Jump to ContentJump to Main Navigation
Aristotle’s Categories in the Early Roman Empire$

Michael J. Griffin

Print publication date: 2015

Print ISBN-13: 9780198724735

Published to Oxford Scholarship Online: March 2015

DOI: 10.1093/acprof:oso/9780198724735.001.0001

Show Summary Details
Page of

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (www.oxfordscholarship.com). (c) Copyright Oxford University Press, 2016. All Rights Reserved. Under the terms of the licence agreement, an individual user may print out a PDF of a single chapter of a monograph in OSO for personal use (for details see http://www.oxfordscholarship.com/page/privacy-policy). Subscriber: null; date: 08 December 2016

(p.233) Appendix 3 An Outline of Aristotle’s Categories

(p.233) Appendix 3 An Outline of Aristotle’s Categories

Aristotle’s Categories in the Early Roman Empire
Oxford University Press

The structure, textual unity, authenticity, and function of the Categories, as well as its place in Aristotle’s philosophy, have been considered in much greater detail elsewhere.1 My intention here is just to survey the organization of the treatise as it presented itself to the readers discussed throughout this book, beginning with Andronicus of Rhodes.

  1. 1. The Ante-praedicamenta. The opening section, traditionally labelled the ‘ante-praedicamenta’ (literally ‘what comes before the categories’), comprises the first three traditional chapter headings (corresponding to Bekker numbers 1a1–15, 1a16–1b9, and 1b10–24 respectively).

    1. a. Chapter 1: The ‘-onymies’. Ch. 1 (1a1–15), sometimes called the ‘-onymies’ in modern scholarship, introduces three kinds of items. These items are labelled, respectively, ‘homonyms’, ‘synonyms’, and ‘paronyms’. Roughly speaking, this chapter explores how, in ordinary Greek, one term successfully picks out multiple referents, either equivocally (homonymy), univocally (synonymy), or with a change in the grammatical accidence of the term (paronymy). Things are then carved up from the perspective of these ordinary linguistic phenomena: for instance, a human being and an ox can both be called ‘synonyms’ because, in ordinary Greek, they are both correctly addressed by the term ‘ζῷο‎ν’ (animal).

      The chapter proceeds as follows:

      1. i. ‘Homonyms’ are items that are related to each other only in that they can be addressed by the same referring term (τοὔνομα‎) in ordinary Greek. The examples offered are a human being and a painting, both of which can be validly referred to by the noun ‘ζῷον‎’ (which in ancient Greek is equivocal between ‘animal’ or ‘painting’). Besides the fact that I can validly address that painting and that person using the word ‘ζῷο‎ν’, the painting and person have nothing in common.

      2. ii. ‘Synonyms’ are items that are related to each other in two ways: like homonyms, they can be addressed by the same term in Greek, but they are also related in that they share something (p.234) called a ‘λόγος τῆς οὐσίας‎’ in common. It has been contentious since antiquity just what a ‘λόγος τῆς οὐσίας‎’ is, and English translations vary, but ‘statement of being, i.e. definition’ may not be far off. The examples offered are a human being and an ox. They may both be addressed by the term ‘ζῷο‎ν’, but they also share a definition in common, namely the definition answering to the term ‘ζῷον‎’: both persons and oxen are animals, i.e. animated beings.

      3. iii. ‘Paronyms’ are items that can be addressed by a name derived from something else, with a difference of ending. For example, a grammarian is so-called from grammar, and a brave person is so-called from bravery.

      It is today,2 and was in antiquity, hotly debated why the Cat. begins with this chapter. Around the first century BCE, for instance, Andronicus argued that some preliminary notion of homonyms, synonyms, and paronyms was valuable for understanding the rest of the treatise (Simplicius in Cat. 21,21–4), while various critics, such as Lucius and Athenodorus, contended that the first chapter failed to cohere with the rest of the work. An editor may have tacked it on to the central portion of the treatise because he believed its explanation of unfamiliar concepts or words would be useful for the remainder. (That is more or less the explanation offered by the commentators, although it is worth noting that the terminology of the first chapter is not employed regularly throughout the rest of the treatise.)

    2. b. Chapter 2: ‘Things that are spoken’ and ‘things that exist’. The second chapter heading deals with two groups of things in sequence:

      1. i. Things that are said (which may be said with combination or without combination). First (1a16–19), Aristotle considers ‘things said’ (λεγόμενα‎). These are divided (perhaps exhaustively) into two: (p.235) there are things said ‘without combination’, i.e. single terms considered on their own, such as ‘person’ and ‘runs’; and there are things said ‘with combination’, i.e. complete sentences that include both a noun and a verb, such as ‘person runs’. This division is echoed in the De Interpretatione, where it is explored in much greater detail.3 It is very important to Andronicus and the later philosophers discussed in this book, who consider the former class (‘things said without combination, i.e. simple terms signifying simple items) to be the scope of the Categories, and anticipate the Neoplatonic argument that for this reason the Categories describes the building blocks of Aristotelian logic—since Aristotle proceeds in De Int. to construct from these building blocks ‘combined’ sentences (signifying states of affairs), then syllogisms (Prior Analytics), and finally demonstrations (Posterior Analytics). (For a thoughtful criticism of these arguments, see Morison 2005; for their historical origins, see Bodéüs 2001: xiv–xvi, and the introduction to Hadot 1990.)

      2. ii. Things that exist (which may be SAID-OF but not IN a subject, or IN a subject but not SAID-OF a subject, or both SAID-OF and IN a subject, or neither SAID-OF nor IN a subject). In this part of ch. 2 (1a20–b9), Aristotle seems to be concerned with real beings (ὄντα‎), chiefly those whose names4 can validly stand in the predicate position of a sentence like ‘Socrates is human.’ He is interested in sentences that include a subject term (‘Socrates’), a copula (the verb ‘is’), and a predicate term (‘human’): schematically, ‘A is B.’ (It seems clear that Aristotle means to talk about predicative sentences, although the technical verb κατηγορῆτα‎ι does not appear (p.236) until 1b10.) He divides beings whose names can stand in the predicate position of such a sentence into four kinds. This fourfold division is constructed from two relationships that these beings might hold towards the being whose name stands as the subject (ὑποκείμενον‎) of the sentence.

  2. 1. The two relationships: SAID-OF AND IN

    1. a. The first relationship is called being SAID-OF5 something as a subject (καθ’ ὑποκειμένου τινὸς λέγεται‎). It indicates an ontological relation that is typically, but not invariably, reflected in everyday speech. In general, B is SAID-OF A as a subject just in case we validly talk about ‘A’ as ‘a B’ (in Greek, about ‘A’ as ‘B τις‎’). Thus Socrates is a human, human is an animal, and grammar is a (branch of) knowledge.6 (Interestingly, Aristotle takes the fact that one can validly say ‘A is a B’ in ordinary Greek as a good indicator, at least heuristically, that A really is a B.) In the framework of ten κατηγορία‎ι that Aristotle will later outline, the SAID-OF relationship operates intra-categorically, or so to speak ‘vertically’: it relates the universals within a category (those items which are ‘said of many things’, De Int. 17a39–40)7 to the items of which they are said in the same category. For instance, within the category of REALITY (οὐσία‎, traditionally rendered ‘substance’),8 genera like ‘animal’ are SAID-OF species like ‘human’, while species are SAID-OF individual members of the species, which are primary realities like Socrates (ch. 5, 2a11–18); likewise, in the category of QUALITY (ποῖα‎), knowledge is SAID-OF a piece of grammatical knowledge. Andronicus of Rhodes (ap. Simplicium in Cat. 54,8–21) noticed that Aristotle’s linguistic heuristic fails to guarantee this intra-categorial relation in some cases: for example, we can validly say that ‘Socrates is (p.237) an Athenian’, implying that Athenian is SAID-OF Socrates—but ‘Athenian’ seems to pick out something in the category of QUALITY, while ‘Socrates’ picks out a REALITY.

    2. b. The second relation is called being IN a subject (ἐν ὑποκειμένῳ‎), which Aristotle describes as being ‘in something, not as a part, and unable to exist apart from what it is in’. For instance, some grammatical knowledge is IN my mind, and pallor is IN Socrates’ body (assuming his body is pale). Aristotle’s description stresses that the pallor is not constitutive of Socrates’ body—the body is not a whole of which this colour is a constitutive part—and likewise, my knowledge of Greek (say) is not constitutive of my mind. This relation is trans-categorial: it applies in cases where some quality, quantity, or another of the latter nine categories ‘inheres’ in a reality (οὐσία‎) such as Socrates or Bucephalus. There is a linguistic check that we can deploy to confirm this, since inherence of B in A applies when I can say that ‘A is B’ but I cannot say that ‘A is a B’: Socrates is not a pale (but pallor is in him), and my mind is not a knowledge (but knowledge is in it).

This distinction between two kinds of predication was taken to be crucial in antiquity, especially as interest in Aristotelian logic began to collect momentum after the first century BCE. Someone who accepts this subdivision of predication into the SAID-OF and IN relations, based on the linguistic heuristics that Aristotle highlights, might then be more willing to distinguish the ontological inherence of a non-essential feature in a subject (such as pallor in Socrates) from a subject’s essentially belonging to a kind (such as Socrates’ belonging to the human species). In this way, these relations also pave the way for essentialism, which is critical for subsequent Aristotelian logic and metaphysics. Ancient Peripatetics like Alexander of Aphrodisias criticized the Stoics for failing to appreciate just this distinction in their logic, and treating all attributes as equal (Alex. in Top. 42,27–43,2: see LS 32E). I argue in ch. 2 that Andronicus placed considerable weight on the lessons of this chapter, and was partly motivated by the perceived failure of (p.238) his predecessors to take proper stock of the difference between essential and accidental predication.

  1. 2. The fourfold division

    Cat. 2, 1a20–b9

    Beings IN a subject

    Beings not IN a subject

    Beings SAID-OF a subject

    Genera and species of items in non-οὐσία‎ categories

    Genera and species of primary οὐσίαι‎ (i.e. secondary οὐσίαι‎)

    Beings not SAID-OF a subject

    Individual items in non-οὐσία‎ categories, such as ἡ τις γραμματική‎ (‘some particular knowledge of grammar’)

    Primary οὐσίαι‎, individuals, e.g. ὁ τις ἄνθρωπος‎ (‘some particular human being’)

    1. a. Aristotle next uses these relations to produce a grid9 of four kinds of beings, illustrated in the following table (adapted from Reinhardt 2007: 515).

      1. i. Beings that are IN a subject will prove to belong to the non-οὐσία‎ categories (quantity, quality, relation, and so on), and may be universals (if they are SAID-OF a subject) or particulars (if they are not SAID-OF a subject).

      2. ii. Beings that are not IN a subject are realities (οὐσίαι‎), and may also be universals (if they are SAID-OF a subject) or particulars (if they are not SAID-OF any subject). This latter class—beings that are neither IN nor SAID-OF a subject—are called ‘primary οὐσίαι’‎: these are particular beings such as Socrates and Bucephalus.

One of the most influential implications of this fourfold division in antiquity was the statement that particular beings, such as Socrates and Bucephalus, count as ‘primary realities’. Some Platonist critics of the Categories discussed in this book, such as Lucius and Nicostratus (ch. 4), were alarmed, partly on this basis, that the Categories omitted the intelligible Platonic Forms as primary beings altogether. Aristotelian readers, like Boethus of Sidon (ch. 5), introduced the question how this account of primary οὐσία‎ as ‘neither SAID-OF nor IN a subject’ could jibe with the hylomorphic analysis of οὐσία‎ in terms (p.239) of Form and Matter in Aristotle’s Metaphysics (Simplicius in Cat. 78,4–20).

  1. 2. The Praedicamenta.

    1. a. Initial list. With Cat. 4 (1b25–2a4), Aristotle introduces a list of ten items ‘said without any combination’, which signify (σημαίνει‎) either (1) realities (οὐσίαι‎), (2) quantities, (3) qualities, (4) relations, (5) where, (6) when, (7) position, (8) having, (9) acting, or (10) being acted-on, and provides examples of each. We seem to have here not just ten, but twenty, items. First, there are ten kinds of simple, referring terms (‘things said without combination’, which, as Aristotle stresses from 2a4–10, cannot produce an affirmation or denial). Secondly, there are ten kinds of beings to which these terms refer. This latter list of ten falls under the fourfold division just offered above: the οὐσίαι‎ are those predicates that are not IN any subject; the remaining non-οὐσία‎ items are those predicates that are IN subjects. (Within each categorial ‘silo’ there will also be universal items SAID-OF more specific items in the category.) The use of the verb σημαίνει‎, coupled with the preceding divisions of ‘things said’ and ‘things that are’, suggests that we have to do here with a relationship between simple words meaning things. This implication was especially important in antiquity, as the entire treatise Categories was taken to be concerned with how simple words signify or refer to simple things—that is, to be primarily concerned with semantics (cf. Simplicius in Cat. 13,11–18), or with referring terms just insofar as they refer.

    2. b. Detailed descriptions of the distinguishing characteristics of the categories. These collected reflections on what is distinctive of each category do not seem to constitute a proper definition of any one of them, in terms of genus and differentiae, but then it would seem to be impossible to ‘define’ an ultimate genus, which has no differentiae. The ancient critics with whom this book is concerned, such as Boethus of Sidon (ch. 6), also seem to have found in these descriptions a valuable, non-technical ‘sketch account’ that could serve to ‘actuate’ beginners’ preconceptions of the highest genera, beginning the road to logic.10 As Menn (1995) suggests, these brief, non-technical discussions of the most distinctive features of each genus are certainly helpful in a (p.240) pragmatic way, providing a set of rules of thumb for a dialectician to recognize the genus to which a given term belongs.

      1. i. Chapter 5: οὐσία‎. Aristotle states that primary οὐσίαι‎ are distinguished by being neither SAID-OF nor IN a subject (2a11–18, etc.)—which sparked some of the Platonist criticisms mentioned above, by readers such as Lucius and Nicostratus. Species and genera of οὐσίαι‎ are real, but secondary to the primary realities themselves. Aristotle goes on to describe several other distinguishing features or propria of οὐσίαι‎: for instance, Aristotle notes that everything that carries the name of an οὐσία‎ is so named synonymously (3a33–b9, picking up the vocabulary of ch. 1); that every οὐσία‎ seems to signify a certain ‘this’ (3b10, suggesting that again we might have to do here with both a word, such as ‘human’, and a being, this Socrates right here, and the semantic relationship between them); that nothing is contrary to οὐσίαι‎ (3b24–32); (iv) that they do not admit of a more or less (3b33–4a9); and most of all, that what is numerically one and the same is able to receive contraries (4a10–21).

      2. ii. Chapter 6: QUANTITY. Aristotle divides quantities into the discrete and continuous, and those composed of parts with position and those that are not so composed; he also gives a number of examples. Among the distinguishing features of quantity that he mentions, he notes that quantity has no contrary (5b11) and does not admit of more and less (6a19). There were some criticisms of Aristotle’s account of this category in antiquity, especially the decision to include time and place under quantity; but they did not have the broad range of the more fundamental questions about substance, essentialism, and the general purpose of the book, noted above.

      3. iii. Chapter 7: RELATIVE. Aristotle offers a short, crisp description of a relative at 6a36–7: ‘We call relatives all such things as are said to be just what they are, of or than other things, or in some other way in relation to something else’ (tr. Ackrill). He gives examples such as the ‘larger’, which is larger than something, and offers another selection of distinguishing features of the category: relatives have contrariety (6b15); they admit of more and less (6b19); they must reciprocate (if slaves are of masters, then masters are of slaves, 6b28 and following); they are simultaneous by nature (7b15). But Aristotle raises a problem about whether the description given at the start of the chapter can show that relatives are never spoken of as substances, and this prompts him to offer what looks like a second and different description of the relative at 8a28–34, as items ‘for which being is the same as being related to something’. This second description prompted all sorts of trouble in antiquity; Andronicus (p.241) and his contemporary Ariston, for example, proposed that it was circular (because the definition contained the definiendum), and advanced an alternative definition of the relative as that whose being was somehow related to something different (see ch. 2).11

      4. iv. Chapter 8: QUALITY. Aristotle divides qualities into states and conditions, capacities and incapacities, affective qualities and affections (which seem to have interested Andronicus, although this is largely beyond the scope of the present study), shape, and (as a borderline case) texture. He uses the terminology of paronymy from ch. 1 to explain the relationship between a quality (like pallor) and the qualified (the pale man), and offers an interesting reflection on cases where ordinary language fails to reflect the ontological reality (10a27–b11). Some of the features of qualification include the following: it admits contrariety (10b12) and more and less (10b26), and most distinctively, only in virtue of qualities ‘are things called similar and dissimilar’ (11a15–19).

      5. v. Chapter 9: ACTING and BEING ACTED ON. Very briefly (11b1–9), Aristotle offers an account of two important features of these two categories: they admit of contrariety and more and less. The rest of the chapter (11b10–14)—which explains why there is no further account of the categories’ position, when, where, and having—is an interpolation.

  2. 3. The post-praedicamenta (‘what comes after the categories’). After a short interpolated introduction, which simply states that ‘something must be said about opposites…’ (11b15–16), a new discussion begins, treating opposition, priority and simultaneity, kinds of movement or change, and the usage of the word ‘have’. Andronicus of Rhodes already maintained that this section was appended ‘against the purpose of the treatise’ (ap. Boethium in Cat. 263B).


(1) Frede 1987: 11–48, Bodéüs 2001: xi–clxxxviii, Mann 2000, Wedin 2000, and Menn 1995; see also Ackrill 1963 for English translation and commentary.

(2) Michael Wedin (2000: 11–37) cites this as a question posed by early scholars: ‘The book begins without introductory remarks that give its purpose and object’ (Brandis 1833: 267), while this peculiar introduction makes for ‘the lack of any very definite information as to Aristotle’s precise object in formulating’ the theory of categories (Ross 1924, 1, lxxxii). It is also posed by Simplicius. One answer is found in Ackrill (1963: 69), who follows the ancient Greek and Latin commentary tradition in describing 1a1–15 as containing ‘certain preliminary points and explanations’ presupposed by the main body of the work. John Rist (1989a: 94), by contrast, lays aside 1a1–15 as ‘loosely connected’ with the rest of the text. According to other interpretations, as Barrington Jones rhetorically suggests (1972: 117), the first chapter appears as ‘an incidental excrescence on the work’. Yet many critics, such as Jones (1972), Dancy (1975), Frede (1987), Furth (1988), Lewis (1991), Wedin (2000), and Mann (2000), hold that a coherent account of 1a1–15 is valuable for the interpretation of the text.

(3) Just what Aristotle means by ‘things spoken’ here is also a matter of heated discussion: they may be the objects or states of affairs signified by linguistic utterances, or the utterances themselves. The latter would be a reasonable guess, since the ‘combination’ described seems to apply more intuitively to words (the noun ‘person’ and verb ‘runs’ are ‘combined’ to make a sentence) than to things.

(4) I try to distinguish clearly here between use and mention. As for the modern complaint that Aristotle disregards the use–mention distinction, one might agree with Bäck (2000) when he remarks that Aristotle appears to ‘switch back and forth from speaking about words to speaking about real things’ and does so ‘quickly and blithely’ with confusing rapidity (133–4). Two recent approaches to this ambiguity are evident in Wedin (2000) and De Rijk (2002:I). As a policy, Wedin will ‘follow Aristotle’s practice of disregarding use–mention boundaries in formulating a number of his theses’ (12 n. 6). By contrast, De Rijk, who expresses this equivocation as ‘the absence of a clear-cut borderline between a linguistic expression […] taken as a linguistic tool, and its significate’, proposes that Aristotle exploits it intentionally and systematically, and christens this ‘the rule of indiscriminate reference’ (63–4)—a necessary law and ‘main rule’ for his exegetical approach to Aristotelian semantics.

(5) I am here following Lewis (1991) and Mann (2000) in capitalizing these names to flag their technical usage by Aristotle.

(6) For example, Aristotle states that human (ἄνθρωπος‎) is said-of an individual human being (τοῦ τινὸς ἀνθρώπου‎), like Socrates, as a subject, and knowledge (ἐπιστήμη‎) is said-of an individual branch of knowledge, like grammatical expertise (τῆς γραμματικῆς <ἐπιστήμης>‎). More examples follow in ch. 3 (1b10–15), now using the technical verb ‘predicate’: ‘human is predicated of the individual human (ἄνθρωπος κατὰ τοῦ τινὸς ἀνθρώπου κατηγορεῖται‎), and animal of human…’

(7) Items that can be ‘said of many things’, in the language of De Int. 17a39–40.

(8) I will sometimes avoid the traditional Latinate label ‘substance’, although it is firmly entrenched, just because it is rather misleading in English; see Loux 1991: 15–16. Best of all is to leave ousia untranslated, but that can also be off-putting to the Greekless reader. As Loux points out, ‘reality’ is too much like ‘being’ to clearly draw out Aristotle’s distinction between ousia and on, but it has the advantage of being an abstract noun in English, and it is reasonably clear.

(9) This may not literally have been intended as a diagram, although it is possible: for Aristotle’s use of diagrams and tables in teaching, see recently Natali 2013: 114–15.

(10) I refer several times in this book to Simplicius in Cat. 159,10–15, where Boethus argues that it was not possible to give proper, formal definitions of the primary genera at the early stage represented by Cat., but that Aristotle provides ὑπογραφαί‎ or sketch descriptions that can actuate or ‘stir up’ (ἀνακινεῖν‎) our concept (ἔννοια‎) of a category like the Relative. (That the central idea is Boethus’, I think, is strongly suggested by this passage and others like 163,28–9; I also think that the verb ἀνακινεῖν‎ is Boethus’ own, although it could also be a paraphrase by Simplicius or his source; it has strong undertones in the Platonic tradition.)

(11) Other ancient readers began the tradition that the first ‘definition’ of the relative might apply secundum dici, cued by the introductory phrase ‘We call…’, while the later ‘definition’ might apply secundum esse.