitly the meter as the genus of poetry.

Since meter is not the genus, it is reasonable to take it as a differentiating poetry into a species. Some poetry is in meter, other in written or spoken language; still other in song or dance: all these media mark species of poetry. The proposal that meter is the genus of poetry amounts to a confusion of a differentia for a genus. Further, insisting that there would be no common name for mimes and dialogues even if they were written in the same meter, Aristotle is claiming, in effect, that the species "poetry that imitates in meter" cannot be further differentiated by distinguishing the various subjects treated. Distinctions of content cannot differentiate a type of meter; for they are not proper differentiae, and the same differentiae would reappear under other species of poetry.

In short, in the course of examining the various materials in which poetry could be made, Aristotle implicitly considers whether meter is not the best definition of this genus and rejects it. In the process we come to understand what does constitute the genus of poetry and what some of its differentiae are, namely, differentiae that are the matter "in which" the imitation is made.

Let us turn now to the second differentia, the object to be imitated. This Aristotle divides into good, bad, and average characters, and imitations of these types can appear in different materials. That is, the object imitated and the first differentia, that "in which" the imitation is made, combine. Since the same object can be imitated in several materials, the second differentia serves to differentiate several coordinate species of poetry. This is contrary to Metaphysics Z.12's doctrine of proper differentiation: each differentia belongs in one species or, anyway, in one line of differentiation and, therefore, implicitly includes the higher species that it differentiates. Obviously, this specification is impossible if the same differentia falls under different coordinate species.(5) But such differentiation is understandable in light of Metaphysics H.2's recommendation to define artifacts and natural composites by giving both form and matter. In such cases, the differentia does not suffice because it could be present in different matters. So, too, the same object could be imitated in different matter; to define a species of poetry, it is necessary to mention both object and kind of matter.

The third differentia, the manner (narrative or dramatic), must also be added to distinguish some species of poetry, in particular, tragedy and epic. Though the last and narrowest differentia, manner is not the most characteristic and important, as we might have expected, but relatively unimportant; for Aristotle declares that the poetic effect can be achieved with or without a performance (6.1450b15-20).


These three differentiae form the basis for the definition of tragedy that Aristotle offers in chapter 6. The definition is:

Tragedy is the imitation of serious and complete action having magnitude in pleasing speech, with each of the kinds of speech used separately in its parts, dramatized rather than narrated, and accomplishing through pity and fear the purgation of these emotions. (1449b24-28)

Aristotle goes on to explain that "pleasing speech" refers to rhythm and song and that "the kinds of speech [being] used separately" refers to some parts having only meter and others also song (1449b28-31; cf. b33-34). In other words, the "in which" is divided into the song and diction—the choral and spoken parts of the tragedy. Further, the "manner" is now termed the spectacle. The "object imitated" is most properly a noble action and the imitation of it is the plot; the action stems from the character and thoughts of the agents. Thus, the initial three differentiae are refined into six parts of the tragedy: spectacle, song, diction, thought, character, and plot.

What is particularly puzzling about these six parts and the complex definition of tragedy is their multiplicity. In the Topics Aristotle declares that "for each being the essence (h œϤ˜  jœ e) is one" (VI.4.141a35). He uses this principle to exclude the possibility of multiple definitions of the same thing (141a35-b1), and it is the basis for a lengthy presentation of topics pertinent to definitions that contain a plurality of parts (VI.13-14).

It is just this sort of problem that Aristotle addresses in Metaphysics H.3. Recall that in the preceding chapter he had recommended defining certain entities, artifacts and natural composites, by both form and matter; for example, a house by its function of providing shelter and also by its being boards and bricks. In H.3 he claims that we cannot think of these (the boards and brick and their position) as if they were coordinate components (1043b4-6). If they were, there would have to be something besides them to unify them—and this most important constituent would have been left out of the definition (1043b11-13). It is rather the case that one component is form and the other matter (1043b30-32). The point is that the form is not like the material constituents, but something of a different sort—the principle that unifies the material parts (cf. Z.17). An artifact is, thus, to be defined by its material constituents and the organization or function that serves as their principle of unity.

When we bring this lesson from the Topics and Metaphysics back to the text of Poetics 6, Aristotle's remarks about the definition suddenly make sense. First, his claim that "plot is principle and soul of tragedy" (1450a38-39) is not the throw-away comment it seemed: soul is the form of body and plot is the principle of tragedy in the sense that it is the formal cause. As the form, plot can provide unity to the other five parts, and Aristotle seems intent on showing that that is the case. Thus, the arguments to show that plot ("the structure"—˜ ) is more important than character (1450a15-38) have the effect of assigning the job of unification to plot rather than to character. Both character and thought are justifications for the actions of the protagonists; the tragedian wants characters who are either habituated to do what he wishes them to do in the tragedy or who can be led to do certain actions through reasons. In either case, it is the action to be performed that is primary and the character or reasoning are formulated to justify the action. Hence, plot unifies thought and character. When Aristotle discusses these parts later (chs. 14, 19), he emphasizes in more detail how they contribute to plot.

There is significantly less discussion of the material elements of tragedy in the Poetics, but at every turn Aristotle subordinates them to plot: Plots should be summarized first and only then brought to completion through speech (ch. 17), and elaborate speech should not obscure character or thought (24.1460b2-5). Likewise, the chorus (song) should be part of the whole and contribute to the performance (18.1456a25-27). Spectacle is artless, belonging partly to the designer; it should serve plot (6.1450b16-20).

In sum, plot is the formal principle of unity of the formal parts, thought and character; and the material parts are appropriately used when they contribute to the plot. Plot itself ought to be a unity (ch. 7). This is best done not by making plot simple but by organizing many events. In brief, the plot is one if it imitates a single action, and many events count as a single action when they share a single end. Determining what counts as a single action is a central problem of the Poetics. For present purposes, the details of this discussion need not concern us; what is important for us to realize is the significance unity of the plot has for Aristotle.

Approaching the Poetics with the criterion of the Topics in mind enables us to see that the character that would allow tragedy to be defined, the unity of its parts, is a major theme of the work. What might have seemed to be scattered remarks can now be understood to contribute toward the task of showing that tragedy admits of definition and, consequently, of being known.

Claims about the unity of tragedy's parts function normatively as well as descriptively. Aristotle aims, after all, not only to describe good tragedy but also to explain what makes it be so. In a word, the more a tragedy is one, the better it is as a tragedy. It is not any conjunction of plot, character, etc. that makes a tragedy; the tragedy exists only if the plot functions to unify the other parts. Only then is the tragic effect achieved. Aristotle satisfies the criterion set out in the Topics with a move that parallels that of Metaphysics H.3. The dialectic of the Topics is not an explicit concern here, but it is presupposed at every juncture.


Let us now return to the points of contrast between poetry and history. Recall that there are two. First, poetry represents universals and history particulars, a difference that makes poetry more philosophical than history (9.1451a38-b7). Second, whereas poems (specifically, tragedies and epics) imitate a single, complete and whole action, a history concerns all the events in a single time that happened to one or more people and that relate to each other at random and have no single end (23.1459a17-30). Since a universal is a kind of unity (see Met. ƒ.26.1023b29-32), both contrasts are ways of saying that poems are unities and histories are not or are much less so. Aristotle compares the tragedy or epic to an animal (23.1459a20; also 7.1450b34-1451a6); he means to say that its parts have an organic unity in contrast with the events of history.

Whereas tragedy can be defined because it is made one by plot, history has no such unity. It is intrinsically undefinable, intrinsically particular, intrinsically unphilosophical. Yet both passages contrasting poetry and history are followed immediately by injunctions to tragedians to draw upon history as their source material (9.1451b15-32; 23.1459a30-b7). Both tragedy and history concern human actions. Tragedy draws upon history.

The historian aims to gather all the particulars. Were he to organize them into necessary or probable causal sequences, as the poet does, his subject would universal and definable; but he would no longer be doing history. That is to say, history stands to poetry just like the ˜ of physics (Physics ˆ) stands to the rest of the Physics, just like the ˜ of the soul (De Anima I) stands to the rest of the De Anima. In each case, a history is the essential preliminary that occurs before there is a definition. But only in the case of history proper does attaining the definition transform the inquiry into a completely different discipline.

We can appreciate how strange and troubling such a connection is by recalling Aristotle's claims that poetry originates from a natural tendency to imitate (4.1448b4-9) and that we find this pleasurable partly because of what we learn from it (b9-17). A representation draws attention to salient details. That is the reason that it is easier to learn anatomical details by first studying diagrams: under a high powered microscope or in nature the salient details are present with too many other details to grasp them clearly. So, too, the tragedy or the epic does not present all the actions or characters that occurred in a time period—that would be history—but only those pertinent to a single sequence of events. From it we learn the event's causes. Poetry is, in effect, a science of human action, and history the material from which it is derived.

Trouble is, poetry is fictional. Why can there not be a scientific history that studies the true universal causes of human actions? Aristotle's answer has been clear all along: there are no universal in human actions; tragedy creates universals and causes. But how could a discipline that imitates human action, tragedy, have more universality, more unity, than what it imitates?

There is no way to resolve these puzzles entirely, but three points, all observations about science and definitions go some way toward making Aristotle's thought intelligible. First, "history" as Aristotle understands it is not simply the setting out of facts; it also involves making connections and drawing conclusions—we need only look at a "history" that precedes an Aristotelian science to appreciate what it can achieve. Second, Aristotelian science requires an essence. The point of showing that the tragedy is one and, thereby, definable is to show that it has an essence. There is an essence of human action, but it is treated in Ethics and Politics. These consider the generation and destruction of definable conditions, happiness and the state; but such developments follow a regular path that depends upon the condition realized. The details of history are simply accidental to such processes. Finally, concerned with essences, Aristotelian sciences are atemporal. Poetics contains temporal action, but this action imitates the atemporal because it comes to a completion because it contains an end. It imitates not human action but, as we said, good character or, rather, flawed good character. Poetic action is complete when such character gets what it deserves. History, alas, does not work that way.

To conclude, the definition of tragedy advanced in the Poetics presupposes a dialectic but unlike other Aristotelian sciences it does not emerge from a history. History proper, the preliminary to poetics, does not admit of scientific treatment because it contains no essential unities. Poetics, on the other hand, is not seeing unity in human events but creating a unity that is not there. The method at work in the opening of the Poetics is dialectic.


(1) The case of akrasia is frequently cited—N.E. VII.1.1145b2-7.

(2) In contrast, Stephen Halliwell writes, "It is not that life or history cannot furnish coherent structures of action, but just that they do not regularly do so, according the Ar. (though most historical writing is based upon an implicit rejection of this premise)," The Poetics of Aristotle: Translation and Commentary (Chapel Hill: University of North Carolina Press, 1987), p. 105. He may be thinking of 1251b29-33, for there Aristotle suggests that historical events can have happened according to probability. But Aristotle also insists that even if someone represents these probable events he is still a poet; that is, the probability of the events themselves does not diminish the role of the poet in representing them.

(3) G. E. M. de Ste. Croix, "Aristotle on History and Poetry," in Essays on Aristotle's Poetics, ed. Amelie Oksenberg Rorty (Princeton: Princeton University Press, 1992), pp. 28-29, argues that Aristotle's dismissal of history is, "even on [his] own premises, . . . not fully justified." This conclusion is based upon Aristotle's claiming that science is of the universal or what is for the most part and Thucydides' concern with the constancy of human nature and the resulting patterns of human behavior that tend to recur. Ste. Croix thinks that Aristotle should have recognized Thucydides' works (with which he was familiar) as scientific.

(4) The development of tragedy and comedy described in Poetics 4-5 is no more a history than Aristotle's account of the development of a state or a dog. Like the state, imitation arises from nature (4.1448b4-24; cf. Pol. I.2.1252a24- 31), and it had only to reach its pinnacle in tragedy (4.1449a14-15; 5.1449b17-20; cf. Pol. 1252b27-31). The progression of development is a function of the form that ultimately comes to be realized through it. Stephen Halliwell, Aristotle's Poetics (Chapel Hill: The University of North Carolina Press), pp. 93-95, arrives at the same conclusion.

(5) In the Topics Aristotle repeats the conclusion of Z.12 that each differentia could only fall under a single genus, but then quickly reneges and acknowledges that one differentia can fall under distinct genera (VI.6.144b12-30). His example is that "biped" differentiates both "walking-animal" and "flying-animal." It is hard to fathom why Aristotle would state both views here; and in the Parts of Animals (I.5) Aristotle makes a point of denying that the feet of human and of birds are biped in the same way.