Introduction
The Posterior Analytics (APo) contains a treatment of the nature of scientific thought and of science (epistm) itself. In my view it implies
a method for presenting science, a form in which a body of scientific
knowledge is organized and ideally should be set out. The APo is not
primarily about obtaining scientific facts, although it has important
implications about the nature of scientific facts, the kinds of scientific
facts, and the relations among scientific facts, and so it has consequences for the kinds of things that are investigated in scientific research. But
its main purpose is to discuss achieved sciences, not how to achieve
them.
The APo model for presenting sciences is founded on a distinction
between principles (primary facts/propositions) and provables (propositions/facts that depend on the principles1). There are strict conditions on principles. (They must be true, primary, immediate, better
known than, prior to, and explanatory grounds for the conclusions
derived from them.2) Each science treats a distinct subject matter or
genus (genos). There are three different kinds of principles: definitions
APo I 2, 71b20-2
76
Richard McKirahan
and existence-claims, which are proper to a single genus, and common principles that apply to a plurality of genera). In addition, Aristotle places strict conditions on acceptable procedures for deriving
provables from principles: they must be derived by means of syllogistic
arguments based on relevant principles. Such proofs are called demonstrations (apodeixeis).
A traditional problem is how to account for the fact that none of
Aristotles scientific works follows the APo model. This is clearly true.
Even if some parts of some works show a tendency towards APo organization,3 it remains that no entire work of Aristotles is in the form
demanded by the APo for an achieved science. Here are three possible
explanations. First, Aristotle wrote the APo late and didnt go back and
redo his earlier scientific works according to its specifications. But the
APo is regarded as an early work, and I know of no one who has proposed this explanation, nor will I propose it. Second, Aristotle wrote the
APo early and subsequently in doing scientific work he abandoned the
straitjacket imposed by the APo model. Third, Aristotle wrote the APo
early and did not abandon it; some of the APos leading ideas influence
his scientific work, even though he did not arrange his scientific works
in the form described there. This paper is intended to provide support
for this third interpretation.
People who adopt the third interpretation need to face the question
of why Aristotle didnt go all the way and arrange the works in the way
described in the APo. Here are some ingredients of a possible explanation. First, the important scientific work is done in reaching the principles; the remaining task of constructing demonstrations based on the
principles is trivial. To reach the principles we first need to ascertain the
propositions that constitute the science: the body of necessary propositions involving subjects and attributes in the subject genus of the science
in question. This is the research phase. Then comes the organizational
phase4 in which we determine which of the facts are principles and then
construct demonstrations. In terms of the distinction Aristotle makes
between the that (to hoti) and the why (to dihoti),5 we first ascertain
I do not mean to suggest that we must complete all research relevant to the science
before beginning work on organizing the body of knowledge into a deductive system.
that something is the case and then ascertain why it is the case, which
is done by finding a demonstration of the fact in question.
But there is reason to think that the work of constructing demonstrations that begin from already known principles is entirely mechanical
and trivial. Let me explain. Syllogistic arguments are very restricted
in form. Each syllogism consists of two premises and one conclusion.
Premises and conclusion are in subject-predicate form, and each one of
them must have one of the following forms: A necessarily belongs to
all B or A necessarily belongs to no B. The only syllogistic moods acceptable in demonstrations are Barbara and Celarent.6 It follows that the
process of constructing proofs is trivial once we have the principles.7 If
the principles are finite in number, there can be only a finite number of
demonstrations. In view of this, the scientific work is done in ascertaining the principles; the rest is gift wrapping.
Further, Aristotle has philosophical reasons that would lead him
to believe that this situation obtains. For example, he believes that all
propositions in syllogisms predicate an attribute of a subject,8 that all
deductions can be reduced to syllogisms,9 and that proofs are only finitely long.10 He believes that some of a subjects attributes hold of ne-
This is Aristotles basic model, as discussed in APo I 14. It agrees with his conviction that science deals with universals, not particulars, and per se relations, not
accidental ones. It is so restrictive a model that it does not even account for the
most elementary proofs in geometry; and this raises important problems for the
theory of demonstration that go beyond the scope of this paper. See McKirahan
1992: Chapter 12 for discussion of some of them.
We first look for pairs of principles in which the subject term of one is the predicate
term of the other. If both principles are affirmative, then they are the premises of a
syllogism in Barbara which demonstrates its conclusion. If both are negative, then
no syllogism is possible. If one principle is positive and one negative and a syllogism can be formed (that is, if the middle term is the predicate of the affirmative
principle and the subject of the negative principle), then either we have a demonstration in Celarent or we have a syllogism that is not a demonstration (we have a
first step towards a demonstration, where there are one or more middle terms that
need to be filled in). We then follow the same procedure for the set of propositions
consisting of all the principles and all the conclusions of the preceding demonstrations.
APr I 1-4
Every demonstration and every deduction takes place through the three figures
(APr I 23, 40b20-2).
10
APo I 19-22
78
Richard McKirahan
cessity and others do not,11 and of the former he believes that some are
naturally prior to others in the sense that the truth of the others depends on the truth of them.12
My account distinguishes two phases of scientific work: gathering
relevant information, and organizing it into the form of a finished science. The preliminary work (research phase) aims to determine the
facts that fall within the scope of the science, and does so in part on the
basis of observations, as outlined in APo II 19, and in part on the basis
of endoxa the opinions of the many and the wise. The criteria here are
truth and relevancy.
The organizational phase has two parts: first, determining which
facts (perhaps more precisely, the true propositions that express those
facts) are principles, and second, forming the demonstrations, which
we have seen is trivial. The first part of the organizational phase is not
trivial. I believe that it has four parts. First, it requires determining
which facts/propositions follow syllogistically from which. This work
13
falls under the heading of dialectic. When this work has been done,
we have candidate demonstrations, but we do not yet know which candidate demonstrations are genuine demonstrations. The reason for this
is that, as Aristotle recognizes, there are situations where there are two
propositions each of which is a premise in a candidate demonstration
of the other, as in the case of the non-twinkling planets discussed below.
Since a given proposition cannot be both a principle and a provable, the
two candidates cannot both be genuine demonstrations. In other words,
logic is not enough to determine which propositions are principles.
The second part is to determine which among competing candidate
demonstrations give the true explanations of their conclusions. At least
two kinds of work are involved in making this determination. First, Aristotle specifies some features of demonstrations that restrict the field.
Most notably, he holds that a proof that isosceles triangles have angles
equal to two right angles is not a proper demonstration; the property
in question belongs primarily to triangles and in consequence belongs
11
APo I 6, 74b10-12
12
See the discussion of the cause of the fact that the planets do not twinkle at APo I
13, 78a22-b3.
13
Top I 2, 101a36-b4, APr I 30, 46a3-30. This role of dialectic is discussed in McKirahan (1992), 261-2.
14
15
16
APo I 13, 78a30-b4. The treatment of the sphericity of the moon at 78b4-13 is similar.
17
78a37-8
18
19
98b21-2
20
80
Richard McKirahan
The third task in the first part of the organizational phase of a science
is essentially one of collation. It is a matter of seeing which propositions
are premises of one or more demonstrations (and consequently explanatory grounds of the conclusions of those demonstrations) and are the
conclusions of none. These are the indemonstrable propositions, which
are found in the principles. Well, not exactly. There remains the fourth
task, in which we transform the indemonstrables into the three kinds of
principles: definitions, existence-claims, and common principles. I shall
not treat existence-claims or common principles, which do not present
great difficulties.21 But the question of definitions is not so easy, since
the method of forming definitions by collating all the unprovable premises in which the definiendum occurs conflicts with Aristotles standard
view of that definitions are in the form of genus + differentiae, where
the first differentia is a differentia of the genus and each subsequent differentia is a differentia of the immediately previous differentia.22
I am concerned here primarily with the form of definitions.23 The
problem is that there is no guarantee and in fact no reason to suppose
that definitions produced by this method of differentiation (which is
designed to identify the definiendum by stating how it is different from
other entities) I call these thin definitions will provide as much
information as is contained in all the propositions that contain the unprovable facts about the definiendum I call these fat definitions. On
this matter I will summarize an argument I have given elsewhere.24
As I said above, the prevailing model in the APo for the form of
definitions is the genus + differentiae model. APo II 13 presents two
incompatible conceptions of this kind of definition. According to one,
we begin from the genus, divide the genus by coordinate differentiae
that exhaust the genus; begin again from the differentia that includes
the definiendum and divide it by coordinate differentiae of it; proceed
down this line of division until we reach a differentia that is coextensive
with the definiendum. The definition is the proposition that predicates
21
22
See McKirahan 1992, 111-19, with references to Topics IV and Metaphysics Z 12.
23
In particular, largely because of lack of space, my discussion does not address the
treatment of definition in APo II 8-10, which I have discussed elsewhere: McKirahan (1992), Chapter 16, also Chapter 10, especially p.202 and Chapter 13, especially
pp.167-9.
24
the genus and all the relevant differentia of the definiendum (97a23b6). (E.g., an isosceles triangle is a polygon that has three sides, exactly
two of which are equal. Genus: polygon; first differentia: having three
sides; final differentia: having exactly two equal sides.) Since the definition proceeds along only one line of division, in which each differentia
entails all the preceding differentia along with the genus, the definition
boils down to identifying the definiendum by only one term.25
The other conception of definition found in APo II 1326 calls for us
to proceed by identifying attributes each of which holds more widely than the definiendum, but which when taken together do not hold
more widely than it. The definition is the proposition that predicates all
of the differentiae of the definiendum. For example, the number three
is odd, prime in the sense of not being a product of other numbers, and
prime in the non-standard sense of not being the sum of other numbers
one not counting as a number.
A problem for both accounts is that they fail to distinguish between
essential attributes and non-essential necessary attributes. There is
nothing in either account that might exclude the property of having
angles equal to two right angles from being an element in the definition
of isosceles triangle, although the APo makes abundantly clear that this
is a provable attribute of triangles, not a definitional property. In addition, the first approach involves another serious problem. In general
there is no reason to suppose that all the demonstrable facts in which
a given subject or attribute appears are consequences of the fact that a
single differentia is predicated of it.
Aristotle eventually recognized the problem. In de Partibus Animalium I 2-3 he takes up once again the doctrine of genus + differentia
definitions, and offers some devastating objections. One of his objections is relevant to the point just made. Discussing the definition of man
as a succession of differentiae along a single line of division: animal,
footed, two-footed, etc. he says: Proceeding successively it reaches the
final differentia ... . This is having toes only; or the entire combination
if man is the subject of division, for example, it should form the conjunction footed, two-footed, having toes. If man were merely something
that has toes, in that case having toes would prove to be the single
25
26
82
Richard McKirahan
27
PA I 3, 644a2-8
28
PA I 3, 643b23-4
29
30
PA I 3, 643b10-13
31
PA I 3, 643a28-31
84
Richard McKirahan
Aristotles Poetics
The work I have chosen to talk about is an outlier: the Poetics. I will
begin by justifying my decision to consider it in this context, since it is
clear that poetry is not a science as Aristotle defines the term. On the
other hand, it is an art, skill, or craft (tekhn), and the APos account of
how we reach scientific principles is also an account of how we reach
the principles of arts as well. The works title is Peri Poietiks, a prepositional phrase missing its noun. The noun to be supplied is tekhns.32
Aristotles two sustained discussions of how we attain knowledge of
scientific principles, explicitly state that his account holds for principles
of arts as well as of sciences.33 So for Aristotle, arts, like sciences, are
organized bodies of knowledge, the difference between them being that
arts have products separate from the producer, whereas sciences aim at
knowledge, not at anything external.34 Like scientists, artists can give
accounts, justifications, or explanations of what they do. The accounts
are based on knowledge of a certain kind knowledge of principles
and consist in deriving conclusions from those principles.
32
In fact, Diogenes Laertius refers to the work as Pragmateia tekhns poitiks (Lives of
the Philosophers V 24).
33
APo II 19, 100a8-9, where the difference on this score between arts and sciences
is said to be that the subject matter of the former has to do with generation (peri
genesin), while the subject matter of the latter has to do with being (peri to on), and
Metaph A 1, 980b27-2a3, where Aristotle uses an example taken from the art of
medicine to illustrate points common to both arts and sciences.
34
I aim to show that the Poetics focuses first on definition and second
on using the relevant definition as a basis for deriving further features
of poetry. Of the three kinds of principles identified in the APo, only
definitions are prominent in the Poetics.35 If we read the Poetics with this
in mind, we will easily notice that definitions are absolutely central to
the work. Once this is noticed, it may strike us as an odd way to study
36
the nature of poetry. And it is especially odd in a work whose purpose
(among other things) is to tell us how to go about writing good poetry,
an aim announced in the opening sentence of the work (1447a9-10).
After five chapters of introductory remarks about several genres of
poetry, Aristotle announces that he will focus on tragedy, and return
to epic and comedy later on (1449b21-2). In the remaining twenty-one
chapters of the surviving first book of the Poetics tragedy is given seventeen (Chapters 6-22), epic two (Chapters 23-4), and the final two are
devoted to questions that concern both genres and to a comparison of
the two. Comedy was treated in the lost second book.
35
36
A quick look through the poetry section of the local public library revealed many
books of poetry but not a single attempt to define it.
86
Richard McKirahan
37
Spoudaios (opposite of phaulos) in some contexts means serious and in others means
morally good. It is important to keep the translation consistent, and I adopt serious throughout.
38
39
40
Melody and spectacle, two of the six parts of tragedy identified at 1450b7-10, are
absent from epic.
41
This is how I translate mimsis for present purposes. I do not claim that it is the best
translation.
88
Richard McKirahan
not unique in any one of these three respects. Not only tragedy but also
comedy, dithyramb and nomes employ rhythm, words and harmony;
not only tragedy but also epic represents serious people; and not only
tragedy but also comedy is dramatic. Still, tragedy is unique in the particular combination of the alternatives. Let me explain.
Chapter 1 states that the medium, the objects, and the manner are the
only differentiae of literary forms (1447a16) and Chapter 3 concludes
with a formal declaration that the discussion of the number and kinds
of differentiae is complete (1448b2-3). Further, Chapter 1 classifies literary forms by the medium in which they operate (rhythm, words and
harmony), Chapter 2 classifies them by their objects (serious or unserious people, or alternatively people who are better than us, like us, or
worse), and Chapter 3 by their manner (narrative or dramatic).
No single one of these classifications is sufficient to define or demarcate tragedy, epic, or comedy. All three classifications are needed
to distinguish these literary forms from others. In Chapter 1 we learn
that tragedy and comedy along with dithyrambs and nomes are distinguished from the other genres in that they employ all three media
rhythm, words and harmony. Within this division, tragedy and comedy are distinguished from dithyrambs and nomes in that the latter two
genres employ rhythm, words and harmony all together, while tragedy and comedy do not. Aristotle does not emphasize the place of epic
poetry in this arrangement, but he makes it clear enough that it falls
under a different classification (the classification of arts that employ
words and meter i.e., rhythm42 but not harmony43). Chapter 2 informs us that tragedy and Homeric epic depict agents who are better
than us, while comedy and certain other works depict those who are
worse. And Chapter 3 says that in epic the poet sometimes narrates
and sometimes takes the part of the characters, whereas in tragedy and
comedy the characters are presented doing the actions themselves. Aristotle does not proceed as thoroughly or as systematically as we might
wish. Chapter 1 classifies a large number of kinds of mimsis according
to their medium, but the number is markedly reduced in Chapter 2,
while the only genres mentioned in Chapter 3 are Homeric epic, tragedy and comedy. Since Aristotles concern in the Poetics is just these
42
Cf. 1448b21.
43
Cf. 1447b11-20.
90
Richard McKirahan
We are left with the uneasy feeling that Aristotle did not do an exhaustive survey of all forms of mimsis, and that there may be some
forms not covered by the classification; that is to say, the classification
may not be based on a good empirical study of existing genres. On the
other hand, there is but little attempt to prove that the divisions found
in Chapters 1-3 are the only possible divisions, or that they are the only
correct ones, so there are no good theoretical grounds for accepting the
classification, either.
In this connection it is worth calling attention to Aristotles comment
about people who make visual representations like painting, that some
represent through art and others through habit (1447a19-20). This parenthetical phrase, which Lucas (Lucas 1968, 56 ad loc.) calls quite irrelevant, in fact raises a problem of demarcation. Painting, or poetry that
is done through habit, will use the same medium, and may depict the
same characters and do so in the same manner as that which is done
through art. But then the classifications of Chapters 1-3 are not classifications of arts any more than of habits. That is, the differentiae divide
up two distinct genera. A similar problem is raised by the statement
that most of flute- and lyre-playing are mimeseis (1447a15).
We may suppose that Aristotle could have cleared up some of this
confusion had he wished. But apparently he did not wish to, presumably because of his pressing interest in only three forms of poitik: tragedy, comedy and epic. In these circumstances it would be foolish to
assume that the classifications we find in Chapters 1-3 will be without
problems. We may be in a situation where the best interpretation will
be the one that requires the least damage control.
Now to the classifications themselves. Aristotle begins Chapter 1 by
saying that he will treat poitik and its species (1447a1), goes on to list
six forms of poitik (epic, tragedy, comedy, dithyramb, flute-playing
and lyre-playing) (1447a13-15), and specifies that they are all mimseis
(1447a16). He then says that these arts (tekhnai: 1447a21) differ in three
respects: in the media they employ, the objects they represent, and the
manner of their representation (1447a16-18). Next he illustrates the first
of these respects by means of a comparison with other arts: just as some
people represent many things by means of colors and shapes, and others do so by doing things with their voice, so the arts mentioned above
make their representations in rhythm, words and harmony (1447a1822). It is clear that the arts mentioned here for the first time are also
kinds of mimsis. Are they also kinds of poitik? I think the answer must
be Yes, since painters and sculptors certainly make things their art
has a product. And Aristotles Greek would support this answer, since
he uses the words agalmatopoios for sculptor (Pol VIII 5, 1340a38) and
eikonopoios for portrait-sculptor or -painter (Poet 1460b9).
A consequence of this commonly held interpretation is that the extent of poitik is left quite open. There could be other and quite different arts that also count as forms of poitik; indeed, all arts could
conceivably turn out to be forms of poitik. Further, the line between
mimsis and poitik is left unclear. All forms of poitik are mimseis, but
could it not also be that all mimseis are also forms of poitik? I must set
these questions aside. The text does not help us further, for Aristotles
interest in epic, tragedy and comedy leads him to restrict his discussion
of media to the media relevant to those three particular forms of poitik
and mimsis, and rhythm, words and harmony have nothing to do with
painting, sculpting, and making whatever mouth sounds Aristotle may
have had in mind.
Aristotle now returns to the six species of poitik and declares without argument that they all make their mimsis in rhythm, words, and
harmony (1447a21-2). The preposition in (en) signals that Aristotle
is referring to the medium.44 Here begins the division of mimsis, or
of poitik, or, more precisely, of a certain incompletely specified set of
forms of poitik.45 These three media are found either separately or in
combination.46 The distinction separately/in combination is evidently intended to be mutually exclusive47 and jointly exhaustive,48 which is
appropriate for a set of coordinate differentiae of a genus. Separately
refers to cases where only one of the three media occurs: for example,
genres that employ rhythm but no words or harmony. In combination
refers to cases where more than one of the three media are employed.
There are seven possible cases where one or more of these three occur.
Of the seven, Aristotle mentions five, although he says that two of the
five have no names. As to the remaining two, Aristotle may have deliberately avoided mentioning them, supposing that they are not real
44
This word is the distinctive element in the phrase en heterois, with which Aristotle
originally introduced the notion of media (1447a17).
45
Several commentators (Solmsen (1935), Else (1957), 67, and Battin (1975)) have attempted to force this set into the procrustean mold of dichotomous division.
46
47
Any form of poitik that employs the media separately does not employ them in
combination, and vice versa.
48
Any form of poitik employs either one medium or more than one.
92
Richard McKirahan
49
Harmony cannot stand by itself, since all music is associated, if not with words,
with rhythm, and for the same reason words plus harmony cannot exist without
rhythm Lucas (1968), 62.
50
That is not to say that they are empty. To the contrary, the kind of poitik that employs words alone without rhythm or harmony covers a range of works, such as
the mimes of Sophron and Xenarchos and the Socratic dialogues; it simply has no
name, since prose or logos psilos covers non-mimetic writings as well. Likewise,
the kind of poitik that employs words and rhythm or meter without harmony,
covers much of what we call verse (trimeter, elegiac, etc.) but some of the verse it
covers is non-mimetic, such as Empedocles writings, which Aristotle says is natural science rather than poitik. Epic poetry which is composed in dactylic hexameter and is a form of mimsis, falls under this division. I will need to refer to these
two nameless kinds of poitik in what follows. For convenience I shall call them
mimetic verse and mimetic prose.
51
The distinction presents difficulties. The hama category might seem to cover any
genre that employs all three simultaneously in any part, so that comedy and tragedy would qualify, since they employ all three simultaneously in the stasima. And
kata meros might seem to apply to genres that employ all three media, but not all
simultaneously. However, the particles and connectives prove that tragedy and
comedy fall under the heading kata meros, dithyramb and nomes under hama. And
this accords with another and equally acceptable way of understanding hama and
kata meros which I adopt in my interpretation.
52
All three media are used continuously throughout (Lucas (1968), 61).
53
54
Think of Chairemon (1447b20-3). That worthy did not confine himself to a single
meter, but jumbled them all together in his mixed rhapsody The Centaur. He too
should be called a poet (poits), but it will not do to call him an elegiac poet or a
trochaic poet, or to name him by any other single metrical form.
94
Richard McKirahan
must represent people who are either better than or worse than, or
similar to ordinary people55 another division that is intended to be
mutually exclusive and jointly exhaustive. He goes on to say that tragedy and Homeric epic depict agents better than us, while comedy and
certain other works depict agents who are worse. From Chapter 3 we
learn that in epic the poet sometimes narrates and sometimes takes the
part of the characters, whereas in tragedy and comedy the characters
are presented performing the actions themselves.
I do not propose to make sense out of the bafflingly simplistic claim
at the heart of Chapter 2, that some genres, including tragedy, make
their characters better than people of nowadays and others, including
comedy, make them worse (1448a16-17). But whatever Aristotle may
have meant by it, it provides a means to distinguish tragedy from comedy the principal task left over from Chapter 1.
However, it is crucial to notice that Chapter 2 does more than this.
It does not present a dichotomous division of the category forms of
poitik that employ the media of rhythm, words and harmony and do
so kata meros, i.e., tragedy and comedy. The classification it presents is
said to apply to all forms of mimsis (1448a1), and in particular to all the
forms of mimsis previously mentioned: Each of the mimseis that have
been mentioned will have these differentiae, and will be different by
reason of representing different things in this way (1448a7-9). This division is coordinate, not subordinate to Chapter 1s division by media,
and here lies the solution to another problem, namely, why Aristotle introduces a further coordinate division, in terms of manner (dramatic
or narrative) in Chapter 3.
Here is the problem. Chapter 1 groups comedy and tragedy together, apart from epic. Chapter 2 groups epic and tragedy together, apart
from comedy. Chapter 3 again groups tragedy and comedy together
(as dramatic forms), as opposed to epic (which is narrative). Now if the
divisions of Chapters 1-3 were successive rather than coordinate, this
would be a problem, since all the work needed to differentiate epic, tragedy and comedy would be complete by the end of Chapter 2. Chapter
3 would be redundant, and hence unneeded according to the account
of dichotomous definition. But it is not a problem on the reading I am
recommending.
55
beltionas kath hmas. According to Lucas (1968), 64 ad loc, this is equivalent to the
people of nowadays (tn nun): 1448a18.
56
The crucial lines, 1448a20-4, are difficult and have been taken in different ways. On
any account, there are three manners.
57
58
The only authors mentioned in Chapter 3 are Homer, Sophocles and Aristophanes.
I take it that the latter two are representative of tragedy and comedy, and I suppose
Homer is meant to represent epic in the same way. But these are the only three genres
that appear, and unless we can find some way to deal with the remaining forms of
poitik, Aristotle is in trouble. Not only is the division incomplete, but the attempt
to demarcate epic, tragedy and comedy from other forms of poitik founders.
But Aristotle may have had legitimate reasons for not mentioning other
genres. In the first place, the present division may not be meant to be a division
of all forms of poitik but only of those that employ words. Painting, dancing
and instrumental music could simply be excluded from the division. (See p. 97
below.) Further, Aristotle may have thought that the other forms of mimetic verse
which until now have not been distinguished from epic either do not fall into
this division either, or, if they do, that they fall under the non-Homeric manner of
narrative. This claim does not apply to dithyrambs, but this is not a problem for
my interpretation, since they were classified differently than epic in Chapter 1.
But it plausibly applies to elegiac poetry, which is the only other genre of mimetic
verse that Aristotle names. (Chairemons aberrant mixed rhapsody is hardly a
genre (1447b20-2).) And Aristotle might have supposed that no other variety of
mimetic verse had the particular kind of narrative that epic does.
96
Richard McKirahan
that do not employ words may have no place in the division according to manner. But perhaps this is not a problem: it is equally unclear
whether every species of animals will have a place in every one of the
main branches of the classification of animals. Still, it may be a problem,
and at least it is an untidy situation.
Here is one way to tidy it up. Aristotle emphasizes that privatives
can occur as legitimate differentiae in the new method of classifica59
tion. It will then be possible to have none of the above as a differentia at any point in a division. Narrative and dramatic are mutually
exclusive differentiae. If they apply only to epic, tragedy and comedy,
or only to those forms of poitik that employ words, Aristotle will permit the heading of manner to be divided initially by the differentiae
dramatic/narrative/neither, which will make room for all the forms
of poitik. This subdivision of genres that employ all three media may
seem arbitrary and ad hoc, but I do not think it is. I claimed above that
Aristotles special interest in tragedy, comedy and epic leads him to
neglect making a thorough classification of the forms of poitik. Recall
that he identifies flute-playing, lyre-playing and pipe-playing as distinct arts, and recognizes others too. This is proved by the etc. clause
at 1447a24-5. All these arts fall under the heading of genres that employ
rhythm and harmony but not words. In order to distinguish these arts
from one another, Aristotle would have had to draw further distinctions
within that heading, possibly somewhat like modern classifications of
musical instruments into stringed instruments, wind instruments, etc.,
with further subdivisions brought in as needed. I suppose he chose not
to do so because it would have taken him too far afield from the genres
he is set to discuss: tragedy, comedy and epic.
Likewise, Aristotle recognizes forms of poitik that can be identified
by the meter employed (elegiac, trimeter, etc.) what I call mimetic
verse. They all belong to the classification of genres that use rhythm
and words but not music. Again, Aristotles present concern is not to
identify all possible or actual metrical forms, and he does not do so. But
again, had he wanted to, he could have made the attempt. And again I
suppose his reason for not doing so is that it would have taken him too
far from his brief.
But I need to say a little more, since epic is part of his brief, and
epic belongs in this division. How then could Aristotle demarcate epic
59
PA I 3, 643b24-6
98
Richard McKirahan
from other poetic verse forms? An obvious way to differentiate mimetic verse is by the kind of meter employed: elegiac couplets, iambic trimeters, dactylic hexameters, etc. Aristotle says that people do in
fact distinguish genres of verse by their meter, but complains that there
is something misleading in calling anyone who composes in dactylic
hexameter an epopoios a title inappropriate to those like Empedocles
who do compose in that meter, but who are not really poets poitai
because their work is not a form of mimsis. Aristotle invites us to
distinguish between poets who employ hexameters and others who use
the same meter, reserving the title epopoios for the former. Admittedly
epic was not the only genre composed in dactylic hexameters. I do not
know how Aristotle would have dealt with the Homeric Hymns for
example, or the Works and Days. Could he have brushed them aside
as easily as he did Empedocles? But just as tragedy and comedy are
not yet distinguished one from the other at the end of Chapter 1, and
it requires a different line of division (in Chapter 2) to separate them,
epic too can be marked off from other genres of hexameter poetry by
further divisions. Alternatively, the opening lines of chapter 6 suggest
that Aristotle thought of mimtik in hexameters as equivalent to epic,
so the problem I just identified may have not been a problem for him,
or may have been invisible to him.
And now the second problem. We saw that the division of media in
Poetics Chapter 1 (possibly the division in Chapter 2 as well) puts the
multiple differentiae to work in a particular way. It does not simply
take the differentiae as exclusive alternatives as happens in dichotomous division, but it considers their possible combinations one, two,
and three at a time. Is this procedure compatible with the new method
of classification?
I believe it is. The solution is an extension of the solution to the previous problem. If painting, say, employs certain media and represents
certain objects, but does not involve manner as Aristotle conceives it,
we saw that we can deal with that fact by putting it into the none of the
above heading, which is at the very beginning of the division of manner. The same will apply in the present case too. The three differentiae
of medium are not mutually exclusive. So the initial division of media
will be into not three but a maximum of eight mutually exclusive classes, ranging from the class that has rhythm, words and harmony (that
is to say, all three types) through the three classes employing two of
the three differentiae, the three classes employing just one of the three
differentiae, and finally the class that employs none of the three. Some
of these combinations are not instantiated and may be impossible, or
100
Richard McKirahan
60
61
At 1447b25 Aristotle substitutes melos (melody) for harmonia (harmony), but evidently intending no difference in meaning.
62
Lucas (1968), 99
63
102
Richard McKirahan
64
Success and failure refer to and point ahead to the discussion of the structure of
tragic plots later in the work, which gives particular emphasis to stories in which
the main character has a change of fortune from good to bad or vice versa; this
change is the action which the plot represents. Cf. 1450a3-4.
Conclusion
In reading the Poetics a few years ago I was struck by how Aristotelian a
work it is, and in trying to pin down in what way it is Aristotelian and
at the same time why it is so bizarre as either a history of literature or
as a manual on how to write successful tragedies although it claims
to treat those topics too I found that those elements (the historical
and the how-to aspects) are subordinate to the definition of tragedy
which Aristotle takes such great pains to discover. The centrality of
definition, and therefore of essence, is in my opinion the Aristotelian
kernel of the work, and the work done in reaching this kernel and seeing what follows from it are reminiscent of many of Aristotles other
writings. Once this connection became apparent and also the close links
to the APo approach to science, it turned out that the Poetics began to
look more central to Aristotles work than many have thought. Scholars of Aristotle are familiar enough with disparaging remarks about
the Poetics. (I recall my mentor Gwyl Owen once dismissing the Poetics
as an inferior appendix to the Rhetoric.) I see this paper as a step not
only towards rehabilitating that work as significantly Aristotelian (if
such rehab work is even necessary), but also of exploring how themes
prominent in Aristotles early work develop and continue to affect the
65
work of his later years.
65
I wish to thank the organizers of the conference for inviting me and for arranging
the publication of the proceedings. I am also grateful to David Reeve for his useful comments (some of which have led me to rethink and rewrite certain parts of
the paper for the published version) and to the participants in the conference for a
lively and helpful discussion of my paper. I owe particular thanks to Jim Lennox
for an after-hours discussion which clarified my thinking about the relation of APo
II 8-10 to the paper, and to Gisela Striker for references to some passages of Aristotle that have an important bearing on the thesis of my paper.
104
Richard McKirahan
References
Battin, W.P. 1975. Aristotles Definition of Tragedy in the Poetics, Journal of Aesthetics and
Art Criticism 33:155-70, 293-302.
Else, G.F. 1957. Aristotles Poetics: The Argument (Cambridge, MA: Harvard University
Press).
Heath, T.L. 1949. Mathematics in Aristotle (Oxford: Oxford University Press).
Jope, J. 1972. Subordinate Demonstrative Science in the Sixth Book of Aristotles Physics,
Classical Quarterly 22: 279-92.
Lucas, D.W. 1968. Aristotle: Poetics (Oxford: Clarendon Press).
McKirahan, R. 1992. Principles and Proofs. Aristotles Theory of Demonstrative Science
(Princeton, NJ: Princeton University Press).
Ross, W.D. 1949. Aristotles Prior and Posterior Analytics (Oxford: Oxford University
Press).
Solmsen, F. 1935. The Origins and Methods of Aristotles Poetics, Classical Quarterly 29:
192-201.