PLATO’S ACCOUNT OF FALSEHOOD PLATO’S ACCOUNT OF FALSEHOOD PLATO’S ACCOUNT OF FALSEHOOD PLATO’S ACCOUNT OF FALSEHOOD A STUDY OF THE A STUDY OF THE A STUDY OF THE A STUDY OF THE SOPHISTSOPHISTSOPHISTSOPHIST

Paolo Crivelli

CONTENTS

INTRODUCTION 2

CHAPTER I DEFINING THE SOPHIST BY THE METHOD OF DIVISION 4

CHAPTER II PUZZLES CONCERNING NOT BEING 17

CHAPTER III PUZZLES CONCERNING BEING 54

CHAPTER IV FORMS, DIALECTIC, AND NOT BEING 68

CHAPTER V FALSE STATEMENT AND FALSE BELIEF 150

APPENDIX PLATO’S THEORY OF TRUTH AND FALSEHOOD FOR PRIMARY

STATEMENTS: FORMAL PRESENTATION 172

REFERENCES 175

INTRODUCTION

The Sophist ’s professed purpose is to define the sophist (218B7−218C1).1 The dialogue has the structure of a sandwich, with a frame surrounding a core. The frame section (216A1−236D4 and 264B11−268D5) pursues the project of defining the sophist by the method of division. The core section (236D5−264B10) presents and solves certain puzzles connected with the concept of falsehood.

The connection between the frame and core sections is straightforward. A definition of the sophist is attempted which relies on a description of the sophist as someone who makes false statements and thereby induces false beliefs in those who hear those statements. This description clashes with the falsehood paradox, which is summoned by way of objection. The falsehood paradox is a family of arguments whose conclusion is that there are neither false statements nor false beliefs. I say a ‘family of arguments’ because there are many subtly different arguments with this counter-intuitive conclusion. Accordingly, I often speak of a ‘version of ’ the falsehood paradox. Versions of the falsehood paradox appear in other dialogues (Euthd. 284B1−284C6; Cra. 429C6−430A5; Tht. 167A6−8; 187C7−200C7, especially 188C9−189B9, cf. R. V 478B5−478C2): only in the Sophist does Plato solve one (or more) of them, although his earlier presentations already suggest some awareness of the disarming procedure.2 As for statements, the main version of the falsehood paradox in the Sophist goes as follows: making a false statement is stating what is not, which amounts to stating nothing, which in turn amounts to making no statement; so, making a false statement amounts to making no statement; therefore there are no false statements. Parallel steps lead to the result that holding a false belief amounts to holding no belief, and thereby to the conclusion that there are no false beliefs.

The Sophist’s core section (236D5−264B10) divides into an aporetic subsection (236D5−251A4) and a constructive one (251A5−264B10). The aporetic subsection, which goes through puzzles, divides into two parts: the first (236D5−242B5) presents puzzles that concern not being, images, and falsehood in statements and beliefs; the second (242B6−251A4) purports to show that the concept of being is no less problematic than that of not being. The constructive subsection also has two parts: the first (251A5−259D8) contains an analysis of negative predication based on the concept of otherness, and on its basis develops an account of not being that is free from paradox; the second part (259D9−264B10) deploys this account of not being to explain falsehood in statements and beliefs.

1 The references to pages, sections, and lines of Plato’s works are based on the OCTs (Burnet’s edition except for vol. I, where the edition by E. A. Duke, W. F. Hicken, W. S. M. Nicoll, D. B. Robinson, and J. C. G. Strachan is used, and the Republic, where the edition by S. R. Slings is adopted).

2 Cf. Burnyeat (2002).

CHAPTER I DEFINING THE SOPHIST BY THE METHOD OF DIVISION

The task of defining the sophist. The last sentence of the Theaetetus and the first one of the Sophist present the latter dialogue as a continuation of the former. For the last sentence of the Theaetetus is the following (Socrates is speaking): ‘Now I must go to the King’s Porch to meet the indictment that Meletus has brought against me; but let us meet here again in the morning, Theodorus’ (Tht. 210D1−4). The first sentence of the Sophist is the following (Theodorus is speaking): ‘We have come at the proper time by yesterday’s agreement, Socrates’ (Sph. 216A1−2, cf. Plt. 258A3−4).

Theodorus has taken along someone who was not present at the previous day’s discussion. The newcomer is a visitor from Elea (henceforth I shall refer to him by the acronym ‘ES’, for ‘Eleatic Stranger’). He is described by Theodorus as ‘a member of the group who gather around Parmenides and Zeno’ (216A3) and as being ‘very much a philosopher’ (216A4). This remark prompts Socrates to observe that it is difficult to recognize philosophers: just as gods sometimes disguise themselves as humans and roam the earth unrecognized (216C4−6),1 so also philosophers, who are god-like humans (216B8−216C1), cannot be easily identified (cf. 254A8−254B1). In particular, it is hard to distinguish philosophers from sophists and statesmen, whose appearance philosophers sometimes take on (216C8−216D1). The fact that philosophers can be easily confused with sophists and statesmen induces Socrates to ask the ES to discuss them (216D3−217A4, cf. Plt. 257A3−5). The speakers agree (without argument) that there are three reciprocally distinct kinds corresponding to the three names, ‘sophist’, ‘statesman’, and ‘philosopher’ (217A5−217B2). They also acknowledge that it is difficult to define these three kinds (217B2−4) and the ES accepts to attempt to produce definitions of them. He will do this by asking questions to an interlocutor (rather than by holding a long speech on his own). Theaetetus is the chosen respondent for the first part of the enterprise.

The discussion leading to a definition of the sophist is reported in the homonymous dialogue, the Sophist; similarly, the discussion leading to a definition of the statesman is reported in the Statesman. It remains unclear whether Plato planned a third dialogue, the Philosopher, aimed at defining the philosopher. He surely never wrote it.2 It is important to realize that the definition of the sophist is undertaken with a view to defining the philosopher: defining the sophist will enable one to distinguish a merely apparent philosopher from a genuine one. It is perhaps because of their concern with definitions that the Sophist and the Statesman are listed as ‘logical’ dialogues in Thrasyllus’ edition of Plato’s works.3

1 Cf. Od. 17. 484–7: line 486, partially quoted here, appears in its integrity at R. II 381D3−4.

2 Some commentators (e.g. Cornford (1935), 168−9) believe that the Philosopher was planned but never written, others (e.g. Lauer (1966), 146 and Notomi (1999), 24−5) that it was not even planned.

3 Cf. Cavini (1995), 123.

5

An outline of Plato’s method of division. The method of division (tro/poj th=j diairh/sewj, Sph. 235C9, cf. Cra. 424B7−8)4 enables one to find a definition by ‘dividing’ kinds or forms into forms.5 Plato never offers a ‘textbook description’ of the method of division (the passages from the dialogues that come closest to being such a description are Sph. 254D1−254E3 and Phlb. 16B5−18D2).6

The translation of certain technical terms in Plato’s method of division raises peculiar difficulties. In particular, the terms ‘ge/noj’ and ‘eiådoj’ have not only an absolute use, in accordance with which a certain item is called a ‘ge/noj’ or an ‘eiådoj’ in its own right, but also a relative use, in accordance with which a certain item is called a ‘ge/noj’ or an ‘eiådoj’ of a certain item (in particular, the relative uses reciprocate: ‘ge/noj’ is true of x relatively to y just in case ‘eiådoj’ is true of y relatively to x). The double (absolute and relative) use of ‘ge/noj’ may be accommodated by adopting ‘kind’ as a translation for both the absolute and the relative use, ‘genus’ for the relative use only. In the case of ‘eiådoj’, I adopt ‘form’ as a translation in the case of the absolute use and ‘species’ in the case of the relative use.

The following rough characterization of Plato’s method of division seems a fair account of his scattered remarks and of his practice.

To divide a kind is to identify two (or possibly a larger finite number of ) forms that are (1) immediately subordinate to the kind that is being divided, (2) pairwise disjoint, and (3) exhaustive of the kind that is being divided. For instance (cf. Sph. 219A8−219C9), the kind art is divided into the two forms productive art and acquisitive art 7 because the forms productive art and acquisitive art are (1) immediately subordinate to the kind art (since they are both subordinate to the kind art and neither of them is subordinate to some kind that is subordinate to the kind art), (2) disjoint (since nothing falls under both the form productive art and the form acquisitive art), and (3) exhaustive of the kind art (since everything that falls under the kind art falls under either the form productive art or the form acquisitive art). If one is trying to define a form F, one begins by identifying a more general kind G to which F is subordinate and then divides this more general kind G into a number of forms K 1, …, K n (n ≥ 2). One then examines how the form F to be defined is related to the forms K1, …, Kn into which G has been divided and one checks whether F is identical with any of K1, …, Kn: if F is identical with one of them,

4 On one occasion (APo. 2.5, 91b12) Aristotle refers to division as ‘h( dia\ tw½n diaire/sewn o(do/j’, on others (APr. 1.31, 46b26; de An. 1.1, 402a18−20) he describes it as a ‘me/qodoj’.

5 According to some commentators (e.g. Trevaskis (1967), 123−8) the method of division is not necessarily concerned only with kinds or forms.

6 Commentators have traditionally interpreted Sph. 254D1−254E3 as describing the method of division: cf. e.g. Cornford (1935), 266−8; Lauer (1966), 155. Some scholars (e.g. Trevaskis (1967), 120−3; Gómez-Lobo (1977), 36−47; Bordt (1991), 523−4) have however recently denied that this passage is concerned with the method of division.

7 I sometimes refer to a form (or a kind, idea, or species) by writing ‘the form’ (or ‘the kind’, ‘the idea’, or ‘the species’) followed by an inscription which is an italicized version of those which in other contexts normally introduce that form (e.g. ‘the form beautiful ’, ‘the kind beautiful ’, ‘the idea beautiful ’, ‘the species beautiful ’). On other occasions I use an inscription of an abstract noun (e.g. ‘beauty’).

6

the process comes to an end; if instead F is identical with none of K 1, …, K n, it will be subordinate to one of them, K i. One divides this kind K i to which F is subordinate and one continues in the same way until a kind is reached with which F, the form to be defined, is identical. The definition of F is finally obtained by tracing down the series of kinds to which the kind F to be defined was found to be subordinate in the process of division.

The central notion of this rough characterization of the method of division is that of subordination. I shall not attempt to offer a precise definition of what it is for a form to be subordinate to a form. I restrict myself to pointing out that subordination does not boil down to proper extensional inclusion: it is not the case that for all forms F and K, F is subordinate to K just in case K holds of everything of which F holds and other things besides. Subordination must be stronger than proper extensional inclusion: although for all forms F and K, F is subordinate to K only if K holds of everything of which F holds and other things besides, the converse fails. In order for a form F to be subordinate to a form K, a further condition whose satisfaction seems to be required is that K may be truly and appropriately mentioned in answering the question ‘What is it?’ asked with regard to F.

The method of division as dialectic. The method of division gains progressively more importance in Plato’s late philosophy. Something comparable to the method of division may be found in the Gorgias (454D1−455A2 and 463E3−466A6)8 and the Republic (V 454A1−9). The method is precisely described for the first time in the Phaedrus (265C8−266C1, cf. 271C10−271D5 and 277B5−8), where it is identified with dialectic (cf. Sph. 253B9−254B7; Phlb. 16E4−17A5). It is used extensively in the Sophist, the Statesman, and the Philebus. Here is its presentation in the Phaedrus:

SOCR. Everything else in it [sc. in our discussion of love] really appears to me to have been said in play. But it would be quite wonderful to grasp by means of an art the power of two kinds among those that were mentioned with Fortune’s guidance.

265C 265D

PHAEDR. Which ones? SOCR. The first consists in seeing together things that are scattered about everywhere and collecting

them into one kind, so that by defining each thing we can make clear the subject of any instruction we wish to give. Just so with our discussion of love: whether its definition was or was not correct, at least it allowed the speech to proceed clearly and consistently with itself.

PHAEDR. And what is the other kind you are talking about, Socrates? SOCR. This, in turn, is to be able to cut up each kind according to its species along its natural joints

and to try not to splinter any part, as a bad butcher might do. In just this way, our two speeches placed all mental derangements into one common kind. Then, just as each single body has parts that naturally come in pairs of the same name (one of them being called the right-hand and the other the left-hand one), so the speeches, having considered unsoundness of mind to be by nature one single kind within us, proceeded to cut it up—the first speech cut its left-hand part and continued to cut until it discovered among these parts a sort of love that may be called ‘left-handed’, which it correctly denounced; the second speech, in turn, led us to the right-hand part of madness, it discovered a love that shares its name with the other but is actually divine, set it out before us and praised it as the cause of our greatest goods.

265E

266A

266B PHAEDR. You are absolutely right.

8 Cf. Moravcsik (1973a), 158−9.

7

SOCR. Well, Phaedrus, I am myself a lover of these divisions and collections, so that I may be able to think and speak; and if I believe that someone else is capable of discerning a single thing that is also by nature capable of encompassing many, I follow ‘straight behind, in his tracks, as if he were a god’.9 God knows whether this is the right name for those who can do this correctly or not, but so far I have always called them ‘dialecticians’.

266C

Since the method of division obviously deals with kinds or forms, the identification of dialectic with the method of division may be regarded as a sharpening of the Republic’s view that dialectic studies forms (cf. VII 532A1−532B5).

Some characteristics of the method of division: (1) The kinds reached by a division are not arbitrarily chosen. One must follow the objective

articulation of kinds and not ‘splinter any part, as a bad butcher might do’ (Phdr. 265E1−3, cf. Plt. 259D10−11; 287C3−4).10

(2) In the Statesman (262A3−264B6) Plato says that in divisions one should avoid separating off a small species of a genus by contrasting it with the remainder of that genus. For instance, one should avoid dividing the genus animal in such a way as to obtain the form man contrasted with beast. When one divides like this, the remainder of the isolation of the small species within the genus (as beast is the remainder of the isolation of the small species man within the genus animal ) is only a ‘part’ of the genus, not also a ‘species’ of it. One should always divide genera in such a way as to obtain ‘parts’ that are also ‘species’ of them. Every species of a genus is a part of it but not every part of a genus is a species of it. This requirement is probably connected with the one concerning division’s following the objective articulation of kinds.11

(3) One and the same genus can be divided in more than one way. For instance, at the beginning of the Sophist (219A8−219C9) the ES and Theaetetus divide the genus art into the two species of production and acquisition. But at the beginning of the Statesman (258B7−258E7) the ES says that ‘it is not in the same place’ that he can ‘see a cut’ in that genus: this time the inquirers divide the genus knowledge into the two species of practical knowledge and purely cognitive knowledge (in the Statesman the nouns ‘art’ and ‘knowledge’ are interchangeable).12

(4) Sometimes the division leads to more than two subordinate species. What counts is that the number be the smallest possible and finite (cf. Plt. 287C3−5; Phlb. 16C10−16E2).13

(5) The method of division can serve several purposes. (5.1) It can be deployed to define a kind. (5.2) It can be used to produce a classification. The most important difference between the use of the method of division for definition and its application for the sake of classification is that while in the pursuit of a definition the branches to which the kind to be defined is not

9 Hom. Od. II 406.

10 Cf. Philip (1966), 346.

11 Cf. Trevaskis (1967), 126; Cavini (1995), 131.

12 Cf. Cavini (1995), 131; Natali (1995), 140.

13 Cf. Philip (1966), 346−7; Notomi (1999), 77.

8

subordinate are ordinarily mentioned without being subjected to division, in a classification all branches are divided. The use of the method of division to produce classifications need not be regarded as a philosophically uninteresting exercise in the construction of kind-ladders after the manner of Linnaeus. Classifications may illuminate the conceptual structure of certain subject areas and such an illumination may turn out not to be philosophically futile (consider for instance a classification which maps out the various kinds of change or of enjoyment).14 (5.3) The method of division can also be used to disambiguate words to which different speakers attribute different meanings (e.g. ‘just’, ‘good’, ‘love’, and ‘sophist’, cf. Phdr. 263A2−263D1; Sph. 218C1−5).15 (5.4) The method of division can finally be practised simply as an exercise which will improve one’s general powers of philosophical discussion (cf. Phdr. 266B3−7; Plt. 285C8−285D8).16

(6) Division goes on until kinds are reached that count as indivisible (cf. Phdr. 277B7−8; Sph. 229D5−6). If division is deployed in order to attain a definition, then there is only one kind that counts as indivisible, namely the one whose definition was sought. If division is instead used for classificatory purposes, then at least two kinds count as indivisible, namely those which encompass individuals but no subordinate kinds.

(7) The dialectician will not care if some of the branches of the division are somewhat bizarre. Dialectic ignores the emotional reactions one might have in front of the kinds encountered (cf. Sph. 226E8−227B6).

(8) The dialectician is allowed a cavalier attitude with names: if a branch singled out by the division is nameless, the dialectician may either leave it nameless (cf. Sph. 220A1−3; 225B13−225C6; 226D5; 267A10−267B2) or introduce a neologism (cf. Sph. 220C7−8; 222C9−222D2; 223D6−9; 224B4−224C3; 225A9−12; 267D4−267E3; Plt. 261E1−7).17

(9) Plato sometimes associates the procedure of division with the reverse procedure of collection (cf. Phdr. 265C8−266C1). While division starts from a single general kind and reaches two or more subordinate kinds, collection starts from many specific instances (which may be either individuals or kinds) and reaches a single general kind under which they fall. One might expect collection to be carried out before division takes place. For, in order to begin dividing a general kind into two or more subordinate kinds, one must identify the general kind to be divided and this identification might be expected to be based on a collection. However, none of the examples of division offered by Plato begins with a preliminary collection: in all of these examples, the general kind from which the division starts seems to be identified by something like an immediate intuition.18 What instead we do find in Plato’s practice is collection being deployed (not before, but) during a division. For, when one is looking for two or more kinds into which a

14 Cf. Cornford (1935), 171; Ackrill (1970), 104−8.

15 Cf. Trevaskis (1967), 128−9.

16 Cf. Trevaskis (1967), 129.

17 Cf. Notomi (1999), 75−6.

18 Cf. Philip (1966), 338−41.

9

general kind F is to be divided, one often collects many specific instances of F and thereby identifies the kinds into which F is to be divided (cf. Sph. 219A10−219C1; 219C2−9; 222C3−222D2; 226B2−226C9; 226E5−227A10; 267A10−267B2; Plt. 258C3−258E7).19

(10) The method of division was one of the most successful aspects of Plato’s late philosophy. (10.1) Speusippus wrote an extensive work on Similar Things (cf. Ath. Deipnosoph. 2.58; 2.78; 3.32; 3.65; 4.10 etc.). The work is now lost, but the few available fragments suggest that it was a classification of plants and animals based on collection and division. Moreover, Diogenes Laertius (IV 5) mentions titles of lost works by Speusippus that dealt with division.20 (10.2) In his Analytics (APr. 1.31, 46a31−46b37; APo. 2.5, 91b12−92a5) Aristotle criticizes Plato’s method of division: he claims that it fails to provide a method of valid argument to establish conclusions from premisses and that it provides no clue as to how one should separate the traits that are essential to the kind that is being defined from those that are merely accidental. Aristotle’s criticism presupposes that he regards the method of division as one of the most important aspects of Plato’s philosophy and as the only competitor to his own great logical discovery, the theory of deduction. Also note that Aristotle himself extensively discusses and applies the method of division in his own biological works and that he enriches it with technical distinctions and terminology (e.g. the idea that a genus joins up with a differentia to constitute a species). (10.3) In a fragment from an unidentified play by the comic poet Epicrates (handed down by Ath. Deipnosoph. 2. 59C−59F), Plato, Speusippus, Menedemus, and their disciples in the Academy are described as intent on applying the method of division to define a gourd. Epicrates uses the appropriate technical language in describing the scene. The fact that the passage comes from a comedy presupposes that laymen outside the Academy would know about the method of division and regard it as one of the most remarkable aspects of Plato’s teaching.

The first six definitions of the sophist based on the method of division. Before attempting to apply the method of division to define the sophist, the ES and Theaetetus give an example of the method by using it in defining the kind angling (218C5−221C5). Using this example as a model, they proceed to apply the method of division to define the sophist (221C6−232A7). Such an application in fact yields not one, but six different definitions of the sophist: in the first definition (221C6−223B7), the sophist is characterized as a hunter of rich and prominent young men; in the second definition (223C1−224D3), he is characterized as a merchant of words and learning who buys his goods in one city and sells them in another; in the third and the fourth definition (224D4−224E5), he is characterized as a merchant of words and learning who operates within a single city and either buys or produces the goods he sells (the difference between the two definitions depends on the distinction between buying and producing the goods that are sold); in the fifth definition (224E6−226A5), the sophist is characterized as a verbal fighter; finally, in the

19 Cf. Hackforth (1945b), 142−3; Philip (1966), 341−2; Moravcsik (1973a), 167, 170−1; Notomi (1999), 75.

20 Cf. Philip (1966), 336−7.

10

sixth definition (226A6−231B9), he is characterized as an educator who by means of refutation purifies the soul from its mistaken belief of having knowledge.21

Faced with these six definitions, Theaetetus says ‘I am puzzled [a)porw½]’ (231B9). His puzzlement is ‘due to the fact that the sophist has appeared in many ways’ (231B9−231C1) and it concerns what one could truly and confidently say a sophist really is, i.e. what the essence of the sophist is (231C1−2). Theaetetus’ puzzlement is probably due to the fact that each of the six definitions of the sophist gave him the impression of capturing the essence of the sophist, while he also accepts the reasonable view that the sophist has only one essence. After summarizing these six different definitions (231C9−231E7), the ES and Theaetetus agree that the sophist appears to have many arts or skills and that whoever is taken in by this appearance will find it difficult to spot the single aspect of the sophist’s art in which all these skills converge (232A1−7). A fresh start is needed.

Images, likenesses, and apparitions. The six definitions of the sophist made it clear that the sophist appears to have many arts or skills. Precisely the point that the sophist appears to have certain arts or skills provides the starting point for the new division and definition of the sophist, a definition that turns upon the concept of appearance: the essence of the sophist is exactly the appearance of having certain arts and skills which he in fact lacks.

Before attempting the new, definitive division and definition of the sophist, the ES and Theaetetus embark on a discussion (232B1−236D4) of the sophist’s art that is supposed to provide the background for this division and definition.

The ES begins this discussion by taking up some points that had emerged in the course of the earlier six definitions: the sophist engages in disputes and teaches others to do so (232B1−10). This description raises the question of the range of topics which the sophist claims to make others able to engage in disputes about. The simple answer is that it extends over all topics: divine things that cannot be perceived, perceptible objects both in the heavens and on earth, problems of being and becoming, issues of law and politics, and questions concerning the crafts (232B11−232E5).

In the next step of their discussion (232E6−233D2), the ES and Theaetetus bring in the idea that the sophist has apparent knowledge. Since nobody can know everything, the sophist cannot know about all the topics he engages in disputes about. On the other hand, the sophist brings young people to believe that he does have knowledge of the topics he engages in disputes about: otherwise they would not pay him a fee to be taught. Therefore the sophist has apparent knowledge about all things (although the single expression ‘docastikh\ e)pisth/mh’, 233C10, conveys the idea not only of apparent knowledge, but also of knowledge of appearance).

In the last section of the discussion (233D3−236D4), the ES introduces a model (a para/deigma, 233D3) in order to clarify matters.22 This model institutes an analogy between the

21 The sophist as characterized in the sixth definition looks very much like a philosopher. Recall that at the beginning of the dialogue (216D1) Plato hinted that philosophers sometimes take on the appearance of sophists.

22 This is the second time that the concept of a model is invoked to ease the progress of the inquiry: the earlier model had been that of the definition by division of the angler (cf. 218D9 and 221C5).

11

skill of producing plastic imitations (sculptures) and the sophist’s art. A plastic imitator (a sculptor) can produce plastic imitations (sculptures) of all objects and he can lead ‘those young children who are silly’ (234B8) to think that his plastic imitations are the real objects they imitate and that he is able to produce all objects (i.e. that he is a sort of god);23 analogously, a linguistic imitator (i.e. the sophist) produces linguistic imitations (i.e. false statements) of true statements and he can lead young people ‘who stand even farther away from the truth of things’ (234C4) to think that his linguistic imitations are the true statements they imitate and that he is able to produce all true statements (i.e. that he is wise) (234B5−234D1). Note that the concept of imitation is here used to describe the relationship obtaining between false statements and true statements: false statements are imitations of true statements. Precisely because they are imitations, false statements may induce those who hear them to confuse them with the true statements they imitate.

Let me introduce some conceptual tools that will help to clarify Plato’s conception. Every imitation is an imitation of something and with any imitation of anything there is associated a ‘cognate deception’ which it can induce people to fall for, i.e. the mistake of taking the imitation to be what it is an imitation of (e.g. taking an imitation of Theaetetus to be Theaetetus, or taking an imitation of an apple to be an apple). There are two kinds of imitations, and correspondingly two kinds of cognate deceptions they can induce. On the one hand, there are ‘singular imitations’, like a portrait of Theaetetus. In the case of a singular imitation, the correct answer to the question ‘What is it an imitation of ?’ will be a singular term (e.g. ‘Theaetetus’). In the case of a singular imitation, the cognate deception is a misidentification: if one erroneously takes the imitation to be what it is an imitation of, one misidentifies the imitation with what it is an imitation of (e.g. one misidentifies a waxen statue of Theaetetus with Theaetetus himself ). On the other hand, there are ‘general imitations’, like a waxen apple. In the case of a general imitation, the correct answer to the question ‘What is it an imitation of ?’ will be a general term (e.g. ‘an apple’). In the case of a general imitation, the cognate deception is a misdescription: if one erroneously takes the imitation to be what it is an imitation of, one misdescribes the imitation as being of the kind which would be introduced by a correct answer to the question ‘What is it an imitation of ?’ (e.g. one misdescribes a waxen apple as being itself an apple).

When false statements are regarded as imitations of true statements, what appears to be playing a role is the concept of a general imitation: what I do when I produce a false statement that imitates true statements and thereby induce you to confuse my false statement with a true one is relevantly similar to what I do when I produce a waxen object that looks like an apple and then induce you to believe that it is an apple (it is more similar to this than to what happens when I produce a waxen statue of Theaetetus and then induce you to believe that it is Theaetetus). For, if I produce a false statement and you believe me, your confusion does not amount to thinking that my false statement is numerically identical with a certain true statement from which it is as a matter of fact numerically distinct, but to thinking that my false statement is true whereas it is not.

23 Cf. R. X 596B10−596E11; 598B8−598C4.

12

Plato here is to some extent reducing the concept of ‘propositional falsehood’ (the concept of falsehood according to which what may properly be said to be false is a statement or a belief or a proposition) to that of ‘ontological falsehood’ (the concept of falsehood according to which anything may properly be said to be a false thus and so in so far as it is a fake or defective thus and so): a false statement is a fake true statement, a false true-statement. This is the opposite of what tends to be done by modern philosophers, who prefer to reduce ontological falsehood to propositional falsehood.24 Plato’s reduction is only partial because the concepts of propositional truth and falsehood are presupposed: a false statement is a falsep statement that is a falseo truep statement (where ‘truep’ and ‘falsep’ abbreviate ‘true in the propositional sense’ and ‘false in the propositional sense’, while ‘trueo’ and ‘falseo’ abbreviate ‘true in the ontological sense’ and ‘false in the ontological sense’).

Having introduced the conceptual apparatus of imitation, the ES and Theaetetus draw some distinctions that are intended to be relevant to the division leading to the definition of the sophist (235B8−236D4). The sophist’s art is an imitative art, one that produces imitations or images (‘imitation’, ‘mi¿mhma’, and ‘image’, ‘eiãdwlon’, are treated as synonyms in this context).25 There are two types of images: a likeness (ei¹kw/n) is like (eÃoike) its model; an apparition (fa/ntasma) appears (fai¿netai) to be like its model, but is unlike it.26 In the case of plastic images, i.e. sculptures, likenesses are like their models in that they faithfully reproduce their proportions and colors (ancient sculptures were colored) whereas apparitions are unlike their models in that they distort their proportions and colors, but appear to be like their models by appearing to be faithful reproductions of them (this happens especially with monumentally large sculptures that are supposed to be viewed from a low vantage point). Accordingly, there are two kinds of imitative art: one produces likenesses, the other produces apparitions (235C9−236C8).

At this point, the ES and Theaetetus interrupt their account: they have doubts as to which of these two arts the sophist is to be subsumed under, they lament the obscurity of the subject and they agree that the sophist has found a very confusing hiding-place (236C9−236D4). Their difficulty has to do with appearance and falsehood: the point of the dialogue has been reached where the central ‘digression’ (236D5−264B10) on not being, being and falsehood begins. Of course, one is inclined to suppose that the ES and Theaetetus intend to subsume the sophist’s art under the skill that produces apparitions and this hypothesis is confirmed by some incidental remarks later in the central ‘digression’ (cf. 239C9−239D1; 260D8−9) and by the final division and definition of the sophist (cf. 266D9−267A1; 268C8−268D5).

24 Cf. A. R. White (1970), 5.

25 Cf. Bondeson (1972), 1. The word ‘image’ occurs for the first time in the Sophist at 234C5, in the context of the description of the activity of the sophist. At the corresponding point of the description of the activity of the plastic imitator there is the word ‘imitation’ (234B6).

26 In one passage (240A4−240B12) Plato seems to use ‘image’ (‘eiãdwlon’, 240A5, A7) and ‘likeness’ (‘ei¹kw/n’, 240B11, B12) as if they were synonyms (cf. Robinson (2001), 445). He is probably being careless.

13

We would expect an explanation of what corresponds on the linguistic level to the distinction, drawn on the plastic level, between likenesses and apparitions. But we are not told. We are left to speculate, and the most plausible speculation is the following.

In the case of verbal images, i.e. statements, likenesses are like their models in that they are true statements whereas apparitions are unlike their models in that they are false statements, but appear to be like their models by appearing to be true statements. The sophist is a producer of verbal images, in particular of verbal apparitions, i.e. of false statements that appear to be true. Since the sophist’s statements are false while appearing to be true, they induce those who hear them to believe them to be true. Thus, the sophist makes false statements and induces false beliefs. For instance, the sophist makes the statement ‘Justice is the interest of the stronger’, which is false but appears to be true. Since it appears to be true, it induces those who hear it to believe it to be true. Note that, properly speaking, the false belief induced in the audience concerns the false statement itself: it is the false belief that the statement ‘Justice is the interest of the stronger’ is a true statement, not the false belief that justice is the interest of the stronger. Of course, whoever has the false belief that the statement ‘Justice is the interest of the stronger’ is a true statement is committed to having the further false belief that justice is the interest of the stronger.

The analogy between plastic images and verbal images is brought out by the following diagrams: likenesses true statements

faithfully faithfully imitate imitate

models imitate referents imitate unfaithfully unfaithfully imitate imitate

apparitions false statements

Here are a few remarks on the analogy and the diagrams. (1) The concept of imitation is deployed three times, and on two dimensions. It is employed twice on the ‘horizontal dimension’ to describe the relationship between a

(plastic or linguistic) imitation and its model. Three points are worth noting with regard to the imitations on the ‘horizontal dimension’. (α) The kind of imitation that seems involved in the ‘horizontal dimension’ is singular imitation: just as the statements which will be studied at the end of the Sophist will be true and false singular statements about Theaetetus, so the plastic imitations that are considered here are likely to be singular imitations (e.g. statues of Theaetetus, which can be evaluated on the basis of how faithful they are to Theaetetus, rather than statues of men, which can be evaluated on the basis of how faithful they are as reproductions of men). (β) The distinction between faithful and unfaithful images, i.e. likenesses and apparitions, plays a role on the ‘horizontal dimension’: some monumental statues of men faithfully reproduce the proportions of the men they are reproductions of, others reproduce them unfaithfully; similarly,

14

some statements (i.e. true statements) are faithful to their referents in that they describe them as they are, other statements (i.e. false statements) are unfaithful to their referents in that they describe them as being in ways they are not. (γ) The cognate deception does not seem to play any role in the case of the imitations on the ‘horizontal dimension’: there is no suggestion that someone producing a (faithful or unfaithful) monumental statue of Theaetetus intends to deceive the viewers into believing that the statue itself is Theaetetus; similarly, there is no suggestion that someone producing a (true or false) statement about Theaetetus intends to deceive the viewers into believing that the statement itself is Theaetetus.

The concept of imitation is applied a third time on the ‘vertical dimension’ to describe the relationship between a (plastic or linguistic) apparition and its model, which is a (plastic or linguistic) likeness. In the case of the concept of imitation deployed on the ‘vertical dimension’ exactly the opposite happens with respect to that deployed twice on the ‘horizontal dimension’, correspondingly to each of the three points mentioned in the last paragraph. (α) The kind of imitation that seems involved in the ‘vertical dimension’ is general imitation. If I produce an unfaithful monumental statue of Theaetetus that is an imitation of a faithful monumental statue of him, then the correct answer to the question ‘What is it an imitation of ?’ will be ‘a faithful monumental statue’ or ‘a faithful monumental statue of Theaetetus’ (in any case, a general term). Similarly, if I produce an unfaithful (false) statement about Theaetetus that is an imitation of a faithful (true) statement about him, then the correct answer to the question ‘What is it an imitation of ?’ will be ‘a faithful (true) statement’ or ‘a faithful (true) statement about Theaetetus’ (in any case, a general term). (β) The distinction between faithful and unfaithful imitations plays no role on the vertical dimension: no distinction is drawn between the case of an unfaithful imitation faithfully imitating a faithful imitation and the case of an unfaithful imitation unfaithfully imitating a faithful imitation.27 (γ) The cognate deception does play a role on the vertical dimension: someone producing a monumental statue of Theaetetus which as a matter of fact is unfaithful to him (because it does not accurately reproduce the proportions of Theaetetus) does deceive the viewers into thinking that the monumental statue is faithful (that it does accurately reproduce the proportions of Theaetetus); similarly, someone producing a statement about Theaetetus which as a matter of fact is unfaithful to him (because it describes Theaetetus either as being in a way in which he as a matter of fact is not or as not being in a way in which he as a matter of fact is) does deceive the viewers into thinking that the statement is faithful (that it does describe Theaetetus either as being as he is or as not being in a way in which he as a matter of fact fails to be).

(2) Plato draws an analogy between the faithfulness of an image and the truth of a statement.28 This commits him to a version of a correspondence theory of truth, one that regards the truth of a statement as its being faithful to a model (its referent). As I pointed out earlier, in the initial stage of the analogy between plastic and linguistic imitation the concepts of truth and falsehood were left unanalyzed: all that was said was that false statements are imitations of true statements

27 Cf. Szaif (1998), 402.

28 Cf. Cra. 430A8−431C3; Bondeson (1972), 1, 2.

15

and tend to be regarded as true. Now, in the later stage of the analogy, the concept of imitation is deployed again in order to provide a model for explaining truth and falsehood. This analogy between faithfulness and truth (and, implicitly, between unfaithfulness and falsehood) is however problematic in that faithfulness (and unfaithfulness) admit of degrees whereas truth (and falsehood) do not: a statue can be more or less faithful (or unfaithful) to its model, but a statement cannot be more or less true (or false).29

(3) Later in the dialogue, at 239C9−240C6, Plato presents a paradox concerning images in general (including likenesses as well as apparitions). This paradox probably attaches to the imitation in the ‘vertical dimension’, that whereby apparitions imitate likenesses and tend to be confused with them. For, as I pointed out earlier, the distinction between faithful and unfaithful imitations plays no role on the vertical dimension.

(4) One might be inclined to think that false statements that appear to be true (verbal apparitions) constitute a proper subset of false statements, namely those false statements that are deceptive. On reflection, however, this inclination ought to be resisted: every statement, by virtue of its intrinsic assertoric force, presents itself as true and to that extent appears to be true. False statements therefore coincide with deceptive statements.

(5) As I pointed out earlier, Plato deploys three times the concept of imitation in his analysis of false statements. In fact, in a sense he deploys it four times: for he describes the whole discussion involving the analogy between the skill of producing plastic imitations (sculptures) and the sophist’s art as based on a ‘model’ (‘para/deigma’, 233D3). Since the concept of model is intimately connected with that of imitation (cf. 235D7), Plato is perhaps hinting that the concept of imitation could be used to describe the status of the discussion itself.

The final division and definition of the sophist. The discussion of the sophist’s art issues in the division illustrated by the following diagram:

29 Cf. Bondeson (1972), 2−3.

16

artless with an art

with an acquisitive art with a productive art

divine human

productive productive productive productive of originals of images of originals of images

productive productive of likenesses of apparitions

by means by means of of tools the artist himself

with knowledge without knowledge of the model of the model

without awareness with awareness of the ignorance of the ignorance

by long by private public speeches discussions involving questions and answers

This division is carried out in the second part of the frame section of the dialogue (264B11−268C4). Its summing up leads to the final definition of the sophist (268C5−268D5). The graphic description of the method of division as in the diagram above was probably already used in the Academy: for Aristotle (PA 1.2, 642b12) speaks of certain ‘drawn divisions’ and Plato himself (Sph. 264E1−2, cf. Phdr. 265E4−266B1) speaks of taking always the right-hand branch of a division.

CHAPTER II PUZZLES CONCERNING NOT BEING

The structure of the section dealing with the difficulty of ‘saying what is not’. At the end of the discussion that prepares the final division and definition of the sophist, the ES notes that the sophist has hidden himself in a class that baffles investigation. He has in mind the problems connected with appearing without being and saying something that is not true. He thus introduces a discussion about the difficulty of saying what is not, the paradox of images, and the falsehood paradox (236D5−242B5).

This section of the dialogue may be divided into five subsections: (1) an introduction to the difficulties that bedevil this area of thought (236D5−237B7); (2) three arguments in support of the thesis that it is impossible to ‘say what is not’ (237B7−239C8); (3) an argument for the claim that the concept of image implies a contradiction (239C9−240C6); (4) an argument for the claim that it is impossible to believe or state what is false (240C8−241B4); (5) a summary of the difficulties already encountered and a sketch of the moves that will enable us to overcome them (241B4−242B5).

The introduction to the section dealing with the difficulty of ‘saying what is not’. The core of the introduction to the difficulties raised by the concept of falsehood (236D9−237B7) is the following passage:

ES ãOntwj, wÕ maka/rie, e)sme\n e)n panta/pasi xalepv= ske/yei. to\ ga\r fai¿nesqai tou=to kaiì to\ dokeiÍn, eiånai de\ mh/, kaiì to\ le/gein me\n aÃtta, a)lhqh= de\ mh/, pa/nta tau=ta/ e)sti

236D9 236E1

mesta\ a)pori¿aj a)eiì e)n t%½ pro/sqen xro/n% kaiì nu=n. oÀpwj ga\r ei¹po/nta xrh\ yeudh= le/gein hÄ doca/zein <fa/nai> oÃntwj eiånai, kaiì tou=to fqegca/menon e)nantiologi¿# mh\ sune/xesqai, panta/pasin, wÕ Qeai¿thte, xalepo/n.

E5 237A1

THT. Ti¿ dh/; ES Teto/lmhken o( lo/goj ouÂtoj u(poqe/sqai to\ mh\ oÄn

eiånai: yeu=doj ga\r ou)k aÄn aÃllwj e)gi¿gneto oÃn. Parmeni¿dhj de\ o( me/gaj, wÕ paiÍ, paisiìn h(miÍn ouÅsin a)rxo/meno/j te kaiì dia\

A5 te/louj tou=to a)pemartu/rato, pezv= te wÒde e(ka/stote le/gwn kaiì meta\ me/trwn—

Ou) ga\r mh/ pote tou=to damv=, fhsi¿n, eiånai mh\ e)o/nta: a)lla\ su\ th=sd' a)f' o(dou= dizh/menoj eiårge no/hma.

par' e)kei¿nou te ouÅn martureiÍtai, kaiì ma/lista/ ge dh\ pa/ntwn o( lo/goj au)to\j aÄn dhlw¯seie me/tria basanisqei¿j.

237B1 B2

ES We are really faced, my dear friend, with an extremely difficult inquiry. For this appearing and

seeming, but not being, and saying things, but not true ones, all of this has always been in the past and still now is full of difficulty. For, Theaetetus, it is extremely difficult to see by speaking in what way one should affirm that saying or believing falsehoods really exists, and not to be caught in a contradiction in uttering this.

236D 236E

237A

18

THT. Why? ES This account dared to assume that what is not is: for in no other way would falsehood have

turned out to be. But, my boy, when we were of your age, the great Parmenides from beginning to end testified this, constantly saying so in prose as well as in verse:

Never shall this be proved, that things which are not are, but do thou, in thy inquiry, hold back thy thought from this way.

So we have his testimony and the account itself would reveal this more than anything else if it were appropriately examined.

237B

The above translation is based on David Robinson’s text, which integrates ‘fa/nai’ at 236E4 (the emendation probably goes back to Heindorf ).1 Translations are possible of the text that does not incorporate this integration. One is:

For, Theaetetus, it is extremely difficult to see how it is appropriate that someone who claims that saying or believing falsehoods really exists, in uttering just this should avoid being caught in a contradiction.2

Here ‘oÀpwj’ modifies ‘xrh/’, which governs the whole complex infinitive ‘ei¹po/nta … yeudh= le/gein hÄ doca/zein oÃntwj eiånai, kaiì tou=to fqegca/menon e)nantiologi¿# mh\ sune/xesqai’. Within this infinitive, the subject is the phrase ‘ei¹po/nta … yeudh= le/gein hÄ doca/zein oÃntwj eiånai’, with ‘ei¹po/nta’ governing the subordinate infinitive ‘yeudh= le/gein hÄ doca/zein oÃntwj eiånai’ (where ‘yeudh= le/gein hÄ doca/zein’ is the subject, ‘oÃntwj eiånai’ the predicate); the predicate is ‘e)nantiologi¿# mh\ sune/xesqai’; and the phrase ‘kaiì tou=to fqegca/menon’ is in apposition to the subject (with ‘kai¿’ meaning ‘just’).3

Another translation of the text that does not incorporate Heindorf ’s integration is:

For, Theaetetus, it is extremely difficult to see by speaking in what way one should say or believe that falsehoods really exist, and not to be caught in a contradiction in uttering this.4

Here ‘oÀpwj’ modifies ‘ei¹po/nta’ and ‘xrh/’ governs two infinitives, ‘yeudh= le/gein hÄ doca/zein oÃntwj eiånai’ and ‘tou=to fqegca/menon e)nantiologi¿# mh\ sune/xesqai’. In the first of these infinitives, ‘le/gein hÄ doca/zein’ takes ‘yeudh= … oÃntwj eiånai’ as its object.5

In all translations, the words ‘to see’ (immediately after ‘it is extremely difficult’, ‘panta/pasin … xalepo/n’ at 237A1) are supplied from the context.6 The contorted syntax of the Greek is probably supposed to mirror the difficulty of the topic we are about to address.

1 Cf. Robinson (1999), 147.

2 Cf. Heindorf (1810), 346; Stallbaum (1840), 129; Wagner (1856), 69; Apelt (1897), 113.

3 Cf. LSJ s.v. ‘kai¿’ B 6. Madvig (1871), 380 regards the occurrence of ‘kai¿’ between ‘ei¹po/nta’ and ‘tou=to fqegca/menon’ as unacceptable. He therefore deletes it as having crept into the text from the second syllable of the immediately preceding ‘eiånai’. His emendation is adopted by Schanz.

4 Cf. Campbell (1867), Sph. 81; Fraccaroli (1934), 146; Cornford (1935), 200.

5 In view of the unnaturalness of regarding ‘yeudh= … oÃntwj eiånai’ as the object of ‘le/gein hÄ doca/zein’, Szaif (1998), 395 suggests to delete ‘hÄ doca/zein’ as a gloss that entered the text.

19

The interpretations of the introduction. Most commentators think that 236D9−237B2 contains a precise version of the paradox of false belief. Here are the main interpretations they offer.

(1) According to the ontological interpretation, the argument runs as follows: ‘Suppose that falsehood is; then not being is; but this is a contradiction, so falsehood is not’. This argument involves two senses of ‘to be’ (‘eiånai’): the veridical sense (whereby ‘to be’ means ‘to be true’) and the existential sense (whereby ‘to be’ means ‘to exist’, so that the conclusion makes the paradoxical claim that falsehood does not exist). Therefore the argument is fallacious: in particular, the result that not being is fails to be a contradiction because contradictions require that a predicative expression should be affirmed and denied of the same thing in the same sense (cf. R. V 479B10−479C5; Sph. 256A10−12).7

(2) According to the semantical-gnoseological interpretation, the argument runs as follows: ‘Falsehood involves “saying or thinking what is not”; but it is impossible to “say or think what is not”; therefore there is no falsehood’. The semantical-gnoseological interpretation takes different forms with different commentators, so that it would be more correct to speak of a family of semantical-gnoseological interpretations. Two main branches of this family may be distinguished.

(2.1) According to the indefinability interpretation, the paradox denies the possibility of describing or defining falsehood: ‘To describe or define falsehood, one must speak of what is not; but it is impossible to speak of what is not; so it is impossible to describe or define falsehood’.

(2.2) According to the inexistence interpretation, the paradox denies the existence of falsehood. Two main versions of this interpretation may be distinguished.

(2.2.1) According to the interpretation of propositions, the argument runs as follows: ‘Every statement must concern some object that constitutes its content (the object that constitutes the content of a statement is a proposition); but a false statement concerns what is not, and so lacks content; therefore there are no false statements’. This paradox depends on the ambiguity of the verb ‘to be’, which may mean both ‘to exist’ and ‘to be true’. Frege expounded a paradox that is rather similar to this one, but does not depend on an ambiguity of ‘to be’: Frege considers the possibility that the being of a proposition (a Gedanke) might depend on its truth, in which case a false sentence would express no proposition.8

(2.2.2) According to the interpretation of the collapse of truth conditions on meaning conditions, the argument runs as follows: ‘If a statement is meaningful then there is a portion of reality corresponding to it; if there is a portion of reality corresponding to a statement then the statement is true; therefore if a statement is meaningful then it is true’.9 Three versions of this

6 Heindorf (1810), 346 suggests ‘diideiÍn’ (cf. Phd. 62B5−6) or ‘ei¹peiÍn’ (cf. Sph. 217B1−2) and draws parallels with Tht. 158C2−3 and Ar. V. 1279. Campbell (1867), Sph. 81 suggests ‘gnw½nai’ or ‘le/gein’.

7 Cf. Cornford (1935), 200; Ross (1951), 114−15; Bluck (1963), 60−1; Bondeson (1972), 1; Guthrie (1978), 135; Szaif (1998), 395−6.

8 Cf. Frege (1918−19), 144. Frege notoriously uses ‘Gedanke’ in a sense very close to that in which many later philosophers use ‘proposition’.

9 Cf. Kostman (1973), 193.

20

interpretation have been, or might be, suggested: they differ according to the ‘category’ to which the portion of reality required by meaning and entailing truth belongs.

(2.2.2.1) According to the first version, the portion of reality that is required by meaningfulness and guarantees truth is a fact (a subsistent state of affairs). Here is an example: in order for ‘Theaetetus is flying’ to be meaningful, it must link up with the fact that Theaetetus is flying and this fact guarantees the truth of ‘Theaetetus is flying’.10 This version of the paradox recalls a puzzle that was studied by the early Russell.11

(2.2.2.2) According to the second version, the portion of reality that is required by meaningfulness and guarantees truth is an existent attribute, described as belonging to the item that constitutes the topic of the statement. Here is an example: in order for ‘Theaetetus is flying’ to be meaningful, it must link up with the existent attribute of Theaetetus’ flying and this existent attribute guarantees the truth of ‘Theaetetus is flying’.12

(2.2.2.3) According to the third version, the portion of reality that is required by meaningfulness and guarantees truth is an existent object, the reference of the statement described as having the attribute assigned to it by the statement. Here is an example: in order for ‘Theaetetus is flying’ to be meaningful, it must link up with the existent object which is the flying Theaetetus and this existent object guarantees the truth of ‘Theaetetus is flying’.13

It is very difficult to choose one of these interpretations of 236D9−237B7: different arguments may be brought in support of different solutions. The simple fact that the choice of one interpretation is so difficult suggests that perhaps no such choice should be made. Since the passage is so vague that it admits all these different interpretations, perhaps it is not supposed to be understood in a precise way: it does not contain a precise version of the paradox, but only a very general introduction to it, in which some important points connected with it are briefly touched upon. This hypothesis is confirmed by the fact that at 236E1−2 Plato briefly mentions the difficulty of seeming without being and the difficulty of saying something that is not true: these paradoxes will be fully expounded later on, at 239C9−240C6 and (respectively) 240C7−241B4. This strengthens the feeling that what we are now having is merely a preliminary summary. I shall therefore assume that 236D9−237B7 is the introduction to a long stretch of argument that extends from 237B7 to 241B4.

The first aporetic argument about not being (237B7−237E7). After the introductory subsection, Plato presents three aporetic arguments in support of the thesis that it is impossible to say what is not (237B7−239C8).

Here is the first aporetic argument in support of the thesis that it is impossible to say what is not (237B7−237E7):

10 Cf. Hackforth (1945a), 57; Gulley (1962), 156; von Weizsäcker (1973), 232−3; Przełęcki (1981), 124.

11 Cf. Russell (1910b), 150−1; Russell (1912), 72.

12 Cf. Owen (1971), 244−5.

13 This third version is mentioned by Owen (1971), 245.

21

ES Kai¿ moi le/ge: to\ mhdamw½j oÄn tolmw½me/n pou fqe/ggesqai;

THT. Pw½j ga\r ouÃ;

237B7

ES Mh\ toi¿nun eÃridoj eÀneka mhde\ paidia=j, a)ll' ei¹ spoudv= de/oi sunnoh/santa/ tina a)pokri¿nasqai tw½n a)kroatw½n poiÍ xrh\ touÃnom' e)pife/rein tou=to, to\ mh\ oÃn, ti¿

B10 237C1

dokou=men aÄn; ei¹j ti¿ kaiì e)piì poiÍon au)to/n te kataxrh/sasqai kaiì t%½ punqanome/n% deiknu/nai;

THT. Xalepo\n hÃrou kaiì sxedo\n ei¹peiÍn oià% ge e)moiì panta/pasin aÃporon.

C5

ES ¹All' ouÅn tou=to/ ge dh=lon, oÀti tw½n oÃntwn eÃpi14 to\ mh\ oÄn ou)k oi¹ste/on.

THT. Pw½j ga\r aÃn;

ES Ou)kou=n e)pei¿per ou)k e)piì to\ oÃn, ou)d' e)piì to\ tiì fe/rwn o)rqw½j aÃn tij fe/roi.

C10

THT. Pw½j dh/; ES Kaiì tou=to h(miÍn pou fanero/n, w¨j kaiì to\ tiì tou=to

r(h=ma e)p' oÃnti le/gomen e(ka/stote: mo/non ga\r au)to\ le/gein, wÐsper gumno\n kaiì a)phrhmwme/non a)po\ tw½n oÃntwn a(pa/ntwn, a)du/naton: hÅ ga/r;

237D1

THT. ¹Adu/naton. D5 ES åAra tv=de skopw½n su/mfhj w¨j a)na/gkh to\n tiì

le/gonta eÀn ge/ ti le/gein; THT. OuÀtwj.

ES ¸Eno\j ga\r dh\ to/ ge tiì fh/seij shmeiÍon eiånai, to\ de\ tine\ duoiÍn, to\ de\ tine\j pollw½n.

THT. Pw½j ga\r ouÃ;

D10

ES To\n de\ dh\ mh\ tiì le/gonta a)nagkaio/taton, wj eÃoike, panta/pasi mhde\n le/gein.

237E1

THT. ¹Anagkaio/taton me\n ouÅn. ES åAr' ouÅn ou)de\ tou=to sugxwrhte/on, to\ to\n toiou=ton

le/gein me/n, le/gein me/ntoi mhde/n, a)ll' ou)de\ le/gein fate/on, oÀj g' aÄn e)pixeirv= mh\ oÄn fqe/ggesqai;

THT. Te/loj gou=n aÄn a)pori¿aj o( lo/goj eÃxoi.

E5

E7 ES And tell me: do we somehow dare to express what in no way is? THT. Why shouldn’t we?

237B

ES So if, not for the sake of contention nor of play, but seriously and thoughtfully, one of the hearers were to answer the question to what this name, ‘not being’, should be applied,15 what would we think? What, and what sort of thing, should he apply it to, and show to the questioner?

237C

THT. You asked a difficult question and perhaps it is completely puzzling for one like me to say.

14 I read ‘tw½n oÃntwn eÃpi’ with the main MSS (cf. Cordero (1986−87), 283; Aubenque (1991), 372; Cordero (1993), 233; Fronterotta (2007), 311). Most eds. print ‘tw½n oÃntwn e)pi¿ <ti>’. For ‘e)pi¿’ suffering anastrophe, cf. LSJ s.v. ‘e)pi¿’ D I. For anastrophe in the Sophist, cf. 218C2; 218C4; 221A7; 222D12; 226C1; 230A8; 230B4; 232C8; 233A10; 238D2; 241B5; 243B9; 243C10; 245E7; 248C2; 249D10 (all cases of ‘pe/ri’ in place of ‘peri¿’). For ‘e)pi¿’ with the genitive indicating the goal of motion, cf. LSJ s.v. ‘e)pi¿’ A I 3 b. One should avoid introducing ‘ti’ here because in the next step (237C10−11) the point is made that ‘what is not’ cannot be applied to ti.

15 Cf. 250D7−8.

22

ES But this is at least clear, that ‘not being’ should not be applied to beings. THT. How could it?

ES So, since it cannot be applied to being, by applying it to any something one would again fail to apply it correctly.

THT. How so?

ES This is also16 somehow clear to us, that we always utter this expression, ‘something’, in connection with a being: for it is impossible to utter it on its own, as if naked and isolated from all beings. Is that not so?

237D

THT. Impossible. ES Do you agree when you look at matters this way, that it is necessary that whoever says

something say at least one something? THT. Yes.

ES For you will say that ‘something’ is a sign of one thing, ‘two things’ of two, and ‘things’ of many.

THT. Of course.

ES But, as it seems, it is then most necessary that whoever does not say something say absolutely nothing.

THT. Most necessarily.

237E

ES This, then, should also not be conceded, that such a person speaks, but says nothing. One should instead say that whoever tries to express what is not does not even speak.

THT. The puzzlement generated by the argument would at least reach its completion.17

Two issues of translation need to be discussed.

Issues of translation: (i) The use of ‘fqe/ggesqai’. The Greek verb ‘fqe/ggesqai’ has several uses. I would like to focus on three. (1) The verb can take as its object a linguistic expression introduced by a single noun-phrase in the accusative (cf. Prt. 342E7; R. V 463E3; Sph. 244A6). When it is used in this way, it is most naturally rendered by ‘to utter’. (2) The verb can take as its object two items, one of which is (in most cases) a non-linguistic object while the other is a linguistic expression, both introduced by noun-phrases in the accusative (cf. Sph. 257D11−12; Phlb. 25C5−6). When it is used in this way the verb is most naturally rendered by ‘to call’. (3) The verb can finally take as its object a propositional content introduced either by a declarative sentence (cf. Lg. II 662B7) or by some other device, like a pronoun in the accusative (cf. Cra. 394C8−9; R. VIII 568A11; Plt. 307A5; Lg. II 662C5; X 901A8). When it is used in this way, the verb is most naturally rendered by ‘to express’ or simply ‘to say’. It is in connection with this third use that the verb is employed in sentences like ‘ta)lhqh= fqe/ggesqai’ (‘to say the truth’) (cf. Phlb. 49B8).18

The verb ‘fqe/ggesqai’ occurs twice in our passage: at the beginning Theaetetus candidly accepts that ‘we somehow dare to fqe/ggesqai what in no way is’ (237B7−8) and at the end the ES tells of the dire fate of ‘whoever tries to fqe/ggesqai what is not’ (237E6). Which of the last

16 Cf. 237C7.

17 A pun is probably intended: the puzzlement reaching its completion can be both its coming to an end and its reaching its culmination.

18 At Cra. 430A1−2 ‘a)lhqh= fqe/ggesqai’ and ‘yeudh= fqe/ggesqai’ are used differently: they mean something like ‘to utter sounds in a way that amounts to speaking truly’ and ‘… falsely’.

23

paragraph’s uses of ‘fqe/ggesqai’, if any, is in play? Some translators19 opt for the first use: in their view, what Theaetetus concedes at the beginning is that we dare to utter the phrase ‘what in no way is’ and at the end we learn of what happens to whoever tries to utter the phrase ‘what is not’. This solution is however rather unlikely, in view of several considerations. First, this solution has the consequence that the argument’s conclusion should be that whoever tries to utter the phrase ‘what is not’ fails to carry out an act of saying. But the earlier steps of the argument cannot be easily seen to provide support for this conclusion. Things go back into place if we take the argument’s conclusion to be that whoever tries to express what is not fails to carry out an act of saying (whereby ‘fqe/ggesqai’ is understood in accordance with the third of the last paragraph’s uses). Secondly, the second aporetic argument about not being (238A1−238C12) addresses the issue of the possibility of ‘fqe/ggesqai through the mouth’ and ‘grasp in thought’ what is not (238B6−7) and reaches the conclusion that ‘one cannot correctly fqe/ggesqai nor say nor think of what is not in its own right’ (238C8−9). Given that what is not is here something which one (hypothetically) both fqe/ggetai and thinks of, it is extremely unlikely that fqe/ggesqai what is not could be to utter the phrase ‘what is not’: Plato is surely not raising the question of the possibility of thinking of the phrase ‘what is not’. Thirdly, the solution we are looking at has the consequence that the argument begins by considering someone uttering ‘what in no way is’ and ends speaking about what happens to someone uttering ‘what is not’, which is a different linguistic expression. Of course, it might be retorted that ‘what in no way is’ and ‘what is not’ have the same meaning. Nevertheless, if we take seriously the idea that the argument concentrates on someone producing certain utterances, there is something inelegant in allowing a change of linguistic expression in the course of the argument. On the basis of these reflections I opt for understanding the two occurrences of ‘fqe/ggesqai’ in our passage in accordance with the third of the uses of the verb distinguished in the last paragraph. Accordingly, I translate these occurrences with forms of ‘to express’,20 and I take it that ‘fqe/ggesqai what in no way is’ and ‘fqe/ggesqai what is not’ is ‘to express a propositional content which as a matter of fact is not’. The issue of expressing a propositional content which as a matter of fact is not is obviously relevant to the problem of the possibility of speaking falsely. Why does Plato choose just this verb, ‘fqe/ggesqai’? Perhaps because he wants to concentrate on a version of the paradox that concerns audible utterances of false sentences rather than private false thoughts (cf. 238B6−8).

Issues of translation: (ii) The sentence at 237D6−7. The interrogative sentence ‘aÅra tv=de skopw½n su/mfhj w¨j …;’ (237D6−7) can be construed in two ways. On the first construal, the declarative sentence introduced by ‘w¨j’ is governed by ‘su/mfhj’. The corresponding translation is: ‘In

19 Cf. MacKay (1868), 111; Jowett (1892), IV 365; Fowler (1921), 339; Fraccaroli (1934), 147; Cornford (1935), 203; Warrington (1961), 182; Zadro (1971), 217; Matthews (1972), 222; Cambiano (1981), 433; Benardete (1984), 28; Roggerone (1990), 93; Mazzarelli (1991), 282; Vitali (1992), 67; N. P. White (1993), 26; Bianchini (1997), 92; Ambuel (2007), 205; Fronterotta (2007), 309.

20 Cf. Apelt (1914), 66; Diès (1925), 336; Martini (1931), 254; Taylor (1961), 125; Meinhardt (1990), 85; Giardini (1997), 559; Duerlinger (2005), 103.

24

viewing the matter this way, do you agree that …?’21 On the second construal, the declarative sentence is governed by ‘tv=de’.22 The corresponding translation is either ‘Are you agreeing because you view the matter this way, that …?’23 or ‘Do you agree when you look at matters this way, that …?’.

The linguistic evidence favors the second construal. For: (1) there are no sure parallels in Plato for ‘su/mfhmi’ followed by a declarative sentence introduced by ‘w¨j’ or ‘oÀti’;24 (2) I found no occurrence in Plato of ‘fhmi¿’ construed with a declarative sentence introduced by ‘w¨j’ (although there are examples with a declarative sentence introduced by ‘oÀti’: cf. Grg. 487D5−7; R. III 387D4−6); (3) the formula ‘tv=de skopw½n’ always looks forward (cf. Grg. 497D9; Phd. 73B4; R. I 351A7; Tht. 163A7; Sph. 254C2; Ti. 89E3; Virt. 378C6).25

As I indicated earlier, two translations are possible when one adopts the second construal: the sentence may be rendered either by ‘Are you agreeing because you view the matter this way, that …?’ or by ‘Do you agree when you look at matters this way, that …?’ A decision between these two translations is important: on the first translation, the proposition expressed by the declarative sentence introduced by ‘w¨j’ is supposed to be a ground supporting what comes before in the argument; on the second translation, the proposition expressed by the declarative sentence is supposed to be a step forward in the argument. The decision will also affect how the declarative sentence introduced by ‘wj’ is to be rendered.

One reflection that might be deployed to reach a decision is based on parallel arguments in other dialogues. On a few occasions Plato deploys arguments that start with a certain predicate being said of something (ti), progress to that predicate being said of at least one something (eÀn ge/ ti) and conclude with that predicate being said of something that is (oÃn ti) (cf. Tht. 188E4−189A14; R. V 478B6−478C2). On the reasonable assumption that an argument of this sort is being put forward also in our Sophist passage, it may be inferred that the first translation is the correct one. For, on the first translation, the proposition expressed by the declarative sentence introduced by ‘wj’ is supposed to be a ground supporting what comes before in the argument, namely the point that ‘something’ is correctly applied only to beings: Plato may be taken to

21 Cf. Ficino (1484), 350; MacKay (1868), 112; Jowett (1892), IV 365; Diès (1925), 337; Martini (1931), 254; Fraccaroli (1934), 148; Arangio-Ruiz (1951), 128; Gentile and Plebe (1965), 50; Taylor (1961), 126; Zadro (1971), 217; Cambiano (1981), 434; Meinhardt (1990), 87; Roggerone (1990), 94; Vitali (1992), 67; Cordero (1993), 125; Bianchini (1997), 92; Duerlinger (2005), 103; Ambuel (2007), 206; Fronterotta (2007), 312.

22 Cf. Apelt (1897), 116.

23 Cf. Apelt (1914), 67; Fowler (1921), 341; Cornford (1935), 204; Warrington (1961), 182; Matthews (1972), 223; Mazzarelli (1991), 282; N. P. White (1993), 26; Brann, Kalkavage, and Salem (1996), 41.

24 Cf. Owen (1971), 226. One passage, Phlb. 28E7−29A1, might provide such a parallel, but the syntax is uncertain and the witnesses provide variant readings for points of the sentence that are relevant to its overall interpetation. The earliest example I have found of ‘su/mfhmi’ followed by a declarative sentence introduced by ‘wj’ is Aristid. Or. I 132, 4 J. (2nd century AD).

25 In this paragraph I am indebted to discussions with Robert Parker and David Robinson.

25

accomplish this feat by pointing out that uttering ‘something’ commits one to uttering ‘at least one something’, leaving it to the reader to supply the premiss that uttering ‘at least one something’ commits one to uttering ‘something that is’.

But another consideration speaks strongly against the first translation. As we have seen, on the first translation the proposition expressed by the declarative sentence introduced by ‘w¨j’ is supposed to be a ground supporting what comes before in the argument, namely the point that ‘something’ is correctly applied only to beings. The only way of extracting this feat from the declarative sentence is to take it to be mentioning the expressions ‘something’ and ‘at least one something’: this was in fact done in my reconstruction in the last paragraph, where I also showed how the goal of supporting the earlier point was achieved. The problem is with what comes later: for, on any reasonable interpretation of the argument, at 237E1 Plato does not mention the expression ‘something’, but uses it. In other words, when he affirms that ‘it is most necessary that whoever does not say something say absolutely nothing’ (237E1−2), Plato is not speaking of a person who is not uttering the expression ‘something’, but of a person who is not saying something. So, the first translation requires a substantial but unannounced shift in the use of the phrase ‘ti le/gein’: in 237D6−7 it must mean ‘to utter the expression “something”’, in 237E1−2 it must mean ‘to say something’. This result has a damning effect on the first translation: although it cannot be excluded that Plato is confused about use and mention, or that he is not confused but he intends the argument to be fallacious, nevertheless an interpretation that avoids such a shift in the usage of a key expression ought to be preferred. I therefore opt for the second translation, which, as we shall see, is able to yield such an interpretation. It is worth noting that the temporal (rather than causal) sense of the participle ‘tv=de skopw½n’ adopted by the second translation has a parallel in a passage of the Phaedo: ‘If, Simmias, you are not convinced by these means, said Socrates, consider whether consensus is reached when you look at matters this way [tv=de/ pv/ soi skopoume/n%26 sundo/cv]’ (73B3−4).

The progression of the first aporetic argument. Here is a step-by-step reconstruction of the first aporetic argument about not being.

The first step (237B7−9) states the initial assumption: we express what in no way is. The second step (237B10−237C9) is the claim that the expression ‘not being’ cannot be

applied to beings. This claim is analogous to the claim that the expression ‘not man’ cannot be applied to men. It is an instance of the general principle that a negative predicative expression consisting of ‘not’ followed by a predicative expression P cannot be applied to anything falling under P.

The third step (237C10−11) is the claim that the expression ‘not being’ cannot be applied to anything that is something (because, as was established in the second step, it cannot be applied to any being).

Theaetetus is bewildered: he asks ‘How so?’ (237C12).

26 No weight is to be attached to the difference between the active ‘skopeiÍn’ and the middle ‘skopeiÍsqai’: cf. LSJ s.v. ‘skope/w’ II 2.

26

The fourth step (237D1−5) is the claim that the expression ‘something’ can only be used in connection with beings. This provides a ground for the third step’s claim that the expression ‘not being’ cannot be applied to anything that is something: could the expression ‘not being’ be applied to anything that is something, then it could after all be applied to beings (because, as the fourth step points out, the expression ‘something’ can only be used in connection with beings), contrary to what had been agreed in the second step. The claim made in the fourth step reflects certain facts of Greek language and in particular of Plato’s Greek: ‘to be something’ (‘eiånai¿ ti’) may be used make existential claims,27 and instances of ‘to call ϕ “something”’ or ‘to call something “ϕ”’ may be employed to make the point which could also have been made by using the corresponding instances of ‘to accept ϕ as existent’.28 Plato’s point might perhaps be described as the claim that the particular quantifier ‘something’ has ‘existential import’, i.e. that ‘… something —’ entails ‘… something that is —’.

The fifth step (237D6−11) makes two claims, a primary and a secondary one. The primary claim is that one says something just in case one says at least one something. The secondary claim is that the word ‘something’ is a sign of unity because it is a singular form of the pronoun, as opposed to its dual and plural forms. The secondary claim supports the primary one. Note that the primary claim is a universally quantified biconditional: its formulation, i.e. the sentence ‘Whoever says something says at least one something’ (cf. 237D6−7), expresses both directions of the biconditional depending on where the stress falls: the most natural reading of the sentence gives the left-to-right direction of the biconditional (with ‘saying something’ in the antecedent and ‘saying at least one something’ in the consequent), but if emphasis is put on the occurrence of ‘something’ in third position then the sentence turns out to mean the same as ‘It is whoever says something that says at least one something’, which gives the right-to-left direction (with ‘saying at least one something’ in the antecedent and ‘saying something’ in the consequent). The secondary claim supports the left-to-right direction of the biconditional (with ‘saying something’ in the antecedent and ‘saying at least one something’ in the consequent); the right-to-left direction (with ‘saying at least one something’ in the antecedent and ‘saying something’ in the consequent) is self-evident. Unless it states a universally quantified biconditional, the fifth step will fail to support the following sixth step.29

The sixth step (237E1−3) makes the claim that one does not say something just in case one says nothing. This is also a universally quantified biconditional: its formulation, i.e. the sentence ‘Whoever does not say something says absolutely nothing’ (cf. 237E1−2), expresses both directions of the biconditional depending on where the stress is put (considerations analogous to those bearing on the sentence at 237D6−7 apply). This universally quantified biconditional follows by contraposition from the universally quantified biconditional stated in the fifth step, on the basis of the implicit equation of ‘nothing’ with ‘not at least one something’. This equation is

27 Cf. 247A10; Hp.Ma. 287C4; Prt. 332C3; Phd. 102B1; R. IX 583C5; Phlb. 37A2; 37A7.

28 Cf. Prt. 332A4; Phd. 103C11; Men. 75E1; 76A1; 76D2; 88A7−88B1.

29 Cf. Szaif (1998), 398.

27

completely natural in Greek, where ‘mhde/n’ (‘nothing’) is both phonetically and graphically indistinguishable from ‘mhd' eÀn’ (‘not even one’), while the negative counterpart of ‘ti’ (‘something’) is the less common ‘ouÃti’.30

The seventh step (237E4−5) adds the intuitive assumption that whoever says nothing fails to speak (i.e. fails to carry out an act of saying).

The eighth step (237E5−6) is the argument’s conclusion: whoever expresses what is not fails to speak.

The structure of the first aporetic argument. How is the conclusion of the first aporetic argument reached? Let us first of all note that the argument’s sixth step is

[1] One does not say something just in case one says nothing,

whereas the seventh step is the claim

[2] Whoever says nothing fails to carry out an act of saying.

[1] and [2] straightforwardly imply

[3] Whoever does not say something fails to carry out an act of saying.

So it is reasonable to expect that the argument’s conclusion is reached on the basis of the following claim (which is not formulated):

[4] Whoever expresses what is not does not say something.

For [4] and [3] immediately entail

[5] Whoever expresses what is not fails to carry out an act of saying,

which is the argument’s intended conclusion, stated in the eighth step. Our target is therefore reduced to obtaining [4]. As I said, this claim is not formulated. However, most of the materials are there which enable the reader to establish it. The fourth step of the argument is the claim that the expression ‘something’ can only be used in connection with beings. I noted that this may perhaps be taken to amount to the claim that the particular quantifier ‘something’ has ‘existential import’, i.e. that ‘… something —’ entails ‘… something that is —’. If we apply this to the case of saying, we obtain warrant for

[6] Whoever says something says something that is.

From [6], by contraposition, we get

[7] Whoever does not say something that is does not say something.

30 Cf. Campbell (1867), Sph. 85; Apelt (1897), 117; Szaif (1998), 398.

28

Now add the prima facie plausible

[8] Whoever expresses what is not does not say something that is.

Propositions [8] and [7] immediately entail [4], which is what we needed.31 The line of reasoning of the first aporetic argument about not being may be summarized as

follows. If you carry out an act of saying, then you say something, so that you say something that is. Therefore if you say what is not then you do not carry out an act of saying (this last move is based on contraposition and relies on taking ‘You say what is not’ to entail ‘It is not the case that you say something that is’). Analogous arguments occur elsewhere in Plato’s work, sometimes with expressions other than the verb ‘to say’ (‘le/gein’): in the Euthydemus (284B1−284C6) there is a vaguely similar argument with ‘to say’ (‘le/gein’); in the Republic (V 478B5−478C2) there is one that is quite close and turns upon ‘to believe’ (‘doca/zein’); in the Theaetetus (188D7−189B6) there are four which are also close and involve ‘to believe’ (‘doca/zein’), ‘to see’ (‘o(ra=n’), ‘to hear’ (‘a)kou/ein’), and ‘to touch’ (‘aÀptesqai’); finally, in the Parmenides (132B8−132C2) there is a vaguely similar one constructed around the noun ‘thought’ (‘no/hma’).

The strength of the first aporetic argument: (i) The ambiguity of ‘to say something’. How compelling is the first aporetic argument about not being?

David Wiggins makes a case that the argument is fallacious. In his view, the fallacy depends on the ambiguity of the sentence ‘You are saying something’. In a first sense, ‘You are saying something’ means that you are carrying out a speech-act with some content: in this first sense of ‘You are saying something’, the inference from ‘You carry out an act of saying’ to ‘You are saying something’ is valid. In a second sense, ‘You are saying something’ means that there is an object that constitutes the target on which your act of saying is directed: in this second sense of ‘You are saying something’, the inference from ‘You are saying something’ to ‘You are saying something that is’ is valid (or at least exempt from the specific fault in which Wiggins is interested). The two senses of the sentence have an important difference with regard to what may or may not be inferred. If ‘You are saying something’ is understood in the first sense, then ‘You are saying …’ is an opaque context and is therefore not open to substitutions of appropriately equivalent expressions (similarly, the fact that ‘You are saying that … is honest’ is an opaque context blocks the inference from ‘You are saying that Nixon is honest’ to ‘You are saying that the most corrupt president is honest’ or ‘You are saying that a crook is honest’). If ‘You are saying something’ is understood in the second sense, then ‘You are saying …’ is a transparent context and is therefore open to substitutions of appropriately equivalent expressions (similarly, the fact that ‘You are saying of … that he is honest’ is a transparent context warrants the inference from ‘You are saying

31 The above reconstruction of the first aporetic argument about not being owes a lot to the excellent accounts offered by Wiggins (1971), 268−72 and Szaif (1998), 397−8 (cf. also Bondeson (1972), 3−4). It is radically different from the reconstruction of Moravcsik (1962), 26, who takes Plato to argue that since ‘what is not’ applies to nothing it has no meaning (the argument would then invalidly infer lack of meaning from emptiness of extension). The interpretation I favored in Crivelli (1990), 26−8 was close to Moravcsik’s.

29

of Nixon that he is honest’ to ‘You are saying of the most corrupt president that he is honest’ and ‘You are saying of a crook that he is honest’).32

The ambiguity spotted by Wiggins might be damaging to the first aporetic argument about not being. In defense of the argument one may say that perhaps it does not depend on Wiggins’s ambiguity. Perhaps the argument is to be understood as an attack on those thinkers who believe that the way a speech-act has content is by being directed on an object (a proposition) which may be properly qualified as being either true or false: such a position may be associated with Frege and constitutes the mainstream framework of modern philosophy of language. An interpretation along these lines restores the validity of the argument (at least with respect to Wiggins’s criticism): it assumes that in the argument the sentence ‘You are saying something’ is understood in only one sense, the second one, so that the argument’s first inference amounts to the inference from ‘You carry out an act of saying’ to ‘You are saying something’ meaning ‘There is an object that constitutes the target on which your act of saying is directed’. This inference is based not on an analysis of how we ordinarily use the sentences ‘You carry out an act of saying’ and ‘You are saying something’, but on a substantial philosophical view about what is involved in carrying an act of saying.

There is some textual support for this interpretation. For, in the similar argument in Republic V (478B5−478C2), we find the inference from ‘to have a belief ’ (‘doca/zein’) to ‘to bring one’s belief to bear on something’ (‘e)pi¿ ti fe/rein th\n do/can’) (478B6): this appears to correspond to the way of understanding the first inference in the Sophist’s argument assumed by the alternative interpretation.

The cost incurred by this interpretation is that the first aporetic argument about not being comes out having a restricted scope: it targets only those philosophers who have a certain conception of propositional attitudes, a conception which is ontologically loaded in that it explains propositional attitudes in terms of a relation obtaining between the attitude’s subject and an object (a proposition). Such a conception of propositional attitudes is not mandatory. One may well avoid claiming that the way a speech-act has a publicly shareable content is by being directed on some object or other: one might claim that the common public content of intentional events boils down to a shared structure of the private mental or linguistic events, a shared structure that might perhaps be expressed by adverbial formulations.

The strength of the first aporetic argument: (ii) The veridical and the existential uses of ‘to be’. Suppose we are satisfied with the defence from the first objection. A different objection may then be raised against the first aporetic argument about not being. Let us go with the argument and accept its conclusion: if you are saying what is not then you do not carry out an act of saying. Is this relevant to the problem of speaking falsely? One might deny it. For one might claim that if you are speaking falsely then you are saying what is not, where what ‘is not’ is what ‘is not true’ (veridical use of ‘to be’), and insist that you need not therefore be saying what is not, where what ‘is not’ is what ‘does not exist’ (existential use of ‘to be’). But the antecedent of the conditional

32 Cf. Wiggins (1971), 271, 279−80.

30

that constitutes the argument’s conclusion involves not the veridical, but the existential use of ‘to be’: when we accepted that if you are saying what is not then you do not carry out an act of saying, we meant that if you are saying what does not exist, or what is not there, then you do not carry out an act of saying.

Again, the first aporetic argument about not being is surely vulnerable to this objection. Nevertheless, the argument may be viewed as presenting a serious philosophical challenge: must we really accept in our ontology propositions that are there even though they are false? Such an ontology is crowded and we would be better off if we were able to account for falsehood without so expensive a commitment.

In sum, the first aporetic argument about not being poses a challenge to those theories of propositional attitudes which are committed to explain propositional attitudes in terms of relations obtaining between the subjects of the attitudes and certain things (propositions) which may be properly qualified as being either true or false. The challenge is that these theories are ontologically uneconomical because they need to postulate abstract entities which are there despite their ‘not being’ in the sense of being false. In resisting this position, Plato is revealing himself as far less of a Platonist than most modern philosophers of language.33

The second aporetic argument about not being (238A1−238C12). The second aporetic argument in support of the thesis that it is impossible to say what is not purports to prove that no predicative expression applies to what does not exist (238A1−238C12). Here is the argument:

ES ‘Mh/pw me/g' eiãpvj’: eÃti ga/r, wÕ maka/rie, eÃsti, kaiì tau=ta/ ge tw½n a)poriw½n h( megi¿sth kaiì prw¯th. periì ga\r au)th\n au)tou= th\n a)rxh\n ouÅsa tugxa/nei.

238A1

THT. Pw½j fv/j; le/ge kaiì mhde\n a)poknh/svj. ES T%½ me\n oÃnti pou prosge/noit' aÃn ti tw½n oÃntwn

eÀteron. A5

THT. Pw½j ga\r ouÃ; ES Mh\ oÃnti de/ ti tw½n oÃntwn aÅra/ pote prosgi¿gnesqai

fh/somen dunato\n eiånai; THT. Kaiì pw½j;

A10 ES ¹Ariqmo\n dh\ to\n su/mpanta tw½n oÃntwn ti¿qemen. THT. Eiãper ge kaiì aÃllo ti qete/on wj oÃn. 238B1 ES Mh\ toi¿nun mhd' e)pixeirw½men a)riqmou= mh/te plh=qoj

mh/te eÁn pro\j to\ mh\ oÄn prosfe/rein.

THT. OuÃkoun aÄn o)rqw½j ge, w¨j eÃoiken, e)pixeiroiÍmen, wÐj fhsin o( lo/goj.

B5

ES Pw½j ouÅn aÄn hÄ dia\ tou= sto/matoj fqe/gcaito aÃn tij hÄ kaiì tv= dianoi¿# to\ para/pan la/boi ta\ mh\ oÃnta hÄ to\ mh\ oÄn xwriìj a)riqmou=;

THT. Le/ge pv=; ES Mh\ oÃnta me\n e)peida\n le/gwmen, aÅra ou) plh=qoj

e)pixeirou=men a)riqmou= prostiqe/nai; B10

238C1 THT. Ti¿ mh/n; ES Mh\ oÄn de/, aÅra ou) to\ eÁn auÅ; THT. Safe/stata/ ge.

33 Cf. below, text to n. 13 of Ch. V.

31

ES Kaiì mh\n ouÃte di¿kaio/n ge ouÃte o)rqo/n famen oÄn e)pi- xeireiÍn mh\ oÃnti prosarmo/ttein.

C5

THT. Le/geij a)lhqe/stata. ES SunnoeiÍj ouÅn w¨j ouÃte fqe/gcasqai dunato\n o)rqw½j

ouÃt' ei¹peiÍn ouÃte dianohqh=nai to\ mh\ oÄn au)to\ kaq' au(to/, a)ll'

eÃstin a)diano/hto/n te kaiì aÃrrhton kaiì aÃfqegkton kaiì aÃlogon;

THT. Panta/pasi me\n ouÅn.

C10

C11 ES ‘Do not yet speak loud’.34 For there is more to come, my dear friend, and it is the greatest

and the first of the puzzles. For it happens to be about the very principle of the matter. THT. What do you mean? Speak and do not hesitate.

238A

ES Some other being could come to be attached to what is. THT. How could it not? ES But shall we say that some being can come to be attached to what is not? THT. How could we?35

ES We posit all of number among beings. THT. If anything else is to be posited as a being.

238B

ES Let us then not try to apply either plurality or unity of number to what is not. THT. We would not be making a correct attempt, it seems, as the argument says. ES In what way could one express through one’s mouth or in any way grasp with thought not

beings or not being without number? THT. Will you say how?

ES If we say ‘not beings’, are we not trying to apply multiplicity of number? THT. So what?

238C

ES And if we say ‘not being’, are we not trying to apply unity, in turn? THT. Most clearly.

ES And we say that it is neither right nor correct to try to apply being to what is not. THT. What you are saying is most true.

ES You then understand that one cannot correctly express nor say nor think of what is not in its own right,36 but it is unthinkable and unspeakable and inexpressible and unsayable?

THT. By all means.

After some introductory remarks (238A1−4), whose purpose is to indicate the importance of the coming puzzle, the first step of the argument (238A5−10) is the claim that while an attribute that implies being may well be assigned to what is, an attribute that implies being does cannot be assigned to what is not. This claim is straightforward: it is analogous to the claim that an attribute that implies whiteness cannot be assigned to what is not white. Plato uses the expression ‘some being’ to quantify over attributes that imply being: in a similar manner one might use the expression ‘some animal’ to quantify over attributes like dog, cat, etc., i.e. attributes that imply the kind animal.

The second step (238A11−238B5) is the claim that numerical attributes (i.e. attributes like one, two, many, etc.) imply being,37 and therefore cannot be assigned to what is not. How is it that

34 Cf. S. Fr. 662,1 Radt (with Stob. V 934, 4−5 W.-H.).

35 Cf. Prm. 163E6−164A1.

36 On ‘what is not in its own right’ cf. below, n. 76 of Ch. IV and text thereto.

37 Cf. Tht. 188E8−10.

32

numerical attributes imply being? Plato does not explain. One point that is probably relevant has to do with views about ‘to be’ which can be plausibly attributed to Plato (here I anticipate some points which will be elaborated upon later).38 Plato seems to think that the incomplete uses of ‘to be’, whereby the verb is employed either to express participation in an attribute or to make an identification, imply the complete use, whereby the verb expresses existence. Briefly: ‘to be ϕ’ implies ‘to exist’. On this conception, all attributes may be regarded as implying being because any application of the incomplete use of ‘to be’ that expresses participation in an attribute implies an application of the complete use of ‘to be’ that expresses existence. So, at least part of the explanation of why numerical attributes imply being is that all attributes imply being. But this is unlikely to be the complete explanation. For Theaetetus’ enthusiastical answer (238B1) to the question whether numerical attributes imply being suggests that the implication of being by numerical attributes is especially intimate and tight. This may be accounted for by highlighting two facts about numerical attributes. First, numerical predicative expressions, which express numerical attributes, are always connected to a (possibly understood) count-noun (one and the same chunk of reality may be described as ‘one book’, ‘one hundred sheets’, ‘two hundred pages’, etc.) and the application of any numerical predicative expression is therefore always taken to presuppose the existence of as many instances of the count-noun as indicated by the numerical predicative expression itself (‘one book’ presupposes the existence of one object falling under the count-noun ‘book’, ‘one hundred sheets’ presupposes the existence of one hundred objects falling under the count-noun ‘sheet’, etc.). This may be plausibly taken to indicate that numerical attributes imply existence.39 Secondly, the fact that it is impossible to count non-existent things (how many non-existent cats are there on the mat?) seems to indicate that numerical attributes, which are assigned in counting, imply existence.40

The third step (238B6−238C7) is the claim that the expressions ‘not being’ and ‘not beings’, being respectively singular and plural, involve numerical attributes and therefore cannot be used to pick out what is not.

The fourth step (238C8−12) states the argument’s conclusion: that what is not is unsayable and unthinkable. The ground for this conclusion might be that no expression can be used to pick out what is not because any expression one might try to use will be either in the singular or in the plural and will therefore involve some numerical attribute ( just like ‘not being’ and ‘not beings’). The extension from speech to thought is warranted by Plato’s view that thought is internal silent speech (cf. 263E3−264B5).

Note that the second aporetic argument does not purport to establish that the expression ‘not being’ is inconsistent or that the concept of not being is inconsistent. Its purpose is merely to show that no thing which is not can be referred to either in language or in thought.

38 Cf. below, paragraph to n. 67 of Ch. IV and subsection to n. 182 of that same chapter.

39 Cf. Ryle (1960), 445.

40 Cf. Gosling (1973), 222.

33

The third aporetic argument about not being (238D1−239A12). The third aporetic argument in support of the thesis that it is impossible to say what is not maintains that the paradox itself cannot even be stated: saying that what is not is unsayable and unthinkable involves attributing unity to what is not (because an expression of singular form is used), contrary to an earlier result (238D1−239A12).

Not being and non-existence. In my reconstruction of the three aporetic arguments for the thesis that it is impossible to say ‘what is not’ I have assumed that ‘what is not’ is ‘what does not exist’. Such an assumption clashes with an interpretation which has enjoyed widespread favor among commentators in recent decades. According to this widespread interpretation, in the three aporetic arguments for the thesis that it is impossible to say ‘what is not’, ‘what is not’ is ‘what has no attributes’, ‘that which for all F is not F ’, an object completely ‘naked’, ‘with all the being knocked out of it’.41

At least three arguments may be adduced in support of the view that in this part of the dialogue what is not is a ‘naked’ object.

(1) At the beginning of the first argument Plato uses the expression ‘what in no way is’ (‘to\ mhdamw½j oÃn’, 237B7−8). Such an expression does suggest that ‘naked’ objects are in the offing.

(2) In the middle of the second puzzle about not being (238A8−10) Plato holds that we cannot attribute something which is to what is not. This is easily understood if what is not is a ‘naked’ object, i.e. an object completely devoid of attributes.

(3) The puzzle concerning images speaks of the contrary of what is true (cf. 240B5) and the puzzle concerning false statements and false beliefs speaks of the contrary of things which are (cf. 240D6−8). Later in the dialogue Plato seems to think that the puzzles depend on the concept of the contrary of what is (cf. 257B3−4, 258E6−259A1). But the contrary of what is may be naturally understood as a ‘naked’ object, something totally devoid of being.

None of these arguments is however compelling. As for argument (1), the expression ‘what in no way is’ may be naturally understood as a strong denial, meaning ‘what absolutely is not’, ‘what no way is’. As for argument (2), I have already offered an interpretation of the second puzzle about not being that works without the assumption that what is not is a ‘naked’ object. As for argument (3), nothing guarantees that the contrary of what is should be a ‘naked’ object. In due course I shall offer an interpretation of Plato’s concept of the contrary of what is that does not make it into a ‘naked’ object, but takes it to be what does not exist.

Moreover, there are considerations that suggest that in the three aporetic arguments for the thesis that it is impossible to say ‘what is not’, ‘what is not’ is ‘what does not exist’.

(1) At the beginning of the first aporetic argument, the ES asks Theaetetus whether ‘we somehow dare to express what in no way is [to\ mhdamw½j oÃn]’ (237B7−8). It is extremely unlikely that the young and philosophically unskilled Theaetetus (cf. 234C2−234E7, 237C5−6) could understand the expression ‘what in no way is’ in the philosophically ‘loaded’ sense of ‘what

41 Cf. Frede (1967), 75; Malcolm (1967), 136−7; Owen (1971), 235−6, 247, 266; Gosling (1973), 222; Bondeson (1976), 8; Lewis (1976), 110; Malcolm (1985b), 520; Bordt (1991), 499.

34

has no attributes’. At that stage of the argument it is much more plausible that he should understand the expression in the sense of ‘what does not exist’.

(2) At 264D1−2 Plato uses the expression ‘in no way to be’ (‘mhdamw½j eiånai’) to express the non-existence of falsity.

(3) Were the expression ‘the things which in no way are’ (‘ta\ mhdamw½j oÃnta’) to mean ‘the things which have no attributes’, then, parallelwise, the expression ‘the things which in all ways are’ (‘ta\ pa/ntwj oÃnta’) would have to mean ‘the things which have all attributes’. Plato uses the expression ‘the things which in all ways are’ (‘ta\ pa/ntwj oÃnta’) at 240E5, in a context where he also uses ‘to be in no way’ (‘mhdamw½j eiånai’, 240E5) and ‘the things which in no way are’ (‘ta\ mhdamw½j oÃnta’, 240E2). But, were this occurrence of ‘the things which in all ways are’ (‘ta\ pa/ntwj oÃnta’) to mean ‘the things which have all attributes’, then the argument to the formulation of which it contributes would be nonsensical.42

The contradictory nature of images. Plato presents two arguments which a sophist could use to avoid being captured by the hunt of the ES and Theaetetus. The first argument (239C9−240C6) purports to establish that the concept of an image implies a contradiction. The argument helps the sophist’s escape because the ES and Theaetetus had deployed the concept of an image in their attempt to define the sophist. The argument opens with the request of a characteristic that belongs to all and only the objects to which the expression ‘image’ applies (239D3−4). After an answer to this question has been offered, the argument develops by steps where one uses the verb ‘to be’ and its negation ‘not to be’ while thinking of some complement for their predicate position, but the complement is dropped out (a ‘predicative elliptical’ use of ‘to be’ and ‘not to be’). Such a predicative elliptical use creates the illusion of a contradiction involving the existential use of ‘to be’.

Consider, for instance, an image of a cat: an image of a cat is an image, but is not a cat. These two claims, ‘An image of a cat is an image’ and ‘An image of a cat is not a cat’, involve the predicative use of ‘to be’ and ‘not to be’. But in the argument actually presented in the Sophist the two complements for the predicate positions of ‘to be’ and ‘not to be’, namely the expressions ‘an image’ and ‘a cat’, are dropped. So, what we actually find is: ‘An image of a cat is’ and ‘An image of a cat is not’. This looks like a flat contradiction involving the existential use of ‘to be’ (cf. 240B12−240C1). In order to disarm this puzzle, Plato will have to point out what is going on: that there is a fallacy involving the dropping of a complement for the predicate positions of ‘to be’ and ‘not to be’ and passing from the predicative to the existential use of the verb.43

An even more effective antidote, which will take care also of possible reformulations of the argument that rely on fallacious passages that might be regarded as sound in certain quarters, will involve showing that a negative statement of predication involving the verb ‘to be’ (i.e. an instance of the schema ‘σ is not ϕ’) does not entail a negative existential statement (i.e. the

42 Cf. Heinaman (1983), 4; Brown (1994), 226. However in the Cratylus (386D3−4) Plato credits Euthydemus with the view that ‘all things always have all attributes simultaneously’.

43 Cf. Malcolm (1967), 137; Brown (1986/99), 468.

35

corresponding instance of ‘σ does not exist’): otherwise the argument could be revamped as moving from the true statement ‘An image of a cat is an image and is not a cat’ to the false (indeed contradictory) ‘An image of a cat exists and does not exist’. Such an antidote will be offered by Plato in his account of negation at 257B1−257C4.44

The impossibility of believing or saying falsehoods. Plato presents a second argument which a sophist could use to avoid being captured by the hunt of the ES and Theaetetus. This argument (240C7−241B4) is a version of the paradox of false belief and false statement, i.e. it purports to show that it is impossible to believe or state falsehoods. This will help the sophist to evade the attempted definition of the ES and Theaetetus because their attempt relies on the idea that the sophist’s art is one deception and therefore leads people to believe falsehoods:

ES Ti¿ de\ dh/; th\n te/xnhn au)tou= ti¿na a)fori¿santej h(miÍn au)toiÍj sumfwneiÍn oiâoi¿ te e)so/meqa;

THT. Pv= kaiì to\ poiÍo/n ti fobou/menoj ouÀtw le/geij;

240C7

ES àOtan periì to\ fa/ntasma au)to\n a)pata=n fw½men kaiì th\n te/xnhn eiånai¿ tina a)pathtikh\n au)tou=, to/te po/teron

240D1

yeudh= doca/zein th\n yuxh\n h(mw½n fh/somen u(po\ th=j e)kei¿nou te/xnhj, hÄ ti¿ pot' e)rou=men;

THT. Tou=to: ti¿ ga\r aÄn aÃllo eiãpaimen; D5 ES Yeudh\j d' auÅ do/ca eÃstai ta)nanti¿a toiÍj ouÅsi doca/-

zousa, hÄ pw½j;

THT. OuÀtwj: ta)nanti¿a. ES Le/geij aÃra ta\ mh\ oÃnta doca/zein th\n yeudh= do/can; THT. ¹Ana/gkh. D10 ES Po/teron mh\ eiånai ta\ mh\ oÃnta doca/zousan, hà pwj

eiånai ta\ mhdamw½j oÃnta; 240E1

THT. Eiånai¿ pwj ta\ mh\ oÃnta deiÍ ge, eiãper yeu/setai¿ pote/ ti¿j ti kaiì kata\ braxu/.

ES Ti¿ d'; ou) kaiì mhdamw½j eiånai ta\ pa/ntwj oÃnta doca/zetai;

E5

THT. Nai¿. ES Kaiì tou=to dh\ yeu=doj; THT. Kaiì tou=to. ES Kaiì lo/goj oiåmai yeudh\j ouÀtw kata\ tau)ta\ nomi-

sqh/setai ta/ te oÃnta le/gwn mh\ eiånai kaiì ta\ mh\ oÃnta eiånai. E10

241A1 THT. Pw½j ga\r aÄn aÃllwj toiou=toj ge/noito; ES Sxedo\n ou)damw½j: a)lla\ tau=ta o( sofisth\j ou)

fh/sei. hÄ ti¿j mhxanh\ sugxwreiÍn tina tw½n euÅ fronou/ntwn,

oÀtan aÃfqegkta kaiì aÃrrhta kaiì aÃloga kaiì a)diano/hta prosdiwmologhme/na vÅ ta\ <mh\ oÃnta kata\ ta\> pro\ tou/twn

A5

o(mologhqe/nta; manqa/nomen, wÕ Qeai¿thte, aÁ le/gei; THT. Pw½j ga\r ou) manqa/nomen oÀti ta)nanti¿a fh/sei

le/gein h(ma=j toiÍj nundh/, yeudh= tolmh/santaj ei¹peiÍn w¨j

eÃstin e)n do/caij te kaiì kata\ lo/gouj; t%½ ga\r mh\ oÃnti to\ oÄn prosa/ptein h(ma=j polla/kij a)nagka/zesqai, diomologh-

241B1

same/nouj nundh\ tou=to eiånai pa/ntwn a)dunatw¯taton. ES ¹Orqw½j a)pemnhmo/neusaj.

B4

44 Cf. below, text to n. 179 of Ch. IV.

36

ES What then? By offering what definition of his art will we be able to be consistent with ourselves?

THT. How do you mean? What do you fear in speaking thus?

240C

ES When we say that he deceives us in connection with apparition45 and that his art is one of deception, shall we say that our soul believes falsehoods because of his art? Or what shall we say?

240D

THT. This: for what else could we say? ES And a false belief will be one that believes the contraries of the things which are. Or what? THT. Yes, the contraries.

ES Do you then say that a false belief believes the things which are not? THT. Necessarily.

ES By believing that the things which are not are not, or that the things which in no way are somehow are?

240E

THT. It must believe that the things which are not somehow are, if anyone will ever somehow err even for a short time.

ES What then? Does it not also believe that the things which in all ways are are in no way? THT. Yes.

ES Is this then also a falsehood? THT. This too.

ES Then, I think, a statement will also be regarded as false in the same way, by stating that the things which are are not and that the things which are not are.

THT. How else could it come to be such?

241A

ES Perhaps in no way. But the sophist will deny these things. Or what means are there by which anyone of those who are able to reason well should concede them, when, in conformity with our earlier agreements, the things which are not have been additionally agreed to be inexpressible and unspeakable and unsayable and unthinkable? Do we understand, Theaetetus, what he says?

THT. How do we not understand that he will claim that by daring to affirm that falsehoods exist in beliefs and statements we say things contrary to what was said just now? For we are obliged often to apply being to what is not, having agreed just now that this is the most impossible thing of all.46

ES You remembered correctly.

241B

The text handed down by the main MSS and the difficulties it raises. The text at 241A3−7 deserves some discussion. One of the main MSS, W, omits ‘ou)’ at 241A3 and ‘hÄ’ at 241A4. At 241A5−7 the main MSS read ‘oÀtan aÃfqegkta kaiì aÃrrhta kaiì aÃloga kaiì a)diano/hta prosdiwmologhme/na vÅ ta\ pro\ tou/twn o(mologhqe/nta’ (the only variant is provided by T, which has ‘prodiwmologhme/na’ instead of the ‘prosdiwmologhme/na’ of the rest): recent eds. have variously emended these words (cf. below). Finally, at 241A7, all the main MSS have ‘aÁ le/geij’ (T had written ‘oÁ le/geij’, but corrected himself ): most recent eds. emend this to ‘aÁ le/gei’ (which is found in two later MSS: Venetus 186 offers this reading as a correction, while Parisinus 1812 presents it over an erasure).

The text of the main MSS is hard to make sense of. For, on the one hand, it requires that in the sentence at 241A5−7 ‘ta\ pro\ tou/twn o(mologhqe/nta’ be the grammatical subject of

45 At 236B4−7 the ES characterized an apparition as an image which appears to be like its model but actually is not like it. An apparition therefore leads people to form a false belief (i.e. the belief in its being what it appears to be but is not).

46 Cf. 238A8−10; 238E8−239A2.

37

‘prosdiwmologhme/na vÅ’ (or ‘prodiwmologhme/na vÅ’) with ‘aÃfqegkta kaiì aÃrrhta kaiì aÃloga kaiì a)diano/hta’ as complement. Therefore the sentence at 241A5−7 must be translated by something like ‘… when the things agreed before these have been additionally agreed to be inexpressible and unspeakable and unsayable and unthinkable’ (or ‘… have been earlier agreed to be …’, if the reading ‘prodiwmologhme/na’ is adopted). On the other hand, what was earlier agreed to be inexpressible and unspeakable and unsayable and unthinkable was what is not (cf. 238C8−12 and 238E5−239A7). So the text of the main MSS seems to presuppose that ‘the things agreed before these’ are identical with ‘what is not’. But such an identification is very implausible. Heindorf and Campbell do attempt to justify it, by claiming that Plato wants to avoid the expression ‘what is not’ and therefore replaces it with the phrase ‘the things agreed before these’. But how could ‘the things agreed before these’ be identical with ‘what is not’? The answer of Heindorf and Campbell is that the expression ‘what is not’ and variants of it occured frequently in the discussion of the nature of images at 239C9−240C6 (cf. 240B7, 240B12, 240C2, and 240C5) and by referring back to this discussion Plato would be alluding to ‘what is not’.47 To be sure, the words ‘the things agreed before these’ (‘ta\ pro\ tou/twn o(mologhqe/nta’, 241A6−7) probably allude to agreements made before the discussion of 240C7−241A2: for the ‘tou/twn’ at 241A6 picks up the ‘tau=ta’ at 241A3, which in turn refers to the propositions agreed by the ES and Theaetetus in the course of the discussion at 240C7−241A2 (because it is the object of ‘o( sofisth\j ou) fh/sei’ at 241A3−4). So, the words ‘the things agreed before these’ could well refer to the discussion of the nature of images, as Heindorf and Campbell assume. But it is hard to see how a reference to the discussion of the nature of images could be reasonably taken to be a reference to ‘what is not’. Moreover, it remains unclear why Plato should want to avoid the expression ‘what is not’ at 241A6: he surely did not shy from using it in the immediately preceding lines. So, the justification of the reading of the main MSS offered by Heindorf and Campbell is far-fetched.

Can the text of the main MSS be retained? The most plausible attempt to retain the text of 241A3−7 handed down by the main MSS is based on a way of understanding it which is radically different from the one which gives rise to the difficulties mentioned in the last subsection: treat ‘aÃfqegkta kaiì aÃrrhta kaiì aÃloga kaiì a)diano/hta prosdiwmologhme/na’ (or ‘aÃfqegkta … prodiwmologhme/na’) as a single phrase that constitutes the grammatical subject of the sentence, take ‘vÅ’ as a copula separate from ‘prosdiwmologhme/na’ (or ‘prodiwmologhme/na’) and regard ‘ta\ pro\ tou/twn o(mologhqe/nta’ as the complement of the copula. The resulting translation of 241A3−7 is: ‘But the sophist will deny these things. Or what means are there by which anyone of those who are able to reason well should concede them, when things which have been additionally [or: earlier] agreed to be inexpressible and unspeakable

47 Cf. Campbell (1867), Sph. 97.

38

and unsayable and unthinkable [sc. the things which are not] are the things agreed before these [sc. have the attribute of being believed to be and that of being said to be]?’48

This exegesis is open to two objections. First, the separation of ‘vÅ’ from ‘prosdiwmologhme/na’ (or ‘prodiwmologhme/na’) (241A6) is unnatural. Secondly, it is not plausible to take ‘ta\ pro\ tou/twn o(mologhqe/nta’ (241A6−7) to refer to the attributes mentioned in the remarks of the preceding lines. For ‘ta\ pro\ tou/twn o(mologhqe/nta’ (241A6−7) in the ES’s question seems to be picked up by the words ‘toiÍj nundh/’ (241A9) in Theaetetus’ reply and this in turn requires that ‘the things things agreed before these’ should coincide with the prohibition to apply being to what is not (cf. ‘diomologhsame/nouj nundh\’ at 241B2−3). Moreover, it is plausible to assume that the ‘tou/twn’ in ‘ta\ pro\ tou/twn o(mologhqe/nta’ (241A6−7) picks up the immediately preceding ‘tau=ta’ (241A3) which in turn refers to the remarks made in the preceding lines:49 this would rule out the possibility that ‘ta\ pro\ tou/twn o(mologhqe/nta’ could refer to the attributes mentioned in the remarks of the preceding lines.

In view of these difficulties, I think that the attempted defence to retain the text of 241A3−7 handed down by the main MSS fails and that some emendation is necessary—a view shared by all recent eds.

Earlier emendations of the text of the main MSS. The text at 241A3−7 has been variously emended. Friedrich Schleiermacher moves the question mark from after ‘o(mologhqe/nta’ (241A7) to

after ‘prodiwmologhme/na vÅ’ (241A6) and reads ‘aÁ le/getai’ at 241A7 in place of the ‘aÁ le/geij’ of the main MSS (which must be emended anyhow).50

At 241A6 Karl Friedrich Hermann reads ‘pro\j diwmologhme/na’ in place of the ‘prosdiwmologhme/na’ and the ‘prodiwmologhme/na’ of the main MSS, regarding the isolated ‘pro\j’ as an adverb.51 This emendation may help to dispel the uneasiness which could arise from the fact that ‘prosdiomologe/omai’ is not attested elsewhere in the Greek literature, but contributes nothing to solving the problem outlined in the last subsection.

Johan Nicolai Madvig deletes ‘aÃfqegkta kaiì aÃrrhta kaiì aÃloga kaiì a)diano/hta’: he thinks that these words crept into the text from an interpreter’s gloss on ‘ta\ pro\ tou/twn

48 The above defence of the text of the main MSS may be described as the result of a joint effort of several Oxford classicists and philosophers (specifically David Charles, Juliane Kerkhecker, and Benjamin Morison) some of whom attended a presentation in which my thoughts on this portion of text were made public.

49 This is nothing more than a plausibility because if the reading of W is adopted then the ‘tau=ta’ of 241A3 does not refer back to the remarks made in the preceding lines, but forwards to the following question (attributed to the sophist): ‘This is what a sophist will say: “What means are there … ?” ’.

50 Schleiermacher put forward this emendation in the first edition of his translation of Plato’s works (see Schleiermacher (1807), 187, 490). In the third edition he rejected it without giving any reason (see Schleiermacher (1857), 130, 337).

51 Cf. Hermann (1851), XXVI, 376; LSJ s.v. ‘pro/j’ D.

39

o(mologhqe/nta’ based on 238C8−12.52 Madvig’s emendation is endorsed by Schanz, Apelt, Burnet, and Robinson (all except Robinson adopt the reading ‘prosdiwmologhme/na’, while Robinson chooses ‘prodiwmologhme/na’).

Auguste Diès instead retains the words expunged by Madvig and deletes ‘ta\ pro\ tou/twn o(mologhqe/nta’ while preferring the variant ‘prodiwmologhme/na’ (according to Diès ‘ta\ pro\ tou/twn o(mologhqe/nta’ was a gloss on ‘prodiwmologhme/na’ which at some point got incorporated into the text).53

Criticism of the emendations of Schleiermacher and Diès. The emendations of Schleiermacher and Diès share a trait: they both require that the grammatical subject of ‘prodiwmologhme/na vÅ’ be retrieved from the ‘tau=ta’ of 241A3. Since, as I pointed out earlier, the ‘tau=ta’ of 241A3 refers to the propositions about the things which are not agreed by the ES and Theaetetus in the course of the discussion of 240C7−241A2, both emendations require that the sentence at 241A5−7 be translated by something like ‘… when they [sc. the propositions just agreed] have been earlier agreed to be inexpressible and unspeakable and unsayable and unthinkable’. This raises a problem: earlier, at 238C8−12 (cf. 238E5−239A7), it was agreed that what is not is inexpressible and unspeakable and unsayable and unthinkable, but at no earlier point was it agreed that propositions about what is not are inexpressible and unspeakable and unsayable and unthinkable. An attempt might be made to solve this problem by assuming that the description ‘inexpressible and …’ is transferred from what is not to propositions about it. But this attempt is unsatisfactory, for two reasons. First, the transfer of the description ‘inexpressible and …’ from what is not to propositions about it requires this description to undergo a modification in meaning, a requirement which sits awkwardly with the fact that the description in question has almost the status of a technical phrase. Secondly, on one occasion later in the dialogue (260D2−3) the idea is recalled that what is not is unsayable and unthinkable and in another passage (258E7−259A1) the idea surfaces again that what is not is unsayable. It would be very strange if in our passage the description ‘inexpressible and …’ were applied not to what is not, but to propositions about it. The emendations of Schleiermacher and Diès should therefore be rejected.

Criticism of Madvig’s emendation. Madvig’s emendation faces three difficulties. First, the phrase ‘aÃloga kaiì a)diano/hta’ is probably attributed to Plato as early as the 2nd

century AD by the grammarian Julius Pollux (II 120, 2−3).54 But, apart from our Sophist passage, the phrase occurs nowhere in the Platonic corpus. So, if the words ‘aÃfqegkta kaiì aÃrrhta kaiì aÃloga kaiì a)diano/hta’ are an incorporated gloss, the corruption occurred early on. Of course,

52 Cf. Madvig (1871), 381−2.

53 Cf. Diès (1925), 343.

54 I use the cautionary adverb ‘probably’ because the extant manuscripts of Pollux’s Onomastikon are derived from incomplete and interpolated copies. The phrase ‘aÃloga kaiì a)diano/hta’ is also attributed to Plato by Tim. Lex. s.v. ‘aÃloga’ (between the 1st and the 4th century AD) and by the Lexica Segueriana (see Bekker (1814−21), I 385, 16−17).

40

this is not impossible—but it is unlikely, a fact that detracts plausibility from Madvig’s emendation.

The second difficulty faced by Madvig’s emendation comes from a remark made by the ES late in the dialogue. Before unleashing his final attack, based on a definition of statement and culminating in an account of falsehood, the ES recalls that ‘the sophist denied altogether that falsehood even exists’ (260D1−2) on the grounds that ‘one neither says nor thinks what is not since what is not does not partake at all in any way of being’ (260D2−3). This late remark probably refers back to the argument we are now dealing with, i.e. the argument of 240C7−241B4: this is the only earlier passage in the Sophist which matches the late remark’s description.55 Within this late remark, the sentence ‘One neither says nor thinks what is not [to\ … mh\ oÄn ouÃte dianoeiÍsqai¿ tina ouÃte le/gein]’ (260D2−3) corresponds to the words ‘unsayable and unthinkable [aÃloga kaiì a)diano/hta]’ in the portion of text deleted by Madvig. What is worse, if Madvig’s expunction is accepted, it is no longer clear that the claim that what is not is unsayable and unthinkable plays any role in the argument.

The third difficulty faced by Madvig’s emendation is philosophical. If Madvig’s emendation is correct then (independently of whether at 241A7 one reads ‘prosdiwmologhme/na’ or ‘prodiwmologhme/na’) the argument reaches its conclusion only with Theaetetus’ remark at 241A8−241B3: it is here that an explanation is offered of how the points made by the ES and Theaetetus at 240D1−241A2 clash with earlier agreements. The explanation is that ‘by daring to affirm that falsehoods exist in beliefs and statements’ (241A9−241B1) the two inquirers ‘are obliged often to apply being to what is not’ (241B1−2). But how is it that they are obliged to do this? Several suggestions have been made by commentators. According to some,56 the ES and Theaetetus ‘apply being to what is not’ when they say that falsehoods exist in beliefs and statements: since falsehoods are things which are not, by saying that falsehoods exist in beliefs and statements they imply that things which are not are. This exegesis can hardly be right because it makes the portion of argument 240E1−241A2 irrelevant. According to other commentators,57 the ES and Theaetetus apply being to not being when they describe false belief (resp. false statement) as believing (resp. stating) that the things which are not are or that the things which are are not. This interpretation however raises a further difficulty: how could the suggested description of false belief and false statement generate a difficulty for the ES and Theaetetus? According to some commentators,58 a difficulty arises because by admitting that false belief (resp. false statement) believes (resp. states) that the things which are not are or that the things which are are not the ES

55 The only other passage to which the ES might be referring is 237A3−4: ‘This account dared to assume that what is not is: for in no other way would falsehood have turned out to be’. However, since it says nothing about the impossibility of saying and thinking what is not, this passage cannot be the target of the ES’s back-reference.

56 Cf. de Rijk (1986), 91−2; Movia (1991), 237.

57 Cf. Arangio-Ruiz (1951), 138; Detel (1972), 69; Keyt (1973), 291; Szaif (1998), 409−10; Notomi (1999), 189; Robinson (1999), 149−50.

58 Cf. Szaif (1998), 410−11.

41

and Theaetetus are taken to be committed to grant that the things which are not are or the things which are are not. The inference would involve a passage from ‘being believed (resp. stated) to be’ to ‘being’ and a passage from ‘being believed (resp. stated) not to be’ to ‘not being’. But such an inference enjoys the dubious privilege of being both blatantly fallacious and unparalleled in Plato: it cannot be reasonably taken to be what generates a difficulty for the ES and Theaetetus. Other commentators59 suggest instead that a difficulty arises because the description offered by the ES and Theaetetus of people holding false beliefs (resp. making false statements) lends itself to be understood as a description of people holding contradictory beliefs (resp. making contradictory statements), i.e. of people holding beliefs (resp. making statements) whose contents could be expressed by the sentences ‘The things which are not are’ and ‘The things which are are not’: such an understanding of the ascription of false belief (resp. false statement) performed by the ES and Theaetetus would be based on a confusion between the transparent and the opaque readings of certain formulations involving the verb ‘to believe’ (resp. ‘to state’) and would probably not be unparalleled in Plato (cf. Tht. 189D4−190E4).60 But this interpretation also cannot be right, for two reasons. First, Plato says that by being obliged to ‘apply being to what is not’ (241B1−2), the ES and Theaetetus have committed themselves to ‘the most impossible thing of all’ (241B3): but it does not look as if someone who describes people as having contradictory beliefs (resp. making contradictory statements) may be fairly described as committing himself or herself to ‘the most impossible thing of all’. (It would be very strange if Plato were here relying on Aristotle’s dubious argument that it is impossible to believe contradictions, cf. Metaph. Γ3, 1005b25−32.) Secondly, the interpretation in question requires that when Plato describes the ES and Theaetetus as ‘obliged often to apply being to what is not’ (241B1−2), the expression ‘to apply being to what is not’ is used very loosely, as meaning something like ‘to mix being with not being’ or ‘to connect “to be” with “not to be” in the same sentence’. But, in the context of these pages of the Sophist, the expression ‘to apply being to what is not’ cannot be taken in such a loose sense, but is more specific: it means something like ‘to apply the concept of being to that which as a matter of fact is not’ or ‘to apply the expression “being” to that which as a matter of fact is not’ (cf. the occurrences of ‘to apply’, ‘prosa/ptein’, at 238E8 and 239A3). So, if Madvig’s emendation is correct then it remains utterly unclear how it is that the two inquirers ‘are obliged often to apply being to what is not’ (241B1−2).

In view of these difficulties, Madvig’s emendation should be resisted.61

A new emendation. Both Madvig and Diès emend the passage by deleting bits of it which they regard as surreptitiously incorporated glosses. Perhaps one should go the opposite way, i.e. integrate rather than delete. One possibility is to read ‘oÀtan aÃfqegkta kaiì aÃrrhta kaiì aÃloga kaiì a)diano/hta prosdiwmologhme/na vÅ ta\ <mh\ oÃnta kata\ ta\> pro\ tou/twn

59 Cf. Crombie (1963), 506−7.

60 Cf. Crivelli (1998), 13−20.

61 Madvig’s emendation is rejected also by Cambiano (1981), 439, Rosen (1983), 202, and Movia (1991), 237.

42

o(mologhqe/nta’ and assume that ‘mh\ oÃnta kata\ ta\’ dropped out by homoeoteleuton. The translation is: ‘… when, in conformity with our earlier agreements, the things which are not have been additionally agreed to be inexpressible and unspeakable and unsayable and unthinkable’. Another possibility is to read ‘oÀtan aÃfqegkta kaiì aÃrrhta kaiì aÃloga kaiì a)diano/hta prosdiwmologhme/na vÅ <e)keiÍna kata\> ta\ pro\ tou/twn o(mologhqe/nta’ and assume that ‘e)keiÍna kata\’ dropped out. The translation is: ‘… when, in conformity with our earlier agreements, those things [sc. the things which are not] have been additionally agreed to be inexpressible and unspeakable and unsayable and unthinkable’. Both emendations solve completely the problem of the subject of the description ‘inexpressible and …’: according to both readings, the ES is straightforwardly saying that the things which are not are inexpressible etc. For ‘kata/’ with the accusative meaning ‘in conformity with’ in the Sophist cf. 216A1, 245C5, 256C7−8, 257E9, and 265E5. In particular, the conjectured ‘kata\ ta\ pro\ tou/twn o(mologhqe/nta’ has a near parallel in ‘kata\ to\n eÃmprosqen lo/gon o(mologhqe/nta’ at Lg. IX 854A4 (‘kata/’ with the accusative of a phrase from the root ‘o(molog-’ is frequent in Plato: cf. Sph. 216A1; Prt. 350E5; Grg. 468E2; R. IV 443A6−7; Criti. 106B6; Lg. XI 920D1−2; 921C1−2). These emendations are bold, but not more, in my view, than those proposed by Madvig and Diès. I opt for the first one because it is slightly more explicit and it can be more easily justified from a paleographical point of view.

A rough outline of the argument. The argument begins by asserting that a false belief believes things which are not. At this point one might expect the train of thought to progress in the direction of the claim that it is impossible to believe things which are not. But, surprisingly, a different direction is taken: we are told that a false belief believes either that the things which are not are or that the things which are are not (the two cases are obviously those of affirmative and negative false beliefs). A parallel account is then introduced for statements: a false statement states either that the things which are are not or that the things which are not are (the two cases are clearly those of negative and affirmative false statements). These results are understood as involving existential claims: what has been putatively established is that a false belief believes either that the things which do not exist exist or that the things which exist do not exist, and that a false statement states either that the things which exist do not exist or that the things which do not exist exist. So, certain false beliefs are about things which do not exist and certain false statements are about things which do not exist. This however clashes with the second aporetic argument about not being (238A1−238C12), whose conclusion is that ‘what is not in its own right […] is unthinkable and unspeakable and inexpressible and unsayable’ (238C9−11). This conclusion of the second aporetic argument is now described as having been ‘additionally agreed’ in accordance with ‘earlier agreements’, which may be plausibly identified with agreements to the effect that being should not be applied to what is not (cf. 241B1−3).

The last part of our passage (241A8−241B4) reports the putative comments of a sophist: ‘by daring to affirm that falsehoods exist in beliefs and statements’ (241A11−241B1) the ES and Theaetetus are ‘obliged often to apply being to what is not’ (241B1−2). One might wonder how it is that the ES and Theaetetus are ‘obliged often to apply being to what is not’. Very roughly, the reason why the ES and Theaetetus are compelled to apply being to what is not has to do with their commitment to concede that what does not exist can be thought of and spoken of (cf.

43

above) and with the fact that the second aporetic argument about not being (238A1−238C12) putatively proves that whoever grants that what does not exist can be thought of and spoken of is compelled to apply being to what is not. Plato probably expects us to realize nothing more than this. Nevertheless it is worthwhile to work out the details of the argument by filling out its main steps on the basis of the second aporetic argument about not being. The points made at 240D1−241A2 commit the ES and Theaetetus to grant that some false believers have thoughts about things which do not exist and that some false speakers mention things which do not exist (this, as we were reminded at 241A4−7, clashes with the conclusion of the second aporetic argument, according to which what does not exist is unthinkable and unspeakable). The ES and Theaetetus are therefore committed to concede that some thinkers produce inner silent utterances of expressions that refer to things which do not exist and that some speakers produce audible utterances of expressions that refer to things which do not exist. It follows that the ES and Theaetetus are committed to grant that certain expressions pick out things which do not exist. Since all expressions are of either singular or dual or plural form, the ES and Theaetetus are committed to concede that either unity or multiplicity holds of things which do not exist and that therefore number also holds of the things which do not exist (cf. 238B6−238C4). Since number entails existence (cf. 238A11−238B1), the ES and Theaetetus are committed to grant that existence holds of what does not exist and therefore to apply existence to what does not exist. In this sense, the ES and Theaetetus are obliged to apply being to what is not and they are obliged to apply it ‘often’ because the application is required for false beliefs as well as false statements.

The descriptions of false beliefs. In the course of the argument of 240C7−241A2, Plato offers two descriptions of false beliefs.62 First he offers a ‘simple’ description:

[δ1]63 A false belief believes the things which are not (cf. 240D9−10).

Simple descriptions of false beliefs occur elsewhere in Plato, both in the Sophist (260C3−4) and in other dialogues (cf. Phdr. 262B2−3; Tht. 167A7−8; 188D3−4; 188D8−10; 189A10−189B6). In our Sophist passage, Plato moves on from the simple description of false belief to a ‘polarized’ description:

[δ2] A false belief believes either that the things which are not are or that the things which are are not (cf. 240E1−9).

To the best of my knowledge, the polarized description of false beliefs does not occur elsewhere in Plato. As I pointed out earlier, the polarized description brings in a distinction between affirmative and negative beliefs.

62 Cf. Szaif (1998), 406; Notomi (1999), 187.

63 The occurrences of ‘δ’ at the beginning of the codes abbreviate ‘do/ca’, those of ‘λ’ abbreviate ‘lo/goj’. The asterisk marks the descriptions that do not occur in our Sophist passage.

44

Plato does not describe true beliefs. Had he done so, he would probably have introduced a simple as well as a polarized description corresponding to those of false beliefs:

[δ3]* A true belief believes the things which are.

[δ4]* A true belief believes either that the things which are not are not or that the things which are are.

The simple description of true beliefs occurs elsewhere in Plato’s works (cf. R. III 413A8; Tht. 171A9; 199B8−9). The polarized description, to the best of my knowledge, is never explicitly formulated by Plato.64

The descriptions of false statements. When he comes to false statement, Plato omits the simple description:

[λ1]* A false statement states the things which are not.

The simple description of false statements occurs elsewhere in Plato, both in the Sophist (260C3−4) and in other works (cf. Euthd. 284C2−5; 286A2−3; Cra. 385B10; 429D5−6). In our passage, instead of the simple description Plato offers a polarized one:

[λ2] A false statement states either that the things which are are not or that the things which are not are (cf. 240E10−241A2).

A (partial) polarized description of false statements occurs also later on in the Sophist (263B7−10).65 As in the case of beliefs, both the simple and the polarized description of true statements are omitted:

[λ3]* A true statement states the things which are.

[λ4]* A true statement states either that the things which are are or that the things which are not are not.

The simple description of true statements occurs elsewhere in Plato (cf. Euthd. 284A5−8; 284C6; Cra. 385B10; R. III 389C4−5), while the polarized description does not.66

64 The polarized description of true belief is however hinted at by the double occurrence of ‘not to be’ at 240E1.

65 Some would probably say that Cra. 385B7−8 also contains a (partial) polarized description of false statements. However, on my interpretation of this passage (cf. below, n. 45 of Ch. V), its account of true statement is of a rather different sort.

66 Several commentators would probably say that Sph. 263B4−6 and Cra. 385B7−8 contain (partial) polarized descriptions of true statements. However, on my interpretation of these two passages (cf. below, n. 45 of Ch. V and text thereto), the descriptions of true statements they contain are of a rather different sort.

45

Some ancient parallels. If one combines [λ2] with [λ4] and transforms the plural forms into singular ones, the result is Aristotle’s famous definition of truth and falsehood: ‘To state [to\ le/gein] that what is is not or that what is not is [to\ oÄn mh\ eiånai hÄ to\ mh\ oÄn eiånai] is false; to state that what is is and that what is not is not [to\ oÄn eiånai kaiì to\ mh\ oÄn mh\ eiånai] is true’ (Metaph. Γ7, 1011b26–7, cf. Int. 6, 17a26−31).67 The fact that Aristotle defines falsehood before truth might be a trace of, or a tribute to, the Sophist. Note that the order of cases in Aristotle’s definition of falsehood for statements (denials first, affirmations later) is the same as in Plato’s polarized description of false statements at Sph. 240E10−241A1 (for the polarized description of false beliefs Plato adopts the reverse order: affirmations first, denials later, cf. 240E1−9).

A characterization of speaking truly that recalls what we find in our Sophist passage appears in Xenophon’s Anabasis: ‘This man enjoyed the reputation of having spoken truly [a)lhqeu=sai] in many earlier cases of the same sort, saying the things which are as being and the things which are not as not being [ta\ oÃnta te w¨j oÃnta kaiì ta\ mh\ oÃnta wj ou)k oÃnta]’ (IV 4, 15).

Definitions of truth and falsehood ‘by cases’. Our Sophist passage is the earliest text in the extant philosophical literature that presents (albeit implicitly) a definition of truth ‘by cases’, i.e. by specifying a certain condition as necessary and sufficient for the truth of statements of a certain kind and different conditions as necessary and sufficient for the truth of statements of different kinds. Definitions of truth of this sort are still widespread in modern philosophy. The main difference between the modern definitions and their ancient forerunners is that modern definitions of truth are based on recursive techniques (whereby the definientia of the definitions of truth for complex statements of certain kinds mention the truth of statements which are among their components, or are ‘simpler’ in some other rigorously specified way) whereas the ancient definitions in Plato and Aristotle (often repeated with greater or smaller variations by later thinkers influenced by them) rely on the distinction of two classes of statements, namely affirmations and denials, and their definientia do not speak of truth (or at least not of the truth of statements).

Definitions of truth ‘by cases’ may contribute to a correspondence theory of truth. The idea that truth amounts to some sort of correspondence with the world is notoriously vague and can be specified in various ways. If, for the sake of simplicity, we restrict ourselves to beliefs while ignoring statements and we leave on one side the view that for a belief to be true is for it to correspond to facts (because the concept of fact seems to be foreign to ancient reflections on truth), a first step towards a clarification of the idea that truth amounts to some sort of correspondence with the world is to say that a belief is true just in case it believes things to be in the way in which they actually are. A definition of truth by cases may help to make a further step in the direction of clarification: for it can be worked out in such a way that beliefs are sorted into classes according to how they believe things to be and then the beliefs of each class are declared to be true just in case things actually are in a certain way (namely in the way in which the beliefs of

67 Cf. Przełęcki (1981), 124; Movia (1991), 237; Szaif (1998), 407; Notomi (1999), 189; Fronterotta (2007), 332.

46

the given class believe things to be).68 Note that later in the Sophist Plato describes a true statement by saying that it ‘states the things which are as they are’ (263B4).69 Also recall that earlier in the dialogue there was a further indication of a conception of truth as correspondence in the idea (implied, but clearly so) that true beliefs are analogous to faithful imitations of their originals whereas false beliefs are analogous to unfaithful imitations (cf. 235C9−236C8).70

Four readings of the simple descriptions of false and true beliefs and statements. The simple descriptions of true and false beliefs and statements (i.e. [δ1], [δ3], [λ1], and [λ3]) allow at least four different readings. To avoid pointless complications, I concentrate on the first of the simple descriptions, namely [δ1], the simple description of false beliefs.

In the first place, the simple description of false beliefs may be taken to be about states of affairs (or propositions) and to involve the veridical use of ‘to be’:

[δ1.1] A false belief believes in states of affairs which are not the case.

Secondly, the simple description of false beliefs may be taken to involve the predicative use of ‘to be’, with the complement omitted to achieve generality:

[δ1.2] A false belief is about objects which are not ϕ.

Thirdly, the simple description of false beliefs may be taken to be about kinds and to involve the converse of the predicative use of ‘to be’ (whereby it means ‘to hold of ’)71 with the complement dropped out to achieve generality:

[δ1.3] A false belief is about kinds which are not about σ.

Fourthly, the simple description of false beliefs may be taken to involve the existential use of ‘to be’:

[δ1.4] A false belief is about things which do not exist.

A parallel fourfold distinction may be drawn for the other simple descriptions, namely those of true beliefs (i.e. [δ3]) and of false and true statements (i.e. [λ1] and [λ3]).

Four readings of the polarized descriptions of false and true beliefs and statements. The polarized descriptions of true and false beliefs and statements (i.e. [δ2], [δ4], [λ2], and [λ4]) also allow four

68 Cf. Szaif (1998), 501−3. According to Tarski (1944), 117−18, the definition of truth offered by Aristotle at Metaph. Γ 7, 1011b26–7 (cf. above, text to n. 67) amounts to the classical formulation of the correspondence theory of truth.

69 Cf. below, paragraph to n. 45 of Ch. V.

70 Cf. above, paragraph to n. 28 of Ch. I.

71 On the converse use of ‘to be’, cf. below, text to n. 98 of Ch. IV.

47

different readings which correspond to those of the simple descriptions. Again to avoid pointless complications, I concentrate on the first of the polarized descriptions, namely [δ2], the polarized description of false beliefs.

In the first place, it may be taken to be about states of affairs (or propositions) and to involve the veridical use of ‘to be’:

[δ2.1] A false belief believes either that states of affairs which are not the case are the case or that states of affairs which are the case are not the case.

Secondly, the polarized description of false beliefs may be taken to involve the predicative use of ‘to be’, with the complement omitted to achieve generality:

[δ2.2] A false belief believes either that objects which are not ϕ are ϕ or that objects which are ϕ are not ϕ.

Thirdly, the polarized description of false beliefs may be taken to be about kinds and to involve the converse of the predicative use of ‘to be’ with the complement dropped out to achieve generality:

[δ2.3] A false belief believes either that kinds which are not about σ are about σ or that kinds which are about σ are not about σ.

Fourthly, the polarized description of false beliefs may be taken to involve the existential use of ‘to be’:

[δ2.4] A false belief believes either that things which do not exist exist or that things which exist do not exist.

Parallel fourfold distinctions may be drawn for the other polarized descriptions, namely those of true beliefs (i.e. [δ4]) and of false and true statements (i.e. [λ2] and [λ4]). The third reading of the polarized description of false statement is worth spelling out:

[λ2.3] A false statement states either that kinds which are not about σ are about σ or that kinds which are about σ are not about σ.

The interest of [λ2.3] lies in the fact that it is close to the definitive analysis of false statement finally endorsed by Plato at the end of the Sophist ’s core section (262E11−263D5).

The case for the first readings of the descriptions of false beliefs. I make the rather natural assumption that the same kind of reading is to be accepted for the simple and the polarized descriptions. In other words, I assume that for every n (1 ≤ n ≤ 4), reading [δ1.n] is correct just in case [δ2.n] is

48

correct. Which, if any, of the four readings of the simple and polarized descriptions of false beliefs is the one intended by Plato?72

At first glance, the correct reading of the simple description of false beliefs is the first one, namely reading [δ1.1], which brings in states of affairs and the veridical use of ‘not to be’. If this is right then (by the assumption I just made about the same kind of reading being required for the simple and the polarized descriptions) the correct reading of the polarized description of false beliefs should be the first one, namely reading [δ2.1], which also brings in states of affairs and the veridical use of ‘to be’ and ‘not to be’.

The case for the second and the third readings of the descriptions of false beliefs. The introduction of the polarized description of false beliefs may come as a surprise. For, once the ES and Theaetetus have agreed that ‘a false belief […] believes the contraries of the things which are’ (240D6−7) and therefore ‘believes the things which are not’ (240D9), it is surprising that they should go on to say (240E1−4) that not all beliefs which believe the things which are not are false, but only those which believe that the things which are not are. One may get the impression that a change of perspective has taken place in the course of the argument, or that the beginning of the argument had been misleading.

But let us consider the matter more closely, by bringing in the four possible readings of the simple description of false beliefs, namely [δ1.1], [δ1.2], [δ1.3], and [δ1.4]. If we adopt reading [δ1.1], then the introduction of the polarized description of false beliefs really comes as a complete surprise: for, according to reading [δ1.1], the ES and Theaetetus have just agreed that a false belief believes in states of affairs which are not the case. Once they have agreed this, they seem to have no reason for introducing the corresponding reading [δ2.1] of the polarized description of false beliefs, according to which a false belief believes either that states of affairs which are not the case are the case or that states of affairs which are the case are not the case. If we adopt reading [δ1.2], then the introduction of the polarized description not only does not come as a surprise, but is even required. For, according to reading [δ1.2], the ES and Theaetetus have just agreed that a false belief is about things which are not ϕ. One feels the need to specify that a belief about things which are not ϕ is false if it believes that those things are ϕ, and once one has made this move one also feels the need to make it clear that a belief about things which are ϕ is false if it believes that they are not ϕ. A similar argument applies to reading [δ1.3]. In other words, if we adopt reading [δ1.3], then the introduction of the polarized description is not only unsurprising, but even required. For, according to reading [δ1.3], the ES and Theaetetus have just agreed that a false belief is about kinds which are not about σ. One feels the need to specify that a belief about kinds which are not about σ is false if it believes that those kinds are about σ, and once one has made this move one is also inclined to make it clear that a belief about kinds

72 Szaif (1998), 405−9 adopts the readings which involve states of affairs and the veridical use of ‘to be’, namely [δ1.1] and [δ2.1]. McDowell (1982), 132 seems to endorse the readings which involve attributes and the converse of the predicative use of ‘to be’, namely [δ1.3] and [δ2.3]. In Crivelli (1990), 34−6 I chose different readings for the simple description and the polarized description: for the simple description I endorsed the reading which involves states of affairs and the veridical use of ‘to be’, namely [δ1.1], while for the polarized description I favored the reading which brings in the predicative use of ‘to be’, namely [δ2.2].

49

which are about σ is false if it believes that they are not about σ. As for reading [δ1.4], matters are less straightforward: the introduction of the polarized description is not as surprising as in the case of reading [δ1.1], but is not so obviously called for as with readings [δ1.2] and [δ1.3]. So, the introduction of the polarized descriptions of false beliefs suggests that either the second or the third readings of the descriptions of false beliefs, namely readings [δ1.2] and [δ2.2] or readings [δ1.3] and [δ2.3], should be correct. To put it concisely, if the first readings of the descriptions of false beliefs are correct then the introduction of the polarized description is a strange and surprising move; if the second or the third readings are correct then the introduction of the polarized descriptions, far from being surprising, is even required; if the fourth readings are correct then the introduction of the polarized description is not surprising, but is not required either.

The implications of the second and the third readings of the descriptions of false beliefs. I would like to investigate further the implications of the second and the third readings of the descriptions of false beliefs. Consider the way in which the simple description of false beliefs is introduced: the ES and Theaetetus begin by agreeing that ‘a false belief […] believes the contraries of the things which are’ (240D6−7) and they infer that it ‘believes the things which are not’ (240D9). How should these remarks on contrariety and not being be understood, if the second readings of the descriptions of false beliefs are correct? How should they be understood if the third readings are right?

Let us begin by considering the second readings. Suppose that reading [δ1.2] of the simple description of false beliefs is correct, so that reading [δ2.2] of the polarized description of false beliefs is also correct. Then Plato is concentrating on beliefs concerning objects which are not ϕ and the reason why these objects are not ϕ is that they are contrary to what is ϕ. Beliefs of this sort are false if what they believe with regard to the objects they are concerned with is that they are ϕ. Consider, for instance, Jim’s belief that the Taj Mahal is black. This belief is concerned with the Taj Mahal, an object which, being white, is contrary to what is black and therefore is not black. The belief is false because with regard to this object it believes that it is black.

Now let us look at the third readings. Suppose that reading [δ1.3] of the simple description of false beliefs is correct, so that reading [δ2.3] of the polarized description of false beliefs is also correct. Then Plato is focusing on beliefs concerning kinds which are not about σ, and the reason why these kinds are not about σ is that they are contrary to what is about σ. Beliefs of this sort are false if what they believe with regard to the kinds they are concerned with is that they are about σ. Consider again Jim’s belief that the Taj Mahal is black. This belief is concerned with blackness, a kind which is not about the Taj Mahal because it is contrary to whiteness, a kind which is about the Taj Mahal. The belief is false because with regard to the kind it concerns, i.e. blackness, it believes that it is about the Taj Mahal.

The train of thought implied by the third readings might be interpreted in a way that chimes with one of the most widespread exegeses of Plato’s account of false statement at 262E11−263D5. Suppose that the word ‘contrary’ (‘e)nanti¿oj’) in our passage, at 240D6 and 240D8, expresses not polar opposition, namely the relation that obtains between kinds of the same sort which are as different and ‘distant’ as possible (e.g. whiteness and blackness, or largeness and smallness, or hot and cold), but incompatibility, namely a relation obtaining between kinds which cannot both

50

hold together of any object (for instance, incompatibility obtains between whiteness and redness, between yellowness and greyness, and of course also between whiteness and blackness). Then Plato might be hinting that a belief to the effect that a kind K is about an object o is false just in case K is not about o because it is incompatible with some kind which is about o. This would be close to the incompatibility interpretation of Plato’s account of false statement at 262E11−263D5, according to which a statement to the effect that a kind K is about an object o is false just in case K is not about o because it is incompatible with some kind which is about o.73 However tempting this parallel may be, it must be treated with caution. For, first, when he formulates his account of false statement at 262E11−263D5, Plato does not use the word ‘contrary’ (‘e)nanti¿oj’) but the word ‘other’ (‘eÀteroj’), and in the earlier but closely connected passages dealing with negation and not being he uses ‘other’ (‘eÀteroj’) and ‘contrary’ (‘e)nanti¿oj’) to contrast the concepts they express (cf. 257B1−257C4; 258B3−4; 258E6−259A1). Secondly, although in the Sophist the word ‘contrary’ (‘e)nanti¿oj’) is used to express polar opposition between kinds (cf. 247A7; 247B2; 250A8; 255A12; 255B1; 257B4; 259D5), at no point of the dialogue is ‘contrary’ (‘e)nanti¿oj’) used to express incompatibility as a relation obtaining between kinds (although it is used to express incompatibility as a relation obtaining between whole propositions: cf. 238D7; 239A1; 241A8; 241E5; 248D8).

How is a result about what does not exist reached? Concede, for the sake of argument, that any of the first three readings of the descriptions of false beliefs is correct. The question remains of how the argument reaches the conclusion that false beliefs are about what does not exist.

Suppose that the argument relies on the first readings, namely [δ1.1] and [δ2.1], which involve the veridical use of ‘to be’ and ‘not to be’ and bring in states of affairs. Then the argument claims that a false belief believes either that states of affairs which are not the case are the case or that states of affairs which are the case are not the case. The argument then trades either on the fact that ‘to be’ and ‘not to be’ are understood in accordance not with their veridical use but with their existential use (i.e. the crucial move is a sheer equivocation), or on the mistaken assumption that if something is the case then it exists and that if something is not the case then it does not exist (it is the second conjunct of this assumption that is obviously erroneous, while the first conjunct is defensible and probably even endorsed by Plato).74

Suppose instead that the argument relies on the second readings, namely [δ1.2] and [δ2.2], which involve the predicative use of ‘to be’ and ‘not to be’ with omitted complement. Then the argument claims that a false belief believes either that things which are not ϕ are ϕ or that things which are ϕ are not ϕ. In this case, the argument trades either on the fact that the complements of ‘to be’ and ‘not to be’ are omitted, so that their predicative use is easily confused with the existential one, or on the mistaken assumption that if something is ϕ then it exists and that if something is not ϕ then it does not exist (as in the first case, it is the second conjunct of this

73 The incompatibility interpretation of Plato’s account of false statement has encountered widespread favor among commentators: cf. below, n. 27 of Ch. V and text thereto.

74 Cf. below, subsection to n. 182 of Ch. IV.

51

assumption that is obviously erroneous, while the first conjunct is defensible and probably even endorsed by Plato).

Finally, suppose that the argument relies on the third readings, namely [δ1.3] and [δ2.3], which involve the converse of the predicative use of ‘to be’ and ‘not to be’ with omitted complement. Then the argument claims that a false belief believes either that kinds which are not about σ are about σ or that kinds which are about σ are not about σ. In this case, the argument trades either on the fact that the complements of ‘to be’ and ‘not to be’ are omitted, so that the verbs are easily confused with expressions of existence and non-existence, or on the mistaken assumption that if something is about σ then it exists and that if something is not about σ then it does not exist (as in the other two cases, it is the second conjunct of this assumption that is obviously erroneous, while the first conjunct is defensible).

In all cases, a move which it would be sensible for Plato to make in order to block the fallacious inference is clearly to specify that if something is not ϕ it does not follow that it does not exist. As we shall see, this is precisely what Plato does in his account of negation at 257B1−257C4.75

A common trait of the first three readings of the descriptions of false beliefs is that the fallacious step whereby non-existence is introduced occurs after the readings themselves have been presented: in other words, the accounts of false beliefs given in [δ1] and [δ2] are correct and available for later use in the constructive part of the dialogue. Accordingly, the notion of ‘the contraries of the things which are’ (240D6) is not put to use to introduce non-existence.

The case for the fourth readings of the descriptions of false beliefs. The fourth readings of the descriptions of false beliefs are characterized by the existential use of ‘to be’. If the fourth readings are correct then the accounts of false beliefs presented in the argument we are dealing with are paradoxical and are not available for later use in the constructive part of the dialogue: Plato is surely not going to accept [δ1.4] and [δ2.4] as correct accounts of false belief.

One consideration in favor of these readings is that the existential use of ‘to be’ and ‘not to be’ could be indicated by the occurrence of forms of the verb-phrases ‘to be in no way’ (‘mhdamw½j eiånai’), at 240E2 and 240E5, and ‘to be in all ways’ (‘pa/ntwj eiånai’), at 240E5. For, in the aporetic arguments about not being, a form of the verb-phrase ‘to be in no way’ (‘mhdamw½j eiånai’) is used (237B7−8) to introduce the concept of non-existence.76 It may therefore be plausibly assumed that when forms of this verb-phrase reappear in the argument we are presently concerned with, they still introduce the concept of non-existence (note that this verb-phrase never occurs between the aporetic arguments about not being and the argument we are looking at). And, if this is right, the verb-phrase ‘to be in all ways’ (‘pa/ntwj eiånai’), being obviously opposed to ‘to be in no way’ (‘mhdamw½j eiånai’), probably introduces the concept of existence. Note however that if this consideration is sound then one wonders how one should understand the adverb ‘somehow’ (‘pwj’) at 240E1 and 240E3. Perhaps it should be taken as an expression of

75 Cf. below, text to nn. 178 and 180 of Ch. IV.

76 Cf. above, paragraph to n. 41.

52

doubt (‘somehow’ meaning ‘I suppose’ or ‘perhaps’)77 that modifies the whole of the sentences in which it occurs.

Another consideration in support of the fourth readings has to do with the fact that they enable one to establish a connection with the role which Plato attributes to the ‘contrary of what is’ later in the Sophist, at 257B1−257C4. Let me anticipate my interpretation of this difficult and controversial part of the dialogue. Plato remarks that ‘whenever we say “not being”, it seems that we do not call something contrary to what is, but only other’ (257B3−4). Plato is denying that for something not to be ϕ is for it to be contrary to everything that is ϕ (where the occurrence of ‘not to be’ and the occurrence of ‘is’ in penultimate position are either occurrences of forms of ‘not to be’ and ‘to be’ in their predicative use or occurrences of forms of ‘not to be’ and ‘to be’ in their identity use). He is affirming instead that for something not to be ϕ is for it to be other than everything that is ϕ (where the occurrence of ‘not to be’ and the occurrence of ‘is’ in penultimate position are either occurrences of forms of ‘not to be’ and ‘to be’ in their predicative use or occurrences of forms of ‘not to be’ and ‘to be’ in their identity use). The reason why Plato denies that for something not to be ϕ is for it to be contrary to everything that is ϕ is that he wants to avoid a position which would commit him to the claim that what is not ϕ does not exist. For, if something is contrary to everything that is ϕ then it does not exist: to be ϕ is to exist while partaking of ϕness, so that if something is contrary to everything that is ϕ then it does not exist because it is in the state which is as far and opposed as possible with respect to the state of being ϕ, i.e. with respect to the state of existing while partaking of ϕness.78 The fourth readings of the decriptions of false beliefs enable one to find an antecedent of this line of thought. Once Theaetetus has conceded that a false belief believes ‘the contraries of the things which are’ (240D6), namely the contrary of what is the case, or of what is ϕ, or of what is about σ, he is committed to grant that a false belief believes things which do not exist: for, to exist is part of what it is for something to be the case, or to be ϕ, or to be about σ, and being in the contrary state of a given state is being in the state that is as far and opposed as possible with respect to that given state, so that what is contrary to what is the case, or is ϕ, or is about σ, must not exist. Thus, the claim that a false belief believes ‘the contraries of the things which are’ (240D6) immediately yields the claim that a false belief is about things which do not exist, i.e. [δ1.4].

The fourth readings face at least two objections. The first is that Theaetetus has no reason for accepting the first move of the argument, namely the suggestion that a false belief ‘believes the contraries of the things which are’ (240D6−7), if the ‘contraries of the things which are’ are bound not to exist. This objection may be answered by assuming that Theaetetus’ answer is based on understanding ‘contrary’ to mean something like ‘incompatible’, but is then treated as if ‘contrary’ meant ‘polar opposite’. This would in fact be the only invalid move of the argument: once one accepts that false beliefs believe the contraries, i.e. the polar opposites, of the things which are, i.e. of what is the case, or what is ϕ, or what is about σ, one is inescapably saddled with the consequence that false beliefs are concerned with things which do not exist.

77 Cf. Ast (1835−38), s.v. ‘pw½j’ (III 229).

78 Cf. below, subsection to n. 175 of Ch. IV.

53

The fourth readings are exposed to a second objection. If, as the fourth readings assume, the claim that a false belief believes things which do not exist is present already at the beginning of the argument, in the initial simple description of false beliefs, namely in [δ1.4], why introduce the complex polarized description of false beliefs? The argument would already have reached its target, namely the result that false beliefs believe what does not exist, without the need of passing through this further description. The most plausible answer to this objection is that the polarized description of false beliefs enables Plato to locate the non-existent in the item the false belief refers to (be it a state of affairs, an individual, or a kind). By locating the non-existent in the item referred to Plato comes to be in a position to apply the conclusion of the second aporetic argument about not being: for this conclusion denies the possibility of referring to the non-existent (because whatever expression one might use to refer to the non-existent would have to be either in the singular or in the dual or in the plural).

Assessment of the various readings. Which of the four suggested readings of the descriptions of false beliefs should we choose? I suspect that this question relies on a mistaken presupposition. We need not choose one of these readings to the exclusion of the others: a good sophistical argument conceals its fallaciousness under more than one disguise, leaving to the naïve thinker the choice of the trap to fall in. All four readings are allowed and they all lead to the impossible result that some false beliefs are thoughts about what does not exist. However, if one group of readings were to be chosen at the exclusion of the rest, it should be the fourth readings, mainly because of the strong support they get from the text. Perhaps the best way of putting the matter is to say that the fourth readings are those which are most prominent and would most probably be mentioned if an explanation of the argument were called for, while the readings of the other three groups are there for the sophist’s victim to choose if so they prefer.

CHAPTER III PUZZLES CONCERNING BEING

The structure of the section dealing with earlier theories about being. After probing the puzzles that afflict the concept of not being, the ES and Theaetetus turn to examining being. This discussion takes the form of a critical study of earlier theories of being. The section of the dialogue where the ES and Theaetetus undertake this job (242B6−251A4) divides into four subsections: (1) an introduction (242B6−243D5); (2) a criticism of earlier theories that specify how many beings there are (243D6−245E5); (3) a criticism of earlier theories that state what beings are like (245E6−250D4); (4) a conclusion to the effect that being and not being are equally puzzling (250D5−251A4).

The introduction to the discussion of theories about being (242B6−243D5). In the introduction to the section, the ES wonders whether, just as the notion of not being somewhat unexpectedly turned out to be a source of puzzlement, so also the notion of being, of which we think we have a clear grasp, will turn out to be puzzling (cf. 243B7−243C6). We might be in the situation that is typical of the interlocutors of the ‘Socratic dialogues’, where Socrates’ questions lead people to realize that they lack a clear grasp of notions they thought they understood.

A criticism of the theories that specify how many beings there are (243D6−245E5). The ES distinguishes theories that ‘specify just how many beings there are’ (242C5−6) from theories that state ‘what beings are like’ (242C5−6). This distinction is recalled later, at 245E6−246A2, as being between those who ‘make precise reckonings about being and not being’ and those who ‘speak otherwise’.1

Theories that specify how many beings there are are discussed first (243D6−245E5). This discussion begins with (1) a criticism of the pluralists (243D6−244B5). This is followed by (2) a criticism of the monists (244B6−245D11): two arguments are developed, one (244B6−244D13) dealing with the concept of naming, the other (244D14−245D11) dealing with the concept of a whole. Finally we are offered (3) a short conclusion (245D12−245E5). Plato’s criticism of earlier ontological views was probably a model for Aristotle’s method in the first books of the Physics and the Metaphysics (Aristotle also reports the views of his predecessors about what is and distinguishes between monists and pluralists).

The criticism of the pluralists. The position of the pluralists comes under attack (243D6−244B5). The pluralists addressed by Plato hold that there are exactly two beings, namely the hot and the cold (243D7−8) (the argument is supposed to be able to be applied with respect to any pair of contraries). Plato asks what the item is which is expressed by ‘to be’ and is said to hold of both the

1 Cf. Frede (1996), 186; Carchia (1997), 74; Harte (2002), 100.

55

hot and the cold when one says that both the hot and the cold are. Three alternatives are put forward and then discarded.

(α) (243E3−4) the item expressed by ‘to be’ is distinct from both the hot and the cold. This alternative is discarded because, if it were true, there would be not two but three beings.

(β) (243E4−7) The item expressed by ‘to be’ is one of the hot and the cold. This alternative is rejected because ‘by calling one of the two a being you no longer say that both are in the same way [ou) ga/r pou toiÍn ge duoiÍn kalou=ntej qa/teron oÄn a)mfo/tera o(moi¿wj eiånai le/gete]: for in both ways they would probably be one, not two’ (243E4−6).

(γ) (243E8−244A3) The item expressed by ‘to be’ is identical with both the hot and the cold. This alternative is discarded because it implies that the two are one.

Two interpretations are possible of the way in which alternative (β) is rejected. According to the first exegesis, the reason why being cannot be identical with either the hot or the cold is that if it were identical with either of them then since ‘being’ is said of both the hot and the cold in the same way, both the hot and the cold would be identical with being and would therefore be identical with one another, so that the sum total of beings would again be one and not two.2 According to the second interpretation, the reason why being cannot be identical with either the hot or the cold is that if it were identical with either of them then the other would not be (because by being it would enjoy its own contrary, which is impossible), so that the sum total of beings would be one and not two.3

In favor of the first exegesis speak both the fact it gives an important role to the presence of ‘in the same way’ (‘o(moi¿wj’, 243E5) and the circumstance that it makes the argument readily applicable to any theory to the effect that there are exactly n beings (for any n ≥ 2). An interesting feature of the first interpretation is that it takes Plato to be drawing a distinction that is similar to that between statements of identity and statements of predication (because the situation where what the expression ‘being’ stands for is identical with an item is contrasted with the situation where what the expression ‘being’ stands for holds of that item): the distinction between statements of identity and statements of predication will play an important role later in the dialogue.4

Nevertheless, the second exegesis is to be preferred. For later in the Sophist (249E6−250A7) Plato draws an analogy between a criticism of a characterization of being according to which being embraces both change and stability, on the one hand, and his earlier criticism of the pluralists, on the other. The first part of this criticism of the characterization of being (250A8−250C5) is in fact very similar to the criticism of the pluralists if the second interpretation of the rejection of alternative (β) is right: it is much less similar if the first interpretation is correct. This piece of internal evidence tips the scale in favor of the second interpretation.

2 Cf. Moravcsik (1962), 29.

3 Cf. Cornford (1935), 220; Bordt (1991), 501.

4 Cf. Malcolm (1983), 118−19; Roberts (1986), 231.

56

The first argument against the monists (244B6−244D13). The monists say ‘Only one is’ (244B9−11). In saying so, they apply the expression ‘being’ to something (244B12−13). This raises the question (244C1−3) whether ‘being’ and ‘one’ are two names applied to one and the same thing. This question cannot be easily answered by the monists (244C4−7).

To begin with, the monists cannot give what might seem the most straightforward answer, namely that ‘being’ and ‘one’ indeed are two names applied to one and the same thing. This option is precluded to them because by admitting that there are5 two names they would be blatantly contradicting their own admission of no more than one thing (244C8−10).

But matters are even worse because the very notion of a name also turns out to be problematic for the monists (244C11−244D13). For there are three possibilities. The first is that the name applies to something distinct from itself. In this case there are at least two beings, contrary to what the monists maintain. The second possibility is that the name applies to nothing. This is unacceptable because it clashes with the nature of a name. The third possibility is that the name applies to itself only. In this case it will follow that ‘the one will be a name of the one and will also be the one of the name [kaiì to\ eÀn ge, e(no\j oÃnoma oÄn kaiì tou= o)no/matoj auÅ to\ eÁn oÃn]’ (244D11−12).6 In other words, there will still be a duality within the one—a duality of aspects, as it were: it will be both the name of the one and the one named by the name.

The second argument against the monists (244D14−245D11) involves the concepts of whole and part. It relies on the assumption that according Parmenides being is a whole (I shall not address the historical question of whether Parmenides did endorse such a view). The argument divides into two sections: the first is an introduction to the concepts on which the argument turns (244D14−245B3) while the second contains the actual reductio ad absurdum grounded in the foregoing clarifications (245B4−245D11).

In the first section of the argument the ES quotes three lines from Parmenides’ poem (DK B 8, 43−5 = Sph. 244E3−5). On the basis of this textual evidence he argues that Parmenides is committed to the view that being is a whole composed of many parts and is one because the attribute of unity holds of all its parts. Nevertheless, being is distinct from the attribute of unity because ‘what is truly one [to/ ge a)lhqw½j eÀn] must be called completely partless [a)mere\j pantelw½j]’ (245A8−9). This justification of the claim that being is distinct from the attribute of unity is not pellucid. What may be plausibly assumed is that the justification is an argument that being is distinct from the attribute of unity because being has parts whereas the attribute of unity does not. This however leaves one question open: why does the attribute of unity have no parts? One possible answer is that if the attribute of unity had parts then a second attribute of unity would be needed to explain why the first attribute of unity is one despite its having parts, so that an infinite regress would get going. This answer has two drawbacks: it is somewhat far-fetched and it attributes no role to the ES’s hint that the attribute of unity must be ‘truly one’. A second

5 Taking the ‘eiånai’ at 244C8 as existential: Bordt (1991), 504 regards it as expressing identity.

6 I follow Burnet’s text, which does not coincide with any of the readings handed down by the main witnesses and involves emendation. Several different emendations and interpretations have been offered.

57

possible and more plausible answer to the question why the attribute of unity has no parts takes this hint very seriously. According to this second answer, the attribute of unity is one to such an eminent degree as to exclude the possibility of having parts because if it had them then it would enjoy some form of plurality.7 If this second answer is right then the argument presupposes some version of the Self-Predication Principle (according to which every kind enjoys to an eminent degree the characteristics which it causes perceptible individuals that partake of it to have).8 To the extent that a monist might reject the claim that the attribute of unity is ‘truly one’ (and therefore dismiss the Self-Predication Principle), this juncture of the argument is a weak spot.9 The claim that the attribute of unity is other than being while holding of it foreshadows the distinction between statements of identity and statements of predication.10

The second section of the argument, which contains the reductio ad absurdum, opens with a dilemma (245B4−6): ‘Either being will be one and a whole by having the attribute of unity, or we must completely deny that being is a whole’ (245B4−5). Theaetetus regards this as ‘a difficult choice’ (245B6). At 245B7 the ES agrees with Theaetetus that the choice is difficult and proceeds to explain why it is—in fact, the remainder of the argument’s second section (245B7−245D11) is an explanation of why the choice of either horn is problematic. Note that the second horn of this dilemma, i.e. the alternative that being is not a whole, is the negation of the position attributed to Parmenides at the beginning of the refutation. Plato probably does not want this argument to be merely an attack on a particular version of Parmenides’ monism: the main role of the position attributed to Parmenides at the beginning is to provide the concepts with which the attack will be framed.11

The first horn (245B7−10) is excluded on the grounds that if it were true, it would follow that there are at least two distinct items, i.e. being and the attribute of unity that holds of it. We are not told why being would be distinct from the attribute of unity. The reason is probably to be retrieved from the argument’s first section (cf. 245A1−245B3): being has parts whereas the attribute of unity does not. The consequence that there are at least two distinct items is unacceptable for a monist.

The dilemma’s second horn (245C1−245D11), according to which it is not the case that being is a whole by having the attribute of unity,12 is addressed by means of a subordinate dilemma. In the first horn of the subordinate dilemma (245C1−10) it is assumed that the attribute of

7 Cf. Prm. 129B1−130A2.

8 Cf. below, n. 47 of Ch. IV and text thereto.

9 Cf. Harte (2002), 109, 111.

10 Cf. Prm. 149C5−6; 157E5−158B4; Cherniss (1957), 258; Roberts (1986), 231; Bordt (1991), 506, 513, 521.

11 Cf. Harte (2002), 103.

12 I take ‘u(p' e)kei¿nou pa/qoj’ (245C1) to stand for ‘pa/qoj … tou= e(no/j’ (245B4, 245A1) (cf. Cornford (1935), 225; Arangio-Ruiz (1951), 152). I also regard the ‘mh/’ (245C1) as governing the whole clause ‘vÅ … oÀlon dia\ to\ peponqe/nai to\ u(p' e)kei¿nou pa/qoj’ (245C1−2).

58

wholeness exists. In this case, the monist faces two unpalatable consequences. The first (245C1−7) is that since being is not a whole, ‘being will turn out deprived of itself ’ (245C2−3), so that ‘being will be a not being’ (245C6),13 which yields a contradiction. We are not told on what grounds the hypothesis that being is not a whole should entail that being lacks itself. Two explanations are possible: the first is that being should lack being because it lacks the attribute of wholeness, which is a being,14 the second is that being should lack itself because it is not something whole, i.e. complete.15 In neither case is the argument valid: the phrase ‘to lack being’ can mean both ‘to miss a part of being’ and ‘not to be’, and the argument seems to rely on a shift from the first to the second meaning. Plato is perhaps entitled to employ an argument of this sort because his opponents are not in a position to draw the distinction that would show the argument’s invalidity. I have a slight preference for the second explanation because it enables one to reach the conclusion that ‘being will turn out deprived of itself ’ (245C2−3) also on the assumption, made in the second horn of the subordinate dilemma, that the attribute of wholeness does not exist (the first explanation does not allow this): as we shall see, the ES does commit himself to the view that also on the second horn of the subordinate dilemma the consequence follows that ‘being will turn out deprived of itself ’ (245C2−3).

The first horn of the subordinate dilemma entails the further unpalatable consequence that ‘all things become more than one because being and the whole have each their peculiar nature separately’ (245C8−9). The conclusion that ‘all things become more than one’ (245C8) is of course unacceptable for a monist. But how is this conclusion supposed to be entailed by the premiss that ‘being and the whole have each their peculiar nature separately’ (245C8−9)? One possible explanation is that the premiss that ‘being and the whole have each their peculiar nature separately’ (245C8−9) is supposed to entail that being and the attribute of wholeness are two distinct items because they have different natures. We are not told how the nature of being will differ from that of the attribute of wholeness: the most plausible explanation is that being is not a whole (this is the hypothesis of second horn of the main dilemma) whereas the attribute of wholeness is a whole (by a version of the Self-Predication Principle). A second possible explanation is that the premiss that ‘being and the whole have each their peculiar nature separately’ (245C8−9) is reminding us that it is by lacking the attribute of unity that being fails to be a whole (cf. 245C1−2): since it lacks the attribute of unity, it follows that being is more than one (in the first section of the argument, at 244E2−245A4, the ES and Theaetetus had argued that Parmenides is committed to the view that being has many parts that are unified by enjoying the attribute of unity, but the hypothesis which is now being addressed denies that being enjoys the attribute of unity). I have a slight preference for the second explanation because it enables one to reach the conclusion that ‘all things become more than one’ (245C8) also on the assumption,

13 I take ‘ou)k’ (245C6) to govern ‘oÃn’ (245C6) (cf. Arangio-Ruiz (1951), 152). Cornford (1935), 225 takes it instead to govern ‘eÃstai’ (245C6): there will be a being (i.e. the attribute of wholeness) which being is not.

14 Cf. Cornford (1935), 225.

15 Cf. Harte (2002), 103. Parmenides himself (DK B 8, 32−3) asserted that being is not incomplete and lacks nothing.

59

made in the second horn of the subordinate dilemma, that the attribute of wholeness does not exist (the first explanation does not allow this): as we shall see, the ES does commit himself to the view that also on the second horn of the subordinate dilemma the consequence follows that ‘all things become more than one’ (245C8).

In the second horn of the subordinate dilemma (245C11−245D11) it is assumed that the attribute of wholeness does not exist. In this case, we are told (245C11−245D1), being will have the same characteristics which it was shown to have on the first horn. No explanation is given of why it is bound to have these same characteristics. As for the first, i.e. its not being, the most plausible guess is that even in the case that the attribute of wholeness does not exist, being still fails to be whole, i.e. complete, so that being lacks itself and therefore is not. The second characteristic, i.e. being more than one, may attributed to being on the basis of the hypothesis that being is not a whole by lacking the attribute of unity.

However, the second horn of the subordinate dilemma is said to entail additional absurdities (245D1−11). On top of its being the case that being is not, the second horn of the subordinate dilemma also allegedly entails (245D1−7) that being will never come into being because what comes into being does so as a whole. Moreover (245D8−11) whatever is not a whole will not be of any definite size: for whatever is of a definite size is so as a whole. Since the attribute of wholeness does not exist, nothing will enjoy it (for only existent attributes can be enjoyed, cf. 247A5−247B3), so that nothing will be a whole, whence nothing will be of any definite size. This consequence is probably supposed both to be absurd in its own right and to clash with a view to which Parmenides is committed: at the beginning of the refutation (244E3−5), lines of his poem had been quoted where the one being was compared with a ‘well-formed sphere, equal-balanced all ways from the middle’ (DK B 8, 43−4). Moreover, the results just stated with regard to coming-to-be and size may be extended to all those attributes which things enjoy only as wholes. If all attributes are only enjoyed by things as wholes, it will follow that nothing enjoys any attribute.16

The criticism of theories that state what beings are like (245E6−250D4). The subsection of the dialogue dealing with theories that state what beings are like (245E6−250D4) divides as follows: (1) a transition from the discussion of theories that state how many beings there are to that of theories that state what beings are like (245E6−246C8); (2) a description of a ‘battle of the gods and the giants’ (246A4−246C8); (3) the questioning of the giants (246C6−248A3); (4) the questioning of the gods (248A4−248E6); (5) a new characterization of being (248E7−249D8); (6) a criticism of the new characterization of being (249D9−250D4).

The battle of the giants and the gods. Plato describes a fight between certain ‘giants’ and certain ‘gods’, or ‘friends of the forms’. The giants hold that ‘only what can be touched and offers resistance is’ (246A11−246B1) and they propose a definition whereby ‘body and being are the

16 Cf. Bordt (1991), 509−10.

60

same’ (246B1).17 The friends of the forms maintain that only certain intelligible forms are: they contrast being (what is always in the same way) with becoming (what becomes different at different times) and they claim that we communicate with being by means of the soul through reasoning whereas we communicate with becoming by means of the body through perception.

Scholarly debate has raged about the identity of the gods, or friends of the forms. Some commentators think that they defend Plato’s own ‘classical theory of forms’ and that Plato therefore in the Sophist has at least modified, if not abandoned, this theory.18 There is in fact a striking resemblance, even in terminology, between the position of the friends of the forms and the ‘classical theory of forms’ as it is presented in the Republic and the Timaeus.19 Since the Timaeus was traditionally held to be one of Plato’s last dialogues, some commentators argue that it really belongs to Plato’s middle period.20 This revised dating does not however meet with universal approval.21 As we shall see, there is at least one important difference between the position of the friends of the forms and the ‘classical theory of forms’.22 Other commentators instead hold that the gods or friends of the forms represent some other philosophical school, e.g. the Megarians,23 yet others maintain that they are members of the Academy who defended views close but subtly different to Plato’s.24

The cross-examination of the giants. The ES and Theaetetus question both parties and try to bring them to an agreement. They manage to convince (a softened version of ) the giants to admit that certain incorporeal entities exist. Theaetetus acts as a spokesman for the giants, who admit (246E2−247C8) that (1) there are mortal animals, (2) they are ensouled bodies, (3) souls exist, (4) some souls are just and others unjust and some souls are wise and others unwise, (5) just and wise souls are such by the possession and presence of justice and wisdom whereas unjust and unwise souls are such in virtue of the contrary qualities, (6) what is capable of coming to be in something and going away from it exists, (7) justice and wisdom and their contraries therefore exist although they are incorporeal whereas the soul, which also exists, is a body.

Now that they have granted that both bodies and incorporeal things exist, the giants are asked (247C9−247D4) what the being is which is shared by things that are so different. The ES and Theaetetus agree (247D4−248A3) that they will probably accept the following modal

17 Cf. Bordt (1991), 513; Brown (1998), 186.

18 Cf. Grote (1875), II 458; Cornford (1935), 247; Ross (1951), 107; Frank (1985), 9; McPherran (1986), 244.

19 Cf. below, n. 42 and text thereto.

20 Cf. Owen (1953); Ryle (1967), 320.

21 Cf. Cherniss (1957).

22 Cf. below, text to n. 29.

23 Cf. Campbell (1867), Sph. 125−6.

24 Cf. Gerson (2006), 291−2.

61

characterization of being: to be is to have either the power of affecting something or that of being affected by something.

In order to understand what exactly the modal characterization of being put forward by the ES amounts to, we must clarify what it is for something to affect something and what it is for something to be affected by something. Some commentators favor a weak account: they hold that whenever an object x enjoys a property y, y affects x and x is affected by y.25 This interpretation is implausible because if it were correct then the giants’ acceptance of the modal characterization of being would amount to a wholesale abandonment of their position. We need a more meaty account of what it is for something to affect something and what it is for something to be affected by something. One plausible candidate is: for x to affect y is for x to exert a quasi-causal power on y whereby y is made to be in a certain way. The ES is probably presupposing that such incorporeal items as justice, wisdom, and the remaining virtues will count as beings according to this modal characterization: for justice is what makes just people just, wisdom is what makes wise people wise, and similarly with the other virtues (cf. Prt. 332A4−332C3; Men. 72C6−72D1), so that justice, wisdom, and the other virtues may be described as having the power of affecting things in that they have the quasi-causal power of making things be in certain ways.26 The forms of Plato’s ‘classical theory of forms’ would also probably count as beings by the modal characterization of being: forms do have the power to affect things in that they have the quasi-causal power of making things be in certain ways (cf. Phd. 100C9−100E7). This is not to say that by accepting the modal characterization of being the giants are committed to endorsing the forms of Plato’s ‘classical theory of forms’: the giants can still deny that there is any such thing as forms of this sort.

The cross-examination of the friends of the forms. Having convinced the giants to accept the modal characterization of being, the ES tries this characterization out on the friends of the forms (248A4−248E6). Since they maintain that we communicate with becoming by means of the body through perception and with being by means of the soul through reasoning (248A10−248B1), the friends of the forms are asked what this communication is which we have with both becoming and being (248B2−3). Specifically, they are asked whether this communication is a case of affecting or of being affected (248B3−6). The friends of the forms are thereby invited to endorse the modal characterization of being on the basis of their epistemological view that we do cognitively communicate with objects.27

The friends of the forms react (248B6−248C9) by rejecting the modal characterization of being. They recognize that the modal characterization of being would make the things of the world of becoming into beings because the things of the world of becoming have both the power

25 Cf. Moravcsik (1962), 37. Plato does use ‘to be affected by’ (‘pa/sxein’) to say of an object that it enjoys a property: cf. Sph. 245A5. Künne (2004), 309 takes a line which is close, perhaps identical, to that of Moravcsik.

26 Cf. Brown (1998), 199.

27 Cf. Politis (2006), 158, 159.

62

of affecting something and that of being affected by something (248C7−8). This is perhaps for them already one reason for rejecting the modal characterization of being: they do not want the things of the world of becoming to rank as beings.28 But their stronger reason for rejecting the modal characterization of being is that in their view beings, namely kinds, have neither the power of affecting something nor that of being affected by something (248C8−9). Note that by saying this the friends of the forms reject one of the central tenets of Plato’s ‘classical theory of forms’: for according to Plato’s ‘classical theory of forms’, forms do have the power to affect things in that they have the quasi-causal power of making things be in certain ways (cf. Phd. 100C9−100E7).29

Theaetetus asks whether this position of the friends of the forms is sensible (248C10). More probing takes place. The friends of the forms ‘concede that the soul knows and being is known’ (248D1−2). But they assert that both to know and to be known are neither cases of affecting nor cases of being affected, and they make this assertion because otherwise they would contradict their earlier claims (248D4−9). Further explication is offered of the reasoning of the friends of the forms (248D10−248E6).30 The explication takes the form of an examination of what would follow if one of the positions rejected by the friends of the forms were to be correct. In particular, the consequences are examined of admitting that to know is a case of affecting. If to know were a case of affecting, then what is known would necessarily be affected. So, since it is known by knowledge, being would be affected, and therefore change, in so far as it is known. But this could not happen to what is stable (and, of course, the friends of the forms take being to be stable, cf. 248A11−12).

So, the friends of the forms hold that being has neither the power of affecting something nor that of being affected by something. Hence they reject the modal characterization of being. They acknowledge that being can be known by the soul but they deny that to know something is to affect it and on these grounds they deny also that to be known by something is to be affected by it. So, their admission that being can be known by the soul does not commit them to the claim that being has the power of being affected by something.

After being rejected by the friends of the forms, the modal characterization of being disappears from the dialogue.31 Does Plato himself adopt the position he attributes to the friends of the forms? If he does then his view is that the modal characterization of being is to be abandoned and kinds have neither the power of affecting something nor that of being affected by something

28 Cf. Künne (2004), 310−11.

29 Cf. Künne (2004), 311.

30 I regard lines 248D10−248E6 as an explication of the friends of the forms’ reasoning that leads them to assert that both to know and to be known are neither cases of affecting nor cases of being affected. Some commentators (e.g. Jowett (1892), IV 380; McPherran (1986), 244−5) interpret these lines as an attack on the position of the friends of the forms pushing them to concede that being changes in so far it is known.

31 Pace Malcolm (1983), 120−1, there is no evidence that the friends of the forms accept or are committed to the modal characterization of being.

63

(kinds can be known but to know something is not to affect it).32 However Plato could tacitly endorse a different strategy whereby he retains the modal characterization of being. In fact, Plato could adopt at least three alternative strategies whereby he retains the modal characterization of being and is therefore committed to accept that kinds have either the power of affecting something or that of being affected by something. The first alternative strategy is to claim that by knowing the kinds the soul (does not affect but) is affected by them, so that the soul changes whereas the kinds remain changeless. The basic assumption is that if x is affected by y then x changes but it does not follow that y changes.33 The second alternative strategy is to accept that by being known the kinds are affected by the soul but to deny that kinds are thereby changed. In this case, the basic assumption is that if x is affected by something it does not follow that x changes.34 The third alternative strategy is to accept that by being known kinds are affected by the soul and therefore do change. This third strategy can be implemented in two ways. The first is to credit Plato with the view that by being known kinds undergo an intrinsic and genuine change.35 One will thereby assume that Plato’s ‘classical theory of forms’ undergoes a radical transformation. The second way of implementing the third strategy is to hold that according to Plato kinds by being known undergo (not an intrinsic and genuine change but) only an extrinsic and ‘mere Cambridge’ change, which does not affect their remaining intrinsically and genuinely changeless. (‘Mere Cambridge’ change is that spurious change which for instance Socrates undergoes by being at one time taller and at another time smaller than Theaetetus, the situation being one where Socrates suffers no decrease in size whereas Theaetetus grows—‘mere Cambridge’ change is discussed at some length in the Theaetetus, at 154B1−155C10.) This second way of implementing the strategy probably involves no reform of the ‘classical theory of forms’ and the claim that kinds suffer an extrinsic and ‘mere Cambridge’ change that leaves them intrinsically and genuinely changeless could perhaps be accepted even by the friends of the forms.36

Whatever the merits of these alternative strategies, it is clear that the friends of the forms do not take them up: as I said, they adamantly reject the modal characterization of being, which henceforth disappears from the dialogue. So, the attempt to reconcile the friends of the forms and the giants by means of the modal characterization of being runs in the sand.

32 Cf. Ross (1951), 110.

33 Cf. Cornford (1935), 240; Ross (1951), 111; Brown (1998), 199−201.

34 Cf. Vlastos (1970), 309−17.

35 Cf. Teloh (1981), 194−5; Brunschwig (1988), 122; Gerson (2006), 296−302.

36 Cf. Moravcsik (1962), 39−40; Runciman (1962), 81; Owen (1966), 338−9; Reeve (1985), 53−4, 60−1; McPherran (1986), 244−50; Künne (2004), 316−20.

64

The last characterization of being: change and stability both are.37 In order to break the deadlock, a fresh start is made with an independent argument aimed at providing a new characterization of being (248E7−249D8). This argument, addressed mainly to the friends of the forms, relies on the principle that all that is required for understanding (fro/nhsij or nou=j) or knowledge (e)pisth/mh) must be admitted to be: after all, the friends of the forms do concede that being is known by the soul (cf. 248C11−248D3) and philosophers, or ‘lovers of wisdom’, venerate understanding and knowledge and will therefore fight against whoever banishes either of them (cf. 249C6−11). On the one hand (248E7−249B7), understanding belongs to the domain of what is in the most complete way (t%½ pantelw½j oÃnti … pareiÍnai, 248E8−249A1). But understanding requires life, which requires soul, which in turn requires change. Therefore not only understanding but also life, soul, and change belong to the domain of what is in the most complete way. ‘Hence both changing things [to\ kinou/menon] and change [ki¿nhsin] must be acknowledged as beings’ (249B2−3).38 On the other hand (249B8−249C5), stability must also be because the condition of ‘being in the same way and in the same manner and about the same [to\ kata\ tau)ta\ kaiì w¨sau/twj kaiì periì to\ au)to/]’ could not be satisfied without something being stable (249B12−249C2) and if that condition were not satisfied then there would be no understanding (249C3−5).39 Therefore (249C6−249D8) being must be conceived of in such a way that both all changing things and all stable things are.

How is the universal conclusion reached that both all changing things and all stable things are? After all, what had been established in the earlier stages of the argument was merely that some changing things are and some stable things are. The most plausible justification for the universal conclusion is that the obstacles which could have refrained one from asserting it have been removed. For, the claim that all changing things are is prima facie plausible; the only ground for the friends of the forms’ rejection of it in favor of its contrary, the claim that no changing things are, is that all changing things change and change disqualifies them from being; once it has been acknowledged that some changing things are, this ground subsides; the prima facie plausible

37 I translate ‘ki¿nhsij’ and ‘sta/sij’ by, respectively, ‘change’ and ‘stability’ (similarly with the corresponding verbs). Translators often render ‘ki¿nhsij’ and ‘sta/sij’ by ‘motion’ and ‘rest’. The advantage of the translations ‘change’ and ‘stability’ is that they cover not only locomotion and stillness, but also every kind of alteration and lack thereof (cf. Tht. 181C1−181D7).

38 I take it that at 248E7−249B7 Plato establishes that change and all changing things are by arguing that the soul and therefore change belong to the domain of what is in the most complete way (cf. Cornford (1935), 244−7; Ross (1951), 108−11). At least two other interpretations have been offered of 248E7−249B7. According to some commentators, at 248E7−249B7 Plato establishes that change and changing things are by insisting that kinds are changed insofar as they are known (because souls undergo change insofar as they know and they communicate this change to the objects of their knowledge, i.e. kinds) (cf. Moravcsik (1962), 39). Other interpreters believe that in the passage in question Plato establishes that change and changing things are by asserting either that kinds themselves change in so far as they are living beings endowed with understanding (perhaps intellects) (cf. Apelt (1891a), 78−9) or that the system of all kinds is a living being endowed with understanding (cf. de Vogel (1953), 65−7).

39 For the precise interpretation of 249B12−249C5 cf. below, paragraph to n. 50 of Ch. IV.

65

claim that all changing things are may therefore be endorsed after all. A parallel justification can be given for the claim that all stable things are.40

By the end of the discussion with the giants and the friends of the forms, both parties have made concessions whereby their original ontology is expanded: the giants have acknowledged that there are incorporeal items, while the friends of the forms have granted that changing things also are. To this extent, an agreement between the two parties has been reached.41

One of the claims made by Plato in this section of the dialogue seems inconsistent with the metaphysical picture he draws elsewhere. For in several other works Plato appears to endorse a position that is close to the one of the friends of the forms in that he sets the domain of forms in contrast with that of perceptible particulars and he characterizes the former as the realm of being and the latter as the realm of becoming.42 At the present point of the Sophist ’s argument Plato instead seems to claim that all changing things are beings.43 Whether this inconsistency may be described as an evolution in Plato’s metaphysics depends, at least in part, on issues of relative chronology of his works—issues I prefer to keep clear of in the present context.

The refutation of the last characterization of being. But the last characterization of being also comes under attack (249D9−250D4). The ES says (249E6−250A2 and 250A4−6) that it is open to a criticism analogous to the one leveled earlier at the position of the pluralists (cf. 243D6−244B5). Here is a step-by-step reconstruction of the criticism.

First step (250A8−10): change and stability are opposed to one another. Second step (250A11−250B1): change and stability both are (this is the view that is being

criticized). Third step (250B2−250C2): being cannot be identified with change, nor can it be identified

with stability: in the first case, the fact that stability is would imply that stability changes, which is impossible because change and stability are opposed to one another; in the second case, the fact that change is would imply that change is stable, which is again impossible because change and stability are opposed to one another.

Fourth step (250C3−5): being is not both change and stability together but is other than them. Fifth step (250C6−8): by virtue of its own nature, being neither is stable nor changes. Sixth step (250C9−250D4): being neither changes nor is stable—which is absurd.

40 Cf. Brown (1998), 204.

41 Cf. Brown (1998), 202−3; Politis (2006), 155−8 (who discusses how far the simmetry between the concessions made by the giants and those of the friends of the forms goes).

42 Cf. R. VI 509B1−509C2; VII 526E7−8; 534A2−4; Ti. 27D5−28A4.

43 Cf. Owen (1966), 339; Seligman (1974), 37; Teloh (1981), 195; Brown (1998), 204. Several commentators deny that the position presented in the Sophist is incompatible with the metaphysical picture of other dialogues. Some (e.g. Cornford (1935), 246−7; Ross (1951), 110; Guthrie (1978), 145; Frank (1985), 15−18; Silverman (2002), 155, 343) think that in the Sophist being comprises kinds and souls (which change) but not bodies: but such an interpretation cannot be easily reconciled with 249D3−4 and 250B8−9 (cf. Politis (2006), 173−4).

66

The claim (made in the third step) that the reciprocal opposition of change and stability rules out the possibility that change could be stable raises difficulties which I shall address later in connection with a similar argument concerning change, stability, sameness, and otherness (255A4−255B7).44

The wording of the fourth step suggests that the claim it makes ought to contradict the characterization of being endorsed earlier by the ES and Theaetetus. For the fourth step is worded as follows: ‘Being is not both change and stability together [sunamfo/teron]’ (250C3−4). And the wording of the characterization of being endorsed earlier by the ES and Theaetetus is: ‘Whatever things are unchanging and changing, being and the totality of reality should be said to be both together [sunamfo/tera]’ (249D3−4). The occurrence of ‘both together’ (‘sunamfo/teron’) in the fourth step seems to pick up the occurrence of ‘both together’ (‘sunamfo/tera’) in the characterization of being endorsed earlier by the ES and Theaetetus, and this makes it sound as if a contradiction is intended.45 But, if a contradiction is indeed intended, the argument fails. For what the ES and Theaetetus have established in the fourth step by saying that ‘being is not both change and stability together’ (250C3−4) is the claim that being is other than both change and stability, whereas when they said that ‘whatever things are unchanging and changing, being and all that is should be said to be both together’ (249D3−4), the claim they endorsed was that both all changing things and all stable things are: the two claims are neither contradictory nor incompatible. It may well be the case that a contradiction is intended and that the argument fails (as we shall see, the continuation of the argument after the fourth step is also invalid). On the other hand, since the argument does not stop with the fourth step but goes on, it may be the case no contradiction is intended (for, had it been intended, the argument could very well have finished with the fourth step).

The inference from the fourth through the fifth to the sixth step is invalid. In the fourth step, the starting point of this inference, we have a negative statement of identity: ‘Being is not both change and stability together’ (250C3−4) (which means the same as ‘Being is neither change nor stability’). In the sixth step, the conclusion of the inference, we find instead a negative statement of predication: ‘Being neither changes nor is stable’. The role of the intermediate fifth step, ‘By virtue of its own nature, being neither changes nor is stable’, is probably to provide a ‘middle term’ that eases this invalid slide. For, on the one hand, the sentence ‘Being is neither change nor stability’ entails the sentence ‘By virtue of its own nature, being neither changes nor is stable’ taken in accordance with one possible understanding of it, that whereby it means ‘It is not the case that the nature of being is to change and it is not the case that the nature of being is to be stable’ (because if the nature of being were to change then being would be identical with change, and if the nature of being were to be stable then being would be identical with stability). On the other hand, the sentence ‘Being neither changes nor is stable’ is entailed by the sentence ‘By virtue of its own nature, being neither changes nor is stable’ taken in accordance with another possible understanding of it, that whereby it means ‘The form being itself is of such a nature that

44 Cf. below, subsection to n. 20 of Ch. IV.

45 Cf. Stough (1990), 356.

67

it is neither a changing thing nor a stable thing’. The invalidity of the overall inference is due to the fact that the first and the second way of understanding the sentence ‘By virtue of its own nature, being neither changes nor is stable’ are not only different, but even such that the sentence understood in the first way does not entail the sentence understood in the second. So, the argument is vitiated by equivocation. Failure to recognize the difference between statements of identity and statements of predication has caused trouble to the argument.46

46 Cf. Moravcsik (1962), 28; Owen (1971), 261; Malcolm (1985a), 83−4; Bordt (1991), 521.

CHAPTER IV FORMS, DIALECTIC, AND NOT BEING

A puzzle about the possibility of applying many names to a single thing (251A5−251C7). The ES holds that the same thing may be called by many names. But this view is not shared by certain young people and certain elders who have been late to learn (the ‘late-learners’): they claim that if the same thing were called by many names, then the many would be one and the one many, which is absurd. The examples offered presuppose that names should comprise at least common nouns like ‘man’ and adjectives like ‘good’.

Plato does not provide a precise description of the puzzle of the late-learners, nor does he expound the view of language and the conception of statement on which it relies. Several interpretations have been put forward.

According to one exegesis, the late-learners hold that every name is the proper name of a certain entity: ‘man’ is the proper name of the form man, ‘good’ is the proper name of the form good, etc. When we call the same thing by many names, we use the verb ‘to be’ to construct sentences like ‘Man is good’. But when a statement is made up of two proper names connected by the verb ‘to be’, it asserts that the entity referred to by the first proper name is identical with the entity referred to by the second proper name and it is true just in case the entity referred to by the first proper name is in fact identical with the entity referred to by the second proper name (e.g. ‘Tully is Cicero’ asserts that Tully is identical with Cicero and it is true just in case Tully is in fact identical with Cicero). Therefore ‘Man is good’ also asserts that the entity referred to by the proper name ‘man’ is identical with the entity referred to by the proper name ‘good’, i.e. that the form man is identical with the form good and it is true just in case the entity referred to by the proper name ‘man’ is in fact identical with the entity referred to by the proper name ‘good’, i.e. just in case the form man is identical with the form good. But the form man and the form good are reciprocally distinct entities, i.e. are many, and the statement ‘Man is good’ asserts that they are reciprocally identical, i.e. that they are one, and it is true just in case they are in fact reciprocally identical, i.e. just in case they are one. Thus, the statement makes the many one and the one many. On this interpretation, the puzzle of the late-learners depends on taking statements of predication to be statements of identity.1

According to a second interpretation of the puzzle of the late-learners, at the heart of the puzzle lies an argument that turns on concrete examples. Here is one such example:

(1) Socrates is a single thing assumption (2) Socrates is good, snub-nosed, and small assumption (3) Good, snub-nosed, and small are many assumption

1 Cf. Grote (1875), II 445; Apelt (1897), 159; Bordt (1991), 522−3 (Bordt maintains that the statements allowed by the late-learners are definitions, which he regards as statements of identity of a special kind).

69

(4) Socrates is many from (2) and (3) (5) A single thing is many from (1) and (4)

On this interpretation also the puzzle of the late-learners depends on confusing statements of predication with statements of identity (although it probably relies also on further dubious moves, e.g. treating groups of names as single names).2 The advantage of this interpretation is that it brings the multiplicity explicitly ‘inside’ the single thing, rather than having it inferred from the alleged fact that one thing is identical with things it is other than.

According to a third interpretation of the puzzle of the late-learners, the late-learners endorse two theses about language and world: first, that the only speech-act is naming, i.e. mentioning things by means of names; secondly, that whenever one uses a name to carry out a speech-act of naming there is exactly one thing one is naming (e.g. whenever one uses the name ‘Socrates’ to carry out a speech-act of naming there is exactly one thing one is naming, that is, an individual one is referring to). Were one to apply the names ‘man’ and ‘good’ to the same thing, x, one would be naming x by using both the name ‘man’ and the name ‘good’ (because the only speech-act is naming, as the first thesis endorsed by the late-learners says). Then x would be the only thing one is naming by using the name ‘man’ (because whenever one uses a name to carry out a speech-act of naming there is exactly one thing one is naming, as the second thesis endorsed by the late-learners says) and x would also be the only thing one is naming by using the name ‘good’ (for the same reason). Since by using the name ‘man’ one names the form man and by using the name ‘good’ one names the form good, it would follow that x is identical with both the form man and the form good, so that the form man and the form good would be reciprocally identical. But the form man and the form good are in fact reciprocally distinct, so that the many would be one and the one many. According to this third interpretation, the mistake of the late-learners depends on two theses about how language is related to the world, not on confusing statements of predication with statements of identity.3

On its first as well as its second interpretation, the puzzle of the late-learners depends on the mistake of confusing predicative statements with statements of identity. Accordingly, if either of the first two interpretations is correct then one move Plato might reasonably make to solve the puzzle is to distinguish predicative statements from statements of identity. I shall argue that such a distinction is in fact drawn at 256A3−256B5. On its third interpretation, the puzzle of the late-learners depends on the two theses about language and world: the thesis that naming is the only speech-act and the thesis that whenever one uses a name to carry out a speech-act of naming there is exactly one thing one is naming. Accordingly, if the third interpretation is correct then a reasonable course of action for Plato would be to attack either of these theses. Specifically, one possible reaction would be to reject the first thesis, i.e. to claim that there is at least one speech-act other than naming. I shall argue that a claim of this sort is indeed put forward by Plato at

2 Cf. van Eck (1995), 36.

3 Cf. Moravcsik (1962), 57−9; Waletzki (1979), 251; Bostock (1984), 99−100; Stough (1990), 359−60; Moravcsik (1992), 205−6; Szaif (1998), 415−6; Malcolm (2006b), 278.

70

262D4−6, where stating is distinguished from naming. Another possible reaction would be to reject the second thesis, i.e. to insist that when one uses a name to carry out a speech-act of naming one can name more than one thing. Plato might be suggesting something of this sort at 252B9−10, where he says that one can ‘address’ a thing by that means of an ‘affection’ it enjoys.

It is hard to choose among these exegeses. I have a slight preference for the second one, due both to the fact that it is the only one brings multiplicity ‘inside’ the thing that is supposed to be one and to its ability to be transformed into an argument to the effect that physical elements also cannot combine—this second virtue will soon become obvious.

Some kinds combine, others do not. At 251C8−252E8 Plato offers an argument to the effect that some kinds combine whereas others do not. Consider the following formula, a disjunction with three members:

∀x (Kx → ¬Bx) ∨ ∀x (Kx → Bx) ∨ (∃x (Kx & Bx) & ∃x (Kx & ¬Bx))4

It is easy to see that the above formula is a logical law. If its first disjunct, ∀x (Kx → ¬Bx), is false, then ∃x (Kx & Bx) is true. Similarly, if the second disjunct, ∀x (Kx → Bx), is false, then ∃x (Kx & ¬Bx) is true. So, if both the first and the second disjunct are false, then the conjunction that constitutes the formula’s third disjunct, ∃x (Kx & Bx) & ∃x (Kx & ¬Bx), is true. Hence the falsehood of the first and of the second disjunct entails the truth of the third, so that the whole formula must be true. Now read ‘Kx ’ as short for ‘x is a pair of kinds’ and ‘Bx ’ as short for ‘x blends’. The outcome is that the sentence ‘No pair of kinds blends ∨ every pair of kinds blends ∨ (some pair of kinds blends & some pair of kinds does not blend)’ is a logical law. Plato seems to have some awareness of this fact (252D12−252E8). He refutes the first two disjuncts and regards this as sufficient to establish the third disjunct, thereby offering an indirect proof of it.5

The position expressed by the first disjunct, ‘No pair of kinds blends’, is that of the late-learners: this becomes clear when this position is described as being endorsed by ‘those who in no way allow one thing to be called by the communion of an affection of another thing’ (252B9−10). The refutation of this position (251E8−252D1) involves three parallel arguments.

(α) (251E10−252A11) If no pair of kinds blends, being blends neither with change nor with stability, and then neither change nor stability is. Then all the ontological views discussed earlier would be false.6 The argument is ad auctoritatem.

(β) (252B1−7) If no pair of kinds blends, also the physicists who speak of the combination and separation of elements must be wrong. This seems to presuppose that the puzzle of the late-learners can be deployed (perhaps with some minor modifications) to show that physical elements

4 The usual conventions for economizing on parentheses are adopted.

5 Cf. Szaif (1998), 417.

6 As Heinaman (1982−83), 176−84 convincingly argues, the point is not that if no pair of kinds blends then all the formulations of earlier ontologists are meaningless, but that if no pair of kinds blends then all the theories of earlier ontologists are false.

71

also cannot combine. Plato’s second argument is not merely ad auctoritatem: it also shows what price must be paid by those who deny that kinds blend—they must also abandon the intuitive idea that physical elements can combine.

It is not clear how the puzzle of the late-learners could be transformed into an argument to the effect that physical elements cannot combine. The second exegesis offered above of the original puzzle of the late-learners7 seems to allow such a transformation:

(1) Socrates is a single thing assumption (2) Socrates is earth, water, air, and fire assumption (3) Earth, water, air, and fire are many assumption (4) Socrates is many from (2) and (3) (5) A single thing is many from (1) and (4)

Here, in step (2), the ‘is’ involved is that of constitution (as in ‘A wagon is wheels, axle, body, rails, and yoke’, cf. Tht. 207A5−7). In the corresponding step (2) of the original version of the puzzle of the late-learners, the ‘is’ involved was instead that of predication. This ability to transform the puzzle of the late-learners into a difficulty for the combination of physical elements is one of the traits of the second interpretation of the puzzle that makes it preferable to the other two. In the Parmenides (129C4−129D2) Plato presents a puzzle about one and many that is of the same kind as the above puzzle about Socrates, earth, water, air, and fire. Moreover, in the Philebus (14C8−14E4) he draws a parallel between puzzles like the one of the late-learners and puzzles like the one about Socrates, earth, water, air, and fire.

(γ) (252B8−252D1) If no pair of kinds blends, the late-learners cannot even coherently state their own position because to state it they must carry out some blending: their actions, i.e. their speech-acts, contradict their theses (they are involved in a pragmatic contradiction, more like someone saying ‘I cannot speak’ than like someone saying ‘Socrates is a man and Socrates is not a man’).8 In this case too, instead of showing where the argument of the late-learners has gone wrong, Plato points out unacceptable consequences that would follow if it were sound.

What items are the late-learners blending when they state their own position? One possible answer is that they are simply blending linguistic expressions. In other words, it might be that in using a phrase like ‘to be separately from other things on its own’ the late-learners are combining the linguistic expressions ‘to be’, ‘separately’, ‘from other things’, and ‘on its own’.9 If observing this fact is to suffice to refute the late-learners, then their original argument, which concerns kinds, must be able to be transformed into an argument that excludes the combination of linguistic expressions. Such a transformation may be easily carried out if the second interpretation of their original puzzle is the correct one:

7 Cf. above, text to n. 2.

8 Cf. Ackrill (1955), 75.

9 Cf. Szaif (1998), 418.

72

(1) ‘To be separately from other things on its own’ is a single thing assumption (2) ‘To be separately from other things on its own’ is ‘to be’,

‘separately’, ‘from other things’, and ‘on its own’ assumption

(3) ‘To be’, ‘separately’, ‘from other things’, and ‘on its own’ are many assumption (4) ‘To be separately from other things on its own’ is many from (2) and (3) (5) A single thing is many from (1) and (4)

Another possible answer to the question asked at the beginning of this paragraph is that by stating their own position the late-learners are blending kinds. If this is correct, then Plato’s charge against the late-learners is that they are doing precisely what they are forbidding people to do. They are forbidding one to make statements of predication because doing so commits one to a blatant falsehood, but in formulating the thesis at the basis of their prohibition they make a statement of predication.10 The second answer is slightly preferable because it makes Plato’s argument into a more direct attack on the position of the late-learners.

The second disjunct, ‘Every pair of kinds blends’, is refuted more quickly (252D2−11). Were it the case that every pair of kinds blends, change would also blend with stability, so that ‘change itself would be completely stable and stability itself would in turn change’ (252D6−7), something ‘most necessarily impossible’ (252D9−10). Plato surely has good reasons for denying that the form stability changes. But is he in a position to claim that it is impossible for change to be stable? One might doubt it: after all, he seems committed to claim both that all kinds are stable and that change is a kind.11 However, as we shall see,12 there is a way of interpreting Plato whereby he has a coherent position that requires not only that stability should not change but also that change should not be stable.

Dialectic as the science of the communion of kinds. Having shown that some kinds combine and some do not, Plato goes on to argue that there is a science that studies which kinds combine with which. At 252E9−253B8 Plato introduces two analogies whose aim is to establish this result.

The first analogy is between the combinability of kinds and that of letters (252E9−253A12). Some letters combine and some do not, and vowels have a special status among letters in that they run through all others and enable them to combine with one another. To know which letters combine an art is needed, grammar. Plato does not explain what the combinability of letters amounts to. One plausible guess is that it amounts to pronounceability: the letters ‘p’ and ‘b’ cannot combine in that the sequence ‘pb’ is unpronounceable whereas the letters ‘p’ and ‘e’ can combine in that the sequence ‘pe’ is pronounceable.13 Another perhaps more likely hypothesis is

10 Cf. Bordt (1991), 523.

11 Cf. below, text to n. 20.

12 Cf. below, subsection to n. 45.

13 Cf. Galligan (1983), 269.

73

that the combinability of letters amounts to occurrence in immediate succession within words of a given language (ancient Greek in the case at hand).14 Of course, one does not need to be a grammarian to be able to combine letters in pronounceable sequences or to recognize that certain letters occur in immediate succession within words of a given language: children who have learnt to speak have such abilities but they surely do not rank as grammarians. However, in order to have knowledge of which letters combine with which in pronounceable sequences or within words of a certain language one does indeed need an art, the art of grammar. Similarly, since some kinds combine and others do not, a science is needed in order to know which kinds combine with which.

The second analogy is between the combinability of kinds and that of musical notes (253B1−5). Just as some kinds combine and some do not, so also some notes combine and some do not. As in the case of letters, Plato does not explain what the combinability of musical notes amounts to. The most plausible guess is that it amounts to something like the possibility of being produced in succession within a certain tonality. For instance, in traditional modern musical theory, if I am in the tonality of C minor then an occurrence of a D note may be followed by one of an E flat note, not by one of an E note (of course, the musical theory behind Plato’s analogy is different from traditional modern musical theory, but the point he has in mind is probably similar). Just as to know which notes combine an art is needed, music, so also to know which kinds combine an art needed.

Plato develops (253B9−253C5) the first analogy between letters and kinds: just as the combination of certain letters, i.e. the vowels, allows the combination of other letters, i.e. the consonants, so also certain kinds cause the combination or the division of others. Commentators have introduced the expression ‘vowel-kind’ for the kinds responsible for the combination or division of others. Plato does not specify which kinds are the vowel-kinds. The considerations developed later in the dialogue (between 254D9 and 259E3) make it plausible to assume that there are at least two vowel-kinds, being and otherness (sameness is perhaps a third). In particular, it is reasonable to assume that being (and perhaps also sameness) is a vowel-kind that is responsible for the combination of other kinds whereas otherness is a vowel-kind that is responsible for the division of other kinds.15

Plato seems committed to distinguish two levels of combination: the level of immediate combination and that of mediated combination, which takes place thanks to the immediate combination of certain further factors (the vowel-kinds). The combination of two kinds due to a vowel-kind which is a cause of combination probably corresponds to the speech-act of affirmation whereas the division of two kinds due to a vowel-kind that is a cause of division probably

14 Cf. Trevaskis (1966), 115.

15 Cf. Cornford (1935), 261−2; Ross (1951), 113; Owen (1971), 236; Gómez-Lobo (1977), 38, 45; Gómez-Lobo (1981), 82; Szaif (1998), 346−7. Trevaskis (1966), 111−13 argues that the vowel-kinds are verbs. This is implausible for two reasons: first, verbs are words whereas vowel-kinds (like all other kinds) are non-linguistic entities (cf. 261D1−3); secondly, there are many verbs but one gets the impression that the vowel-kinds are few (like the vowels with respect to the consonants). Lentz (1997), 103 tries to show that change and stability also are vowel-kinds, but his argument is far-fetched.

74

corresponds to the speech-act of negation (Aristotle often puts forward the idea that to negate is to divide, cf. Int. 1, 16a9−18; Metaph. Ε4, 1027b29−33).16 This is an important parallel. For suppose that affirmation were the simple combination of two kinds (e.g. of the kind man and the kind understanding, in correspondence with the affirmation ‘Man understands’). Since negation is the opposite of affirmation, we would expect negation to be the complete absence of combination of kinds: the negation of ‘Man understands’ would then amount to the separate utterance of ‘man’ and ‘understands’, which however is something very different from the real negation ‘Man does not understand’. Plato’s model suggests that affirmation and negation are two kinds of composition: affirmation is a combination that unites, while negation is a combination that divides (the uniting combination takes place thanks to the combination with a vowel-kind that is a cause of union whereas the dividing combination takes place thanks to the combination with a vowel-kind that is a cause of division).17 Negation is not the dissolution of the unity of affirmation, but the introduction of a different sort of unity.

Plato establishes a connection (253C6−253E3) between the communion of forms and the processes of division and collection, knowledge of which he identifies with dialectic. The passage describing the workings of division and collection (253D1−253E3) is obscure and has been variously interpreted.18 Plato does not elaborate on how the communion of forms is connected with the processes of division and collection. The most likely guess is that he thinks that knowing the position of a form within a classificatory system allows one to discover in what ways it may combine with other forms: for instance, it provides one with the information that a certain form partakes of all the forms to which it is subordinate and does not partake of any of the forms with which it is coordinate.

The five kinds of the Sophist. Having agreed that some branch of knowledge must study which kinds combine with one another and which do not, the ES and Theaetetus go on to examine the capability of combination of kinds (254B8−255E7). Since they cannot do this with all kinds (because they are too many), they decide to do it with three that are particularly important: being, change, and stability. Being combines with both change and stability (254D10−11), but change and stability do not combine with one another (254D7−9). The observation that each of these three kinds is other than the remaining two and the same as itself (254D12−254E1, cf. 250A8−250C5) prompts the question whether otherness and sameness are two further kinds, or instead are the same as any of the first three (254E2−255A3). In asking this question Plato assumes that sameness and otherness are reciprocally distinct, an assumption he probably makes on the basis of the view that sameness and otherness are contraries (cf. 259D2−5).

Answering the question whether otherness and sameness are two further kinds over and above being, change, and stability leads to a lengthy examination (255A4−255E7), which constitutes in

16 Cf. Gómez-Lobo (1977), 38.

17 By contrast, Cornford (1935), 262 thinks that the division of two kinds caused by a vowel-kind is not a special type of combination, i.e. a dividing combination, but an absence of combination.

18 Cf. n. 6 of Ch. I.

75

fact a digression from the project of studying the capability of combination of kinds. It consists of three arguments that purport to establish that each of the two newly introduced kinds is distinct from each of the first three. The first argument (255A4−255B7) purports to show that both sameness and otherness are distinct from both change and stability, the second (255B8−255C8) that sameness is distinct from being, and the third (255C9−255E7) that otherness is distinct from being. These arguments display a similar structure: they all assume one or more identities which they intend to disprove and they purport to show that if these identities were true then certain consequences would follow which are obviously false. All of these arguments raise difficulties, albeit in different ways: the first two share the problem that on their most natural reading they are rather obviously unsound, the third one is hard to reconstruct because its formulation is very compressed. At 253D2−3 the ES described one of the jobs of the ‘dialectical science’ as to avoid regarding a form ‘which is other as the same’. The arguments for the distinctness of each of the two newly introduced kinds from each of the first three may be regarded as such an exercise of the dialectical science.19

The argument that both sameness and otherness are distinct from both change and stability (255A4−255B7). In order to distinguish both sameness and otherness from both change and stability, Plato offers the following argument:

ES ¹All' ouà ti mh\n ki¿nhsi¿j ge kaiì sta/sij ouÃq' eÀteron ouÃte tau)to/n e)sti.

THT. Pw½j;

255A4 A5

ES àOtiper aÄn koinv= prosei¿pwmen ki¿nhsin kaiì sta/sin, tou=to ou)de/teron au)toiÍn oiâo/n te eiånai.

THT. Ti¿ dh/; ES Ki¿nhsi¿j te sth/setai kaiì sta/sij auÅ kinhqh/setai:

periì ga\r a)mfo/tera qa/teron o(poteronou=n gigno/menon au)toiÍn a)nagka/sei metaba/llein auÅ qa/teron e)piì tou)nanti¿on th=j au(tou= fu/sewj, aÀte metasxo\n tou= e)nanti¿ou.

A10

255B1 THT. Komidv= ge. ES Mete/xeton mh\n aÃmfw tau)tou= kaiì qate/rou. THT. Nai¿.

ES Mh\ toi¿nun le/gwmen ki¿nhsi¿n g' eiånai tau)to\n hÄ qa/teron, mhd' auÅ sta/sin.

THT. Mh\ ga/r.

B5

B7 ES But surely both change and stability are neither otherness nor the same. THT. How so?

255A

ES Whatever we apply in common to change and stability cannot be either of them. THT. Why?

ES Change will be stable and stability, in turn, will change: for any of the two, by coming to be about both, will oblige one of those two to transform into the contrary of its nature, by partaking of its contrary.

THT. Sure.

255B

ES But they both partake of sameness and otherness. THT. Yes.

19 Cf. Gómez-Lobo (1977), 41.

76

ES Let us then not say that change is sameness or otherness, nor stability. THT. Let us not.

If change or stability were sameness or otherness, then, since both change and stability partake of both sameness and otherness, stability would partake of change or change would partake of stability, so that stability would change or change would be stable—both of which are impossible.

Commentators have noticed that this argument is problematic.20 Let us review it in slow motion. Assume that change is identical with either sameness or otherness. Since stability partakes of both sameness and otherness, it follows that stability partakes of change, a result which is false, at least according to Plato. The assumption from which this result follows is therefore false. It may therefore be inferred that change is other than both sameness and otherness. So far the argument is faring well. Now assume that stability is identical with either sameness or otherness. Since change partakes of both sameness and otherness, it follows that change partakes of stability. The argument now assumes that this result is false, namely that change does not partake of stability. But Plato seems to be committed to deny that the result in question is false, or, equivalently, to affirm that change does partake of stability: for he seems to hold that all kinds are stable21 and he asserts that change is a kind.22 Plato therefore seems not to be entitled to infer that the assumption from which this result follows is false, so that he apparently cannot conclude that stability is other than both sameness and otherness. Note that the problem that has just been outlined cannot be solved by assuming that Plato in the Sophist modifies his conception of kinds by adopting the view that all kinds change (because they suffer an affection in that they are known, cf. 248C11−248E6):23 this assumption would commit Plato to the claim that the kind stability changes, thereby creating a parallel problem for Plato’s argument that change is other than both sameness and otherness.24 Nor, for similar reasons, can the problem be solved by assuming that for Plato in the Sophist every kind both changes and is stable.

It cannot be excluded that the problem with the argument at 255A4−255B7 simply escaped Plato’s attention: Plato could have failed to realize that this argument relies on assumptions which he is committed to reject.25 Such an account of the situation might be encouraged by the fact that

20 Cf. Crombie (1963), 405−6; Frede (1967), 34; Vlastos (1970), 276−8; Ketchum (1978), 43, 58; Heinaman (1981), 55−60; Reeve (1985), 49; Stough (1990), 376−7.

21 Cf. 248A11−12; 248D4−248E6; 249C11−249D1; 249B8−249C9; 252A7−8; Smp. 210E4−211A3; 211B1−5; Cra. 439C7−440A5; 440B4−440C1; Phd. 78D1−9; R. V 479A1−3; 479E7−8; VI 484B4−5; Prm. 135B5−135C2; Plt. 269D5−6; Ti. 27D5−28A4; 38A3−6; 51E6−52A4; Phlb. 58A2−3; 59A11−59B6; 59C2−6; 61E1−3.

22 Cf. 254D4−5; Prm. 129D6−129E1.

23 Cf. above, n. 35 of Ch. III and text thereto.

24 Cf. Heinaman (1981), 55.

25 Cf. Guthrie (1978), 150.

77

in other dialogues Plato insists that no kind can be characterized by its own contrary.26 Nevertheless, a more charitable interpretation should be attempted, one that exonerates the argument from the problem I have outlined.

First solution: the ‘ordinary’ and the ‘essentialist’ readings of sentences containing abstract nouns. An interpretation that solves the problem with Plato’s argument at 255A4−255B7, favored by several commentators,27 is based on distinguishing two readings of sentences containing abstract nouns. Sentences containing abstract nouns, e.g. instances of the schema ‘ϕness ψs’, may be read in two ways. On the one hand, they have the ‘ordinary’ reading, whereby an instance of ‘ϕness ψs’ means the same as the corresponding instance of ‘ϕness itself is a thing that ψs’. On the other hand, sentences of the sort in question have the ‘essentialist’ reading, whereby an instance of ‘ϕness ψs’ means the same as the corresponding instance of ‘At least part of what being a ϕ amounts to is to ψ’ (according to the ‘essentialist’ reading, a sentence containing an abstract noun is concerned with the nature, or content, of the concept represented by the abstract noun).

If we apply this distinction to the sentence ‘Change is stable’, we get two readings of it. On the one hand, there is the ‘ordinary’ reading, whereby ‘Change is stable’ means the same as ‘Change itself is a thing that is stable’. Plato must endorse ‘Change is stable’ if it is understood in accordance with this first reading (because, as we have seen, he maintains both that all kinds are stable and that change is a kind). On the other hand, there is the ‘essentialist’ reading, whereby ‘Change is stable’ means the same as ‘At least part of what being a changing thing amounts to is to be stable’. Plato may (indeed must) reject ‘Change is stable’ if it is understood in accordance with this second reading.

There is much to be said in favor of this rescue of Plato’s argument based on the distinction between the ‘ordinary’ and the ‘essentialist’ readings of ‘Change is stable’. The strongest consideration in support of it is the textual evidence both from within the Sophist and from early Academic works. As for the Sophist itself, at one point Plato seems to allude to the distinction between the ‘essentialist’ and the ‘ordinary’ readings of a sentence containing an abstract noun. At 250C6−8 the ES and Theaetetus agree that ‘by virtue of its own nature, being neither is stable nor changes’, a remark which seems to be focusing on the ‘essentialist’ reading of the sentences ‘Being is stable’ and ‘Being changes’ (both sentences are false because neither stability nor change is to be mentioned in a description of the nature of the concept of being). But immediately afterwards (at 250C9−250D4) the ES and Theaetetus treat their earlier admission as if it amounted to the claim that being neither is stable nor changes, which they regard as an impossibility (because everything must either be stable or change). In this development, they seem to have in mind the ‘ordinary’ reading of the sentences ‘Being is stable’ and ‘Being changes’. Since the argument of the ES and

26 Cf. Prt. 330B6−330E2; Hp.Ma. 291D1−5; Smp. 210E2−211A5; Phd. 94A1−7; 102D6−103A2; 103B4−5; 103C7−8; 104B7−8; 105A2−3; Prm. 129B1−3; 129D2−129E4; 131D8−9; 150A7−150B5.

27 Cf. Frede (1967), 30−5; Ketchum (1978), 43−4; Stough (1990), 377. The distinction between the ‘ordinary’ and the ‘essentialist’ readings of sentences containing abstract nouns has an ancestor in the distinction of predicates that can be said of forms drawn by Owen (1968a), 108 (which was not concerned with the Sophist).

78

Theaetetus is probably regarded by Plato as invalid, Plato might be hinting at the difference between the ‘essentialist’ and the ‘ordinary’ readings of the sentences ‘Being is stable’ and ‘Being changes’. As for early Academic works, in the Topics (6.10, 148a14−22, cf. 5.7, 137b3−13; 7.4, 154a16−20) Aristotle discusses the line of attack on definitions based on checking whether one of the characteristics mentioned in the definiens fails to hold of the form corresponding to the definiendum (as for instance mortality, mentioned in the definiens ‘mortal rational animal’, fails to hold of the form man, which corresponds to the definiendum ‘man’). This line of attack on definitions seems to presuppose something like the distinction between the ‘essentialist’ and the ‘ordinary’ readings of sentences containing abstract nouns.28

I am nevertheless unable to accept this rescue of Plato’s argument based on the distinction between the ‘ordinary’ and the ‘essentialist’ readings of ‘Change is stable’. The main reason for my resistance is that the rescue seems to fail to give Plato good arguments for the distinction of change and stability from sameness and otherness. For, if ‘Change is stable’ is understood in accordance with its ‘essentialist’ reading, on which it straightforwardly comes out as false, ‘Change is other’ should also be understood in accordance with its ‘essentialist’ reading and it should moreover come out as true on this reading (otherwise the argument for the distinctness of stability from otherness will be unsound). But it is hard to see on what grounds Plato could hold that ‘At least part of what being a changing thing amounts to is to be other’ is true, or (in an alternative formulation) that the nature of the concept of change can be correctly described as involving otherness. What is worse, Plato seems to deny this claim later in the dialogue, when he says that ‘each one [sc. of the forms] is other than the others not by virtue of its own nature, but by virtue of partaking of the idea of otherness’ (255E4−6).29

Second solution: the ‘ordinary’ and the ‘generalizing’ readings of sentences containing abstract nouns. A second solution for the problem with Plato’s argument at 255A4−255B7 appeals to an ambiguity of the Greek language (to which there corresponds a parallel ambiguity in English). Sentences containing abstract nouns, e.g. instances of the schema ‘ϕness ψs’, may be read in two ways. On the one hand, they have the ‘ordinary’ reading, whereby an instance of ‘ϕness ψs’ means the same as the corresponding instance of ‘ϕness itself is a thing that ψs’. On the other hand, sentences of the sort in question have the ‘generalizing’ reading, whereby an instance of ‘ϕness ψs’ means the same as the corresponding instance of ‘Everything that partakes of ϕness ψs’.30

28 Cf. Owen (1968a), 116−25; Keyt (1969), 11−13.

29 Cf. Ketchum (1978), 48; Heinaman (1981), 61−2; Reeve (1985), 50−1.

30 Cf. Runciman (1962), 96−8; Bluck (1963), 113−4; Crombie (1963), 402−3; Mates (1979), 222−4; Prior (1980), 201−2; Bostock (1984), 104. The distinction between the ‘ordinary’ and the ‘generalizing’ readings of sentences involving abstract nouns is close to the one, drawn by several commentators (cf. Vlastos (1970), 270−4; Gómez-Lobo (1977), 34, 43; Kostman (1989), 343; Bordt (1991), 509, 523, 524−5), between the ‘ordinary predication’ reading and the ‘Pauline predication’ reading of sentences of the sort in question. The only difference is that the ‘Pauline predication’ reading involves a modal operator: on the ‘Pauline predication’ reading, an instance of ‘ϕness ψs’ means the same as the corresponding instance of ‘Necessarily, everything that partakes of ϕness ψs’.

79

If we apply this distinction to the sentences ‘Change is stable’ and ‘Change partakes of stability’, we get two readings of them. On the one hand, there is the ‘ordinary’ reading, whereby ‘Change is stable’ and ‘Change partakes of stability’ mean the same as ‘Change itself is a thing that is stable’ and ‘Change itself is a thing that partakes of stability’. Plato must endorse ‘Change is stable’ and ‘Change partakes of stability’ if they are understood in accordance with this first reading (because, as we have seen, he maintains both that all kinds are stable and that change is a kind). On the other hand, there is the ‘generalizing’ reading, whereby ‘Change is stable’ and ‘Change partakes of stability’ mean the same as ‘Everything that partakes of change is stable’ and ‘Everything that partakes of change partakes of stability’. Plato may (indeed must) reject ‘Change is stable’ and ‘Change partakes of stability’ if they are understood in accordance with this second reading (at least if both being stable and partaking of stability imply being completely still in all respects).

If the ‘generalizing’ reading is at play in Plato’s argument at 255A4−255B7, then the argument comes out as valid.31 Let us consider separately the four subordinate arguments into which the argument at 255A4−255B7 may be subdivided.

(1) The first subordinate argument purports to establish that stability is not sameness. Change partakes of sameness, i.e. everything that partakes of change partakes of sameness (understanding ‘Change partakes of sameness’ in accordance with its ‘generalizing’ reading). The reason why everything that partakes of change partakes of sameness is that everything partakes of sameness with respect to itself (cf. 256A7−8). Suppose that stability were identical with sameness. Then, since everything that partakes of change partakes of sameness, it would follow that everything that partakes of change partakes of stability, i.e. that change partakes of stability (understanding ‘Change partakes of stability’ in accordance with its ‘generalizing’ reading). But the conclusion that everything that partakes of change partakes of stability is obviously false. We may therefore infer the negation of the premiss from which this obvious falsehood is derived: in other words, we may infer that stability is other than sameness.

(2) The second subordinate argument purports to establish that stability is not otherness. Change partakes of otherness, i.e. everything that partakes of change partakes of otherness (understanding ‘Change partakes of otherness’ in accordance with its ‘generalizing’ reading). The reason why everything that partakes of change partakes of otherness is that everything partakes of otherness with respect to something (cf. 255E3−6 and 259A4−6). Suppose that stability were identical with otherness. Then, since everything that partakes of change partakes of otherness, it would follow that everything that partakes of change partakes of stability, i.e. that change partakes of stability (understanding ‘Change partakes of stability’ in accordance with its ‘generalizing’ reading). But the conclusion that everything that partakes of change partakes of stability is false. We may therefore infer the negation of the premiss from which this falsehood is derived: in other words, we may infer that stability is other than otherness.

(3) The third subordinate argument purports to establish that change is not sameness. Stability partakes of sameness, i.e. everything that partakes of stability partakes of sameness (understanding

31 Cf. Bostock (1984), 108−9; Kostman (1989), 346.

80

‘Stability partakes of sameness’ in accordance with its ‘generalizing’ reading). The reason why everything that partakes of stability partakes of sameness is that everything is the same as itself, so that everything partakes of sameness with respect to itself. Suppose that change were identical with sameness. Then, since everything that partakes of stability partakes of sameness, it would follow that everything that partakes of stability partakes of change, i.e. that stability partakes of change (understanding ‘Stability partakes of change’ in accordance with its ‘generalizing’ reading). But the conclusion that everything that partakes of stability partakes of change is false. We may therefore infer the negation of the premiss from which this falsehood is derived: in other words, we may infer that change is other than sameness.

(4) The fourth subordinate argument purports to establish that change is not otherness. Stability partakes of otherness, i.e. everything that partakes of stability partakes of otherness (understanding ‘Stability partakes of otherness’ in accordance with its ‘generalizing’ reading). The reason why everything that partakes of stability partakes of otherness is that everything partakes of otherness with respect to something. Suppose that change were identical with otherness. Then, since everything that partakes of stability partakes of otherness, it would follow that everything that partakes of stability partakes of change, i.e. that stability partakes of change (understanding ‘Stability partakes of change’ in accordance with its ‘generalizing’ reading). But the conclusion that everything that partakes of stability partakes of change is false. We may therefore infer the negation of the premiss from which this falsehood is derived: in other words, we may infer that change is other than otherness.

The main objection faced by this interpretation of Plato’s argument at 255A4−255B7 is the implausibility of its assumption that the argument’s crucial premisses (‘Change partakes of sameness’, ‘Change partakes of otherness’, ‘Stability partakes of sameness’, and ‘Stability partakes of otherness’) are to be understood in accordance with their ‘generalizing’ reading. For we had already encountered the claims that change is the same, that change is other, that stability is the same, and that stability is other. This was when the ES claimed that ‘these [sc. being, change, and stability] therefore turn out to be three’ (254D12) and then inferred that ‘each of them therefore is other than the two but the same as itself ’ (254D14−15). Here the claim ‘Change is other than the remaining two kinds’ seems to be supposed to be understood in accordance with its ‘ordinary’ reading, i.e. as making the claim that change itself is a thing that partakes of otherness with respect to both being and stability. It surely does not look like the claim that everything that partakes of change partakes of otherness with respect to both being and stability. Again, the claim ‘Change is the same as itself ’ seems to be supposed to be understood in accordance with its ‘ordinary’ reading, i.e. as making the claim that change itself is a thing that partakes of sameness with respect to itself. It surely does not look like the claim that everything that partakes of change partakes of sameness with respect to itself. The interpretation of Plato’s argument at 255A4−255B7 in terms of the ‘generalizing’ reading can only be defended by assuming that in the context of the argument, the claims that change is the same and that change is other have a

81

different value with respect to the one they had in earlier contexts. This is an unpalatable assumption, but not an impossible one.32

Some confirmation for the exegesis of Plato’s argument in terms of the ‘generalizing’ reading of the crucial sentences comes from the earlier similar argument proposed by Plato to distinguish being from change and stability (250A8−250C5). The ES begins by asserting that ‘change and stability are most contrary to one another’ (250A8−9). He probably intends this to entail that change is other than stability. He goes on to argue that if change or stability were being, then, since change and stability both are, change would be stable or stability would change—which is impossible. There are several indications that the sentence ‘Change and stability both are’ is to be understood in accordance with its ‘generalizing’ reading. (α) The ES asks Theaetetus: ‘By positing in the soul being as a third item alongside these, in such a way that both stability and change are contained by it, by taking them together, and by looking at their communion with being—in this way you said that they both are?’ (250B8−11). Here ‘Both change and stability are’ seems to mean the same as ‘Both change and stability are contained by being’, which may be fairly paraphrased by ‘Both everything that partakes of change and everything that partakes of stability is’, which in turn means the same as ‘Both change and stability are’ on its the ‘generalizing’ reading.33 (β) The discussion of earlier ontological theories had reached the conclusion that ‘whatever things are unchanging and changing [oÀsa a)ki¿nhta kaiì kekinhme/na], being and all that is should be said to be both together [to\ oÃn te kaiì to\ pa=n sunamfo/tera le/gein]’ (249D3−4, the infinitive ‘le/gein’ is governed by ‘pa=sa a)na/gkh’ at 249C11). Shortly later (128 words down the text) this result is picked up by saying with regard to change and stability that ‘both of them together and each are to the same extent [eiånai … o(moi¿wj … a)mfo/tera au)ta\ kaiì e(ka/teron]’ (250A11−12). This seems to imply that ‘Both change and stability are’ means the same as ‘All changing things as well as all stable things are’, a result that matches what is required by the ‘generalizing’ reading of the first sentence. (γ) In the course of the argument leading to the claim expressed by ‘Both change and stability are’, the ES remarks: ‘Therefore one must acknowledge also what changes and change as beings [to\ kinou/menon … kaiì ki¿nhsin … w¨j oÃnta]’ (249B2−3). This remark perhaps hints that the sentence ‘Change is’ may be taken to mean the same as ‘What changes is’. Such an equivalence is what one expects on the ‘generalizing’ reading of ‘Change is’. (δ) Apart from particular formulations, the claim that both change and stability are is agreed by the ES and Theaetetus on the basis of their previous discussion of earlier ontological theories. What was established by this discussion is that all changing things as well as all stable things are. This generates a presumption that ‘Both change and stability are’ should mean the same as ‘Both everything that partakes of change and everything that partakes of stability is’, as required by its ‘generalizing’ reading.34

32 Cf. Bostock (1984), 109.

33 Cf. Vlastos (1970), 297−9.

34 Cf. Vlastos (1970), 295−7.

82

As in the cases analyzed earlier, the ‘generalizing’ reading of the relevant premisses yields valid arguments to the effect that both change and stability are other than being. Let me spell them out.

(5) Stability partakes of being, i.e. everything that partakes of stability partakes of being (understanding ‘Stability partakes of being’ in accordance with its ‘generalizing’ reading). Suppose that change were identical with being. Since everything that partakes of stability partakes of being, it would follow that everything that partakes of stability partakes of change, i.e. that stability partakes of change (understanding ‘Stability partakes of change’ in accordance with its ‘generalizing’ reading). But the conclusion that everything that partakes of stability partakes of change is false. We may therefore infer the negation of the premiss from which this falsehood is derived: in other words, we may infer that change is other than being.

(6) Change partakes of being, i.e. everything that partakes of change partakes of being (understanding ‘Change partakes of being’ in accordance with its ‘generalizing’ reading). Suppose that stability were identical with being. Since everything that partakes of change partakes of being, it would follow that everything that partakes of change partakes of stability, i.e. that change partakes of stability (understanding ‘Change partakes of stability’ in accordance with its ‘generalizing’ reading). But the conclusion that everything that partakes of change partakes of stability is false. We may therefore infer the negation of the premiss from which this falsehood is derived: in other words, we may infer that stability is other than being.

Since there is independent evidence that in the earlier argument at 250A8−250C5 the sentences ‘Change partakes of being’ and ‘Stability partakes of being’ are understood by Plato in accordance with their ‘generalizing’ reading, we have indirect confirmation that in the later argument at 255A4−255B7 the sentences ‘Change partakes of sameness’, ‘Change partakes of otherness’, ‘Stability partakes of sameness’, and ‘Stability partakes of otherness’ are also to be read in accordance with their ‘generalizing’ reading. Nevertheless, the shift between the ‘ordinary’ reading required of the sentences at 254D14−15 and the ‘generalizing’ reading attributed to those at 255A10 is hard to accept and sheds doubt on the interpretation based on the distinction between the ‘ordinary’ and the ‘generalizing’ readings of sentences involving abstract nouns.

Third solution: a different reconstruction of the logic of the argument. A third interpretation that solves the problem with Plato’s argument at 255A4−255B7 is based on reconstructing the argument differently. Let us consider the alternative reconstruction of the subordinate argument purporting to show that sameness is other than stability. The fundamental assumption is that sameness holds of change in precisely the same way in which it holds of stability. If sameness were the same as stability, then at least part of what it is for sameness to hold of stability would be to be the same as it. Since (by the fundamental assumption) sameness holds of change in precisely the same way in which it holds of stability, it would follow that at least part of what it is for sameness to hold of change would be to be the same as it. Then sameness would be the same as both stability and change, so that change and stability would be the same as one another. But this is false because change is other than stability. Therefore sameness is other than stability. Analogous alternative reconstructions are available for the other subordinate arguments of the argument at 255A4−255B7, which purport to show that otherness is other than stability, that sameness is other than change, and that otherness is other than change, and for the argument at

83

250A8−250C5, which purports to show that being is other than both change and stability.35 Note that the troublesome claim that change is not stable does not come into the picture: its place is taken by the uncontroversial claim that change is other than stability (cf. 255E11−12).36

This interpretation faces an objection. The alleged falsehood reached by the argument at 255A4−255B7 is expressed by Plato by means of the sentence ‘Change will be stable and stability, in turn, will change’ (255A10). By using this sentence Plato seems to make two statements of predication: he seems to say that change enjoys the characteristic of being stable and stability enjoys the characteristic of changing. By using the sentence in question Plato does not seem to make a statement of identity: he does not seem to say that change and stability are the same as one another.37 But the interpretation we are considering requires Plato to be using the sentence in question to make precisely such a statement of identity.

This objection may perhaps be overcome by appealing to an assumption made by some commentators in the context of their discussion of the Self-Predication Principle and its role in the Third Man Argument. According to these commentators, the Self-Predication Principle, according to which every kind enjoys to an eminent degree the characteristics which it causes perceptible individuals that partake of it to have, is one of the premisses of Third Man Argument but is not endorsed by Plato. This exonerates Plato from the difficulty which the Third Man Argument would otherwise create for his theory of forms. The reason why one might be inclined to regard Plato as committed to the Self-Predication Principle is that when he speaks of kinds he sometimes makes statements of identity by using sentences that are normally used to make statements of predication: he sometimes uses instances of ‘ϕness ψs’ or ‘ϕness is ψ’ to make statements which we would find it more appropriate to make by using the corresponding instances of ‘ϕness is ψness’. For instance, Plato sometimes uses sentences like ‘Largeness is large’ to make the statement which we would find it more appropriate to make by using ‘Largeness is largeness’. The use of sentences of this sort seems to commit Plato to the Self-Predication Principle, although of course it does not.38 If these claims about Plato’s linguistic usage are correct then at Sph. 255A10 Plato might well be using the sentence ‘Change will be stable and stability, in turn, will change’ to make the statement of identity that change and stability are the same as one another.39

35 Cf. Moravcsik (1962), 45−7; Moravcsik (1992), 184.

36 If the alternative reconstruction of the argument at 250A8−250C5 is endorsed, then the second option offered to the pluralists (243E4−7) will also have to be interpreted differently: in fact, the exegesis presented in the main text to n. 2 of Ch. III will have to be preferred.

37 Further objections to Moravcsik’s interpretation are raised by Berger (1965), 70−3 and Ray (1984), 62−3.

38 Cf. Cherniss (1957), 258−9.

39 Reeve (1985), 61 also thinks that by using the sentence ‘Change will be stable and stability, in turn, will change’ (255A10) Plato means that change will be identical with stability. Reeve’s textual ground for this interpretation is however unconvincing: the remark that ‘any of the two, by coming to be about both, will oblige one of those two to transform into the contrary of its nature, by partaking of its contrary’

84

This reply to the objection is not completely convincing because later in the Sophist (especially in the section 256A3−256B5) Plato apparently distinguishes statements of predication from statements of identity precisely with regard to kinds. This retort could perhaps be answered by pointing out that the arguments we are concerned with come before the explicit distinction between statements of predication and statements of identity, so that Plato might be intentionally allowing himself the misleading usage.

Another objection which may be raised against the interpretation we are considering is that at other points of the Sophist (252D2−11 and 254D7−9) Plato explicitly denies that change could be stable and that stability could change, and the context excludes the possibility that what he is denying could be that change is the same as stability. This objection is hard to overcome: one could perhaps point out that in the context of this denial Plato brings in extensional considerations to establish whether kinds do or do not partake of one another. Perhaps here Plato is really adopting the ‘generalizing reading’ of sentences containing abstract nouns.

The third interpretation of Plato’s argument at 255A4−255B7 has some attractive features, but at least one of the objections which may be raised against it makes it rather implausible. I therefore think that one should refrain from adopting this exegesis.

Fourth solution: the kind change is not stable because no kind is stable. A fourth interpretation that solves the problem which afflicts the argument at 255A4−255B7 is based on the assumption that according to Plato all kinds neither change nor are stable because kinds are not the sort of items which change or stability may be properly said to hold of. One reason for making this assumption is that in Plato’s view kinds are outside time.40 Plato could then have thought that since they are outside time, kinds neither change (because whatever changes does so in time) nor are stable (because to be stable is to retain characteristics through time).41 If this is right then Plato is committed to the view that the kind change (like all other kinds) is not stable and that the kind stability (like all other kinds) does not change.

One objection faced by this interpretation is that at one point of the Sophist (250C12−250D4) the ES and Theaetetus agree that whatever does not change is stable, a view which rules out the solution adopted by the present interpretation (according to which kinds neither change nor are stable). This objection may be overcome by assuming that the agreement of the ES and Theaetetus at 250C12−250D4 belongs to the aporetic section of the dialogue and does not represent Plato’s own position. Note that in the Parmenides (139A2−3, 139B2−3) Plato takes seriously the claim that stability is not the same as changelessness.

(255A11−255B1) does not show that ‘Change will be stable and stability, in turn, will change’ (255A10) means or even entails that change will be identical with stability.

40 Cf. Keyt (1969), 9. In the Timaeus (37C6−38C3) Plato seems to commit himself to the claim that kinds are eternal beings outside time. The description of the eternal model as ‘remaining in unity [me/nontoj e)n e(ni¿]’ (37D6) may be regarded as a slip of the tongue which does not commit Plato to the claim that kinds are stable.

41 Cf. Mohr (1985), 71−3. The interpretation we are considering presupposes that Vlastos (1970), 272 is wrong when he asserts that in the Sophist ‘stability’ ‘stands for invariance in its most general sense, applying naturally to entities like Plato’s forms, which, strictly speaking, could no more be said to “be stable” than “move” ’.

85

A second objection faced by this interpretation originates with the exchange with the friends of the forms. The friends of the forms endorse the claim that all kinds are stable (cf. 248A11−12, 248D4−248E6, 249C11−249D1, 252A7−8). Moreover in their discussion with the friends of the forms the ES and Theaetetus also seem to endorse the claim that all kinds are stable (cf. 249B8−249C9). If this claim is accepted by Plato, it clashes with the one attributed to him by the interpretation under consideration, namely that no kinds are stable. This objection may be overcome by making two assumptions. The first is that although the friends of the forms do hold that all kinds are stable, this point of their conception of kinds is not taken over by the ES and Theaetetus.42 If this assumption is correct, there is no reason for thinking that Plato accepts the claim that all kinds are stable. The second assumption is that despite appearances, the ES and Theaetetus do not endorse the claim that all kinds are stable. If this assumption is correct then again there is no reason for thinking that Plato accepts the claim that all kinds are stable. The first assumption is straightforward: since the ontological views of the friends of the forms are different from those eventually accepted by the ES and Theaetetus, it is reasonable to assume that the ES and Theaetetus do not take over the friends of the forms’ claim that all kinds are stable. The second assumption is more controversial and must be justified by inspecting the portion of text (249B8−249C9) where the ES and Theaetetus seem to endorse the claim that all kinds are stable. Theaetetus agrees both that the condition of ‘being in the same way and in the same manner and about the same [to\ kata\ tau)ta\ kaiì w¨sau/twj kaiì periì to\ au)to/]’ could not be satisfied without something being stable (249B12−249C2) and that if this condition were not satisfied then there would be no understanding (249C3−5). The reason why Theaetetus agrees that if the condition of ‘being in the same way and in the same manner and about the same’ were not satisfied then there would be no understanding is that he believes that understanding is an activity and as such involves change, but this change must be orderly and regular. The ES and Theaetetus are therefore not committing themselves to the claim that all kinds are stable: the condition of stability they are talking about belongs not to the kinds but to the activity of understanding (which of course may, and perhaps must, be about the kinds).43 Some support for this exegesis comes from the fact that the use of the phrase ‘being in the same way and in the same manner and about the same’ (‘to\ kata\ tau)ta\ kaiì w¨sau/twj kaiì periì to\ au)to/’, 249B12) echoes a passage from the Laws where Plato compares understanding with circular motion in the same place and describes them by using the phrase ‘in the same way and in the same manner and in the same place and about the same and in relation to the same things and in a single ratio and order’ (‘kata\ tau)ta\ … kaiì w¨sau/twj kaiì e)n t%½ au)t%½ kaiì periì ta\ au)ta\ kaiì pro\j ta\ au)ta\ kaiì eÀna lo/gon kaiì ta/cin mi¿an’, X 898A8−898B1).44

A third objection faced by the interpretation in question is that in the course of his reductio ad absurdum the ES says that if change or stability is sameness or otherness then ‘any of the two [sc.

42 Cf. O’Brien (1995), 108.

43 Cf. Mohr (1982), 193−4.

44 Other relevant passages are mentioned below, n. 51.

86

sameness and otherness], by coming to be about both [sc. change and stability], will oblige one of those two [sc. change and stability] to transform into the contrary of its nature, by partaking of its contrary’ (255A11−255B1). This way of presenting the argument, and in particular the emphatic remark that each of change and stability will be obliged ‘to transform into the contrary of its nature’, suggests that in the view of the ES the reason why stability does not change is that stability is stable and the reason why change is not stable is that change changes (not—as the interpretation in question assumes—that change neither changes nor is stable). This objection might be answered by adopting a different interpretation of the ES’s reductio. One might assume that when the ES says that if change or stability is sameness or otherness then ‘any of the two [sc. sameness and otherness], by coming to be about both [sc. change and stability], will oblige one of those two [sc. change and stability] to transform into the contrary of its nature, by partaking of its contrary’ (255A11−255B1), he means that (α) if change comes to hold of stability then it obliges stability itself to change and thereby to acquire a characteristic that is contrary to (i.e. incompatible with) its nature (i.e. its status of kind: all kinds are immune from change) and (β) if stability comes to hold of change then it obliges change itself to be stable and thereby to acquire a characteristic that is contrary to (i.e. incompatible with) its nature (i.e. its status of kind: all kinds are not only immune from change but also foreign to stability).

So the objections faced by the present interpretation of Plato’s argument at 255A4−255B7 may be overcome. But the cost incurred in overcoming them is rather high. I doubt that the present interpretation can be upheld.

Fifth solution: the kind change is not stable but other kinds are. A fifth interpretation that solves the problem with Plato’s argument at 255A4−255B7 relies on crediting Plato with something like the following view. For x to mutate in a given respect (e.g. color, size, or location) is for it to be the case that at one time x has a certain characteristic pertinent to that respect (e.g. a certain color, size, or location) and at a later time x no longer has that characteristic pertinent to that respect but has another one instead (e.g. another color, size, or location). For x to change is for x to mutate in at least one respect. For x to be stable is for x not to mutate in any respect. The kind change changes (because it is a paradigm of change), so that it mutates in at least one respect and it is not stable. However it is not the case that at one time the kind change has the characteristic of changing and at a later time it no longer has this characteristic but has instead that of being stable: hence in at least one respect the kind change does not mutate.45 The kind stability is stable (because it is a paradigm of stability), so that it does not mutate in any respect and it does not change. In particular, it is not the case that at one time the kind stability has the characteristic of being stable and at a later time it no longer has this characteristic but has instead that of changing: hence even in this respect the kind stability does not mutate. So, change is not stable and stability does not change.

This interpretation has several strengths. One is its attribution to Plato of the claim that it is possible for the same item to mutate in one respect while not mutating in another: this claim is

45 Cf. Ritter (1910), 61−2.

87

explicitly formulated in several dialogues.46 A further strength is that another claim this interpretation attributes to Plato, namely that for something to be stable is for it not to mutate in any respect, finds textual confirmation in Theaetetus’ remark that if change were to partake of stability then ‘it would be completely stable [panta/pasin iàstait' aÃn]’ (252D6). Yet another strength of this interpretation is that it commits Plato to the claim that everything either changes or is stable (because everything either mutates in at least one respect or does not mutate in any respect): in the Sophist (250C12−250D4) Plato explicitly endorses this claim.

One remarkable trait of the present interpretation is that it retains what many scholars regard as an important feature of Plato’s theory of forms, namely the Self-Predication Principle, according to which every kind enjoys to an eminent degree the characteristics which it causes perceptible individuals that partake of it to have (so that justice is just, largeness is large, change changes, and stability is stable).47 Some commentators would regard this trait as another strength of the present interpretation, others as a weakness.

But the interpretation at hand also faces an objection. The friends of the forms endorse the claim that all kinds are stable (cf. 248A11−12, 248D4−248E6, 249C11−249D1, 252A7−8). Moreover in their discussion with the friends of the forms the ES and Theaetetus also seem to endorse the claim that all kinds are stable (cf. 249B8−249C5). If this claim is accepted by Plato, it clashes with one of those attributed to him by the interpretation under scrutiny, namely that change is a kind and is not stable.

This objection may be overcome by making two assumptions. The first is that although the friends of the forms do endorse the claim that all kinds are stable, this claim is not part of the ontological views eventually endorsed by the ES and Theaetetus.48 If this assumption is correct then the fact that the friends of the forms endorse the claim that all kinds are stable provides no reason for thinking that for Plato all kinds are stable. The assumption is straightforward: the ontological views of the friends of the forms are different from those eventually accepted by the ES and Theaetetus, which therefore may well not include the claim that all kinds are stable.

The second assumption to be made in order to overcome the objection under consideration is that although in their discussion with the friends of the forms the ES and Theaetetus seem to endorse the claim that all kinds are stable, they do not actually endorse this claim but only the weaker claim that some things are stable. If this second assumption is also correct then every reason for thinking that for Plato all kinds are stable evaporates. This second assumption is rather controversial and calls for some justification: in particular, the passage must be examined which seems to show that the ES and Theaetetus endorse the claim that all kinds are stable.

46 Cf. R. IV 436C4−437A2; Tht. 181C1−181E8; Lg. X 893C4−7; Reeve (1985), 52, 58; Künne (2004), 316.

47 Cf. Prt. 330B6−330E2; Smp. 210E2−211B5; Phd. 102D6−7; 102E5−6; Prm. 132A6; 132A10−132B1; 150B6−7; 150B8−150C1. Several commentators contend that the Sophist is committed to the Self-Predication Principle: cf. Hicken (1957), 51; Seligman (1974), 42, 49, 59; Heinaman (1981), 55; above, text to n. 8 of Ch. III.

48 Cf. above, n. 42 and text thereto.

88

The passage in question is 249B8−249C5. The ES begins by remarking that ‘if we agree that all things are moving and changing, by this argument also we will subtract this same thing [sc. understanding, cf. 249B6] from beings’ (249B8−10). Theaetetus asks for an explanation (249B11). Then, after giving a negative answer (249C2) to the ES’s question ‘Does it seem to you that being in the same way and in the same manner and about the same [to\ kata\ tau)ta\ kaiì w¨sau/twj kaiì periì to\ au)to/] could at all49 come to be without stability?’ (249B12−249C1), Theaetetus agrees (at 249C3−5) that if the condition of ‘being in the same way and in the same manner and about the same’ were not satisfied then there would be no understanding.

This passage is normally taken to portray a situation where the ES and Theaetetus commit themselves to the claim that all kinds are stable. Such an interpretation is surely possible, but is not the only one available. Let me outline an alternative exegesis which avoids the situation portrayed being one where the ES and Theaetetus commit themselves to the claim that all kinds are stable. The reason why Theaetetus agrees that if the condition of ‘being in the same way and in the same manner and about the same’ were not satisfied then there would be no understanding is that he believes that understanding requires that every kind permanently and invariably enjoy the characteristic of which it is a paradigm: for the condition of ‘being in the same way and in the same manner and about the same’ to be satisfied is simply for it to be the case that every kind permanently and invariably enjoys the characteristic of which it is a paradigm.50 When he says that the condition of ‘being in the same way and in the same manner and about the same’ could not ‘at all come to be without stability’ (249C1), Theaetetus is claiming that if the condition of ‘being in the same way and in the same manner and about the same’ is satisfied, i.e. if every kind permanently and invariably enjoys the characteristic of which it is a paradigm, then some things are stable. His ground for making this last claim is that if every kind permanently and invariably enjoys the characteristic of which it is a paradigm then some kinds, indeed most kinds, namely those which are paradigms of characteristics which do not involve any mutation in any respect (kinds like unity, multiplicity, sameness, and perhaps beauty and goodness), must be stable, i.e. not mutate in any respect—because the characteristics of which they are paradigms do not involve any mutation in any respect and they themselves permanently and invariably enjoy these characteristics. Nevertheless, some kinds are not stable while creating no exception to the condition of ‘being in the same way and in the same manner and about the same’. In particular, the kind change permanently and invariably enjoys the characteristic of which it is a paradigm

49 I understand ‘pot' ’ at 249C1 as having intensive force in the question (cf. LSJ s.v. ‘po/te’ III 3).

50 Cf. Cra. 439C6−440C1; Phlb. 59A7−59C7. The remarks attributed to the friends of the forms at Sph. 248A11−12 and the description of their position at 252A7−8 commit them to the view that kinds ‘are always in the same way in the same manner’. I take it that ‘to be in the same way in the same manner’ is to enjoy a characteristic permanently and invariably, e.g. to be beautiful always and, moreover, not more at one time and less at another, nor in the body at one time and in the soul at another, etc. According to the interpretation we are considering, the phrase ‘about the same’ (‘periì to\ au)to/’, Sph. 249B12) means (not ‘around the same thing’ but) ‘concerning the same things’ or ‘holding of the same things’ (cf. Campbell (1867), Sph. 130; Brown (1998), 202): for this use of ‘peri¿’ with the accusative cf. 248E5; 255A11; 256E6; 261B1−2; 263B12. As far as only kinds are involved, kinds are always ‘about the same’ (this fails if perceptible individuals are involved: the kind beauty is about different perceptible individuals at different times).

89

(thereby creating no exception to the condition of ‘being in the same way and in the same manner and about the same’) because it never loses the condition of changing by acquiring that of being stable and it always changes in the same manner (it is not the case that at one time it mutates only with respect to color and at another only with respect to location, nor is it the case that at one time it changes faster or slower than at another). So, although it changes and it is not stable, the kind change permanently and invariably enjoys the characteristic of which it is a paradigm and therefore creates no exception to the condition of ‘being in the same way and in the same manner and about the same’, a condition which the ES and Theaetetus have agreed must be satisfied for there to be understanding. Note that at 249B2−3 the introduction of the kind change is presented as an expansion of the ontology of the friends of the forms: the unreformed friends of the forms did not recognize change as one of the kinds, and they held that all kinds are stable.

The pivotal element of the last paragraph’s reply to the objection faced by the fifth interpretation is that when Theaetetus gives a negative answer to the ES’s question ‘Does it seem to you that being in the same way and in the same manner and about the same could at all come to be without stability?’ (249B12−249C1), he is committing himself merely to the claim that if every kind permanently and invariably enjoys the characteristic of which it is a paradigm then some kinds are stable and therefore some things are stable. One might retort that this is an implausible reconstruction of the argument: surely—the retort goes—the claim to which Theaetetus is committing himself here is that whatever is ‘in the same way and in the same manner and about the same’ is stable. If this retort were right then the last paragraph’s reply to the objection would collapse: the ES and Theaetetus would be committed to the claim that all kinds are stable (because they surely accept that all kinds are ‘in the same way and in the same manner and about the same’). But the retort misfires: Plato would probably deny the claim which the retort takes Theaetetus to be committing himself to, namely the claim that whatever is ‘in the same way and in the same manner and about the same’ is stable. For in the Laws Plato compares understanding with circular motion in the same place and says that they both ‘move in the same way and in the same manner and in the same place and about the same and in relation to the same things and in a single ratio and order’ (‘kata\ tau)ta\ … kaiì wsau/twj kaiì e)n t%½ au)t%½ kaiì periì ta\ au)ta\ kaiì pro\j ta\ au)ta\ kaiì eÀna lo/gon kaiì ta/cin mi¿an … kineiÍsqai’, X 898A8−898B1).51

So the objection faced by the fifth interpretation of Plato’s argument at 255A4−255B7 may be overcome. This fifth interpretation provides the most plausible solution of the puzzles raised by Plato’s argument at 255A4−255B7.

The argument that being is distinct from sameness (255B8−255C8). In order to distinguish being from sameness, Plato offers the following argument:

51 Cf. Plt. 269E1−3; Ti. 36C2−3.

90

ES ¹All' aÅra to\ oÄn kaiì to\ tau)to\n wj eÀn ti dianohte/on h(miÍn;

THT. ãIswj.

255B8

B10 ES ¹All' ei¹ to\ oÄn kaiì to\ tau)to\n mhde\n dia/foron

shmai¿neton, ki¿nhsin auÅ pa/lin kaiì sta/sin a)mfo/tera eiånai le/gontej a)mfo/tera ouÀtwj au)ta\ tau)to\n w¨j oÃnta proserou=- men.

255C1

THT. ¹Alla\ mh\n tou=to/ ge a)du/naton. ES ¹Adu/naton aÃra tau)to\n kaiì to\ oÄn eÁn eiånai. THT. Sxedo/n. C5 ES Te/tarton dh\ pro\j toiÍj trisiìn eiãdesin to\ tau)to\n

tiqw½men; THT. Pa/nu me\n ouÅn.

C8 ES Should we then think of being and sameness as a single thing? THT. Perhaps.

255B

ES But if ‘being’ and ‘the same’ signify nothing different, then, again, by saying that change and stability both are we will in this way be addressing them as being both the same.

THT. But this is surely impossible.

255C

ES It is therefore impossible that the same and being be a single thing. THT. I suppose so.

ES Must we then posit sameness as fourth in addition to the three forms? THT. By all means.

Plato’s argument at 255B8−255C8 is usually taken to go more or less as follows. If being were identical with sameness, then ‘Change and stability are the same’ would follow from ‘Change and stability are’. But the second sentence is true: for it is logically equivalent52 to the true sentence ‘Change is and stability is’. The first sentence is instead false: for it is logically equivalent to the false sentence ‘Change is the same as stability’. So, being is other than sameness.

Is the argument at 255B8−255C8 invalid? This argument is objectionable because the sentence ‘Change and stability are the same’ has two readings: a collective one and a distributive one. On its collective reading, ‘Change and stability are the same’ is logically equivalent to ‘Change is the same as stability’; on its distributive reading, ‘Change and stability are the same’ is logically equivalent to ‘Change is the same as something and stability is the same as something’ (or, perhaps, to ‘Change is the same as itself and stability is the same as itself ’). Obviously, ‘Change and stability are the same’ is false on its collective reading, true on its distributive reading. The argument treats ‘Change and stability are the same’ as false because it adopts the collective reading. But, in the context of the argument, the distributive reading should be preferred to the collective one because the premiss from which ‘Change and stability are the same’ is inferred, i.e.

52 I use ‘logically equivalent’ in such a way that a sentence s is logically equivalent to a sentence s′ just in case both s entails s′ and s′ entails s.

91

‘Change and stability are’, is to be understood in accordance with its distributive reading: it is logically equivalent to ‘Change is and stability is’.53

It cannot be excluded that Plato did in fact propose the argument attributed to him by the usual interpretation but the problem with the argument eluded him. Nevertheless, with charitable spirit, I intend to explore ways of avoiding saddling Plato with an objectionable argument. In particular, I shall look for an interpretation that gives him a non-objectionable argument.

The sentences involved in the argument at 255B8−255C8 have one or more readings. One interpretation that gives Plato a non-objectionable argument is as follows. Suppose that being and sameness were ‘a single thing’ (255B8). Given that two predicative expressions have the same meaning just in case they are correlated with the same kind, it would follow that ‘to be’ has the same meaning as ‘to be the same’, so that the sentence ‘Change and stability are’ would have the same meaning as the sentence ‘Change and stability are the same’ (cf. 255B11−255C2). ‘But this is surely impossible’ (255C3): for ‘Change and stability are’ has only one reading, namely the distributive reading whereby it is logically equivalent to ‘Change is and stability is’ and is therefore true, whereas ‘Change and stability are the same’ has two readings, namely the collective reading whereby it is logically equivalent to ‘Change is the same as stability’ and is therefore false, and the distributive reading whereby it is logically equivalent to ‘Change is the same as something and stability is the same as something’ (or, perhaps, to ‘Change is the same as itself and stability is the same as itself ’) and is therefore true. ‘It is therefore impossible that sameness and being be a single thing’ (255C4).54

The argument given to Plato by this interpretation is a good one. Regrettably, the interpretation is unacceptable because it fails to fit the text. In the text we do not easily find a comparison between the readings of ‘Change and stability are’ and those of ‘Change and stability are the same’, unless an enormous weight is given to the expression ‘in this way’ (‘ouÀtwj’, 255C1), which however is most naturally taken to be merely introducing the apodosis.55 Nor does the text say that ‘Change and stability are the same’ is false on one reading: it simply says that it is false.

The sentences involved in the argument at 255B8−255C8 are elliptical. If we want to avoid saddling Plato with a poor argument, a different interpretation must be looked for. Here is one. Suppose that being and sameness were ‘a single thing’ (255B8). Given that two predicative expressions have the same meaning just in case they are correlated with the same kind, it would follow that ‘to be’ has the same meaning as ‘to be the same’. Then for any use of the sentence ‘Change and stability are’ there would be a corresponding use of the sentence ‘Change and stability are the same’ with the same meaning and the same truth value. But this is not the case. To see it,

53 Cf. Peck (1952), 48; Bluck (1963), 144; Trevaskis (1966), 103; Vlastos (1970), 286−7; Owen (1971), 266; Bostock (1984), 91.

54 Cf. Lacey (1959), 49.

55 Cf. LSJ s.v. ‘ouÀtwj’ I 7.

92

consider the sentence ‘Change and stability are’ as used to say that change and stability are thus and so, e.g. the sentence ‘Change and stability are instantiated’, which says that each of change and stability partakes of being with respect to the kind instantiated: this is true. The corresponding sentence involving ‘to be the same’ is one which says that change and stability are the same as thus and so, e.g. the sentence ‘Change and stability are the same as instantiated’, which says that each of change and stability partakes of sameness with respect to the kind instantiated: this ought to be true, but it is in fact false, indeed impossible (because it entails that change and stability should be the same as one another). Therefore being is other than sameness.

This interpretation also gives Plato a good argument. But it too faces difficulties in that it seems to fail to fit the text. For it requires that the sentence ‘Change and stability are’ (255B12) should involve the predicative elliptical use of ‘to be’:56 the sentence is taken to say that each of change and stability is thus and so in that each of change and stability partakes of being with respect to thus and so. This is implausible because on earlier occasions the sentence ‘Change and stability are’ has been understood as making the claim that change and stability exist, with a complete use of ‘to be’ rather than an incomplete (albeit elliptical) one.57

The sentences involved in the argument at 255B8−255C8 have only collective readings. Yet another interpretation of 255B8−255C8 is based on noticing the presence of the expression ‘both’ (‘a)mfo/tera’). It can hardly be a matter of chance that this expression occurs twice in the passage (at 255B12 and 255C1). It is impossible to give a sure explanation of the role of ‘both’ in the passage, but one as plausible as any is that the role of the expression is to force the collective reading of the sentences containing it (I shall later mention some evidence in support of this explanation).58 If this explanation of the role of ‘both’ in the passage is right then ‘Change and stability both are’ and ‘Change and stability are both the same’ have only collective readings. The expression ‘each’ plays exactly the opposite role with respect to ‘both’ because it forces a distributive reading: ‘Each of change and stability is’ and ‘Each of change and stability is the same’ have only distributive readings (neither of these last sentences occurs in 255B8−255C8—I mention them merely for the sake of generating a contrast that clarifies the issue). Now, ‘Change and stability both are’ (a sentence with only a collective reading) is logically equivalent to ‘Each of change and stability is’ (the corresponding sentence with only a distributive reading). Although logically equivalent, the sentences ‘Change and stability both are’ and ‘Each of change and stability is’ have subtly different meanings: their relationship is a bit like that between ‘Socrates speaks to himself ’ and ‘Socrates speaks to Socrates’, which are also logically equivalent despite having subtly different meanings. ‘Change and stability both are’ and ‘Each of change and stability is’ are both true. On the other hand, ‘Change and stability are both the same’ (a sentence with only a collective reading) is not logically equivalent to ‘Each of change and stability is the

56 On the incomplete elliptical uses of ‘to be’ cf. below, text to n. 67.

57 Cf. 249C10−249D5; 250A11−250B1; 250B8−250C2; 254D10−11.

58 Cf. below, n. 62 and text thereto.

93

same’ (the corresponding sentence with only a distributive reading): in fact, the first is false (because it is logically equivalent to the false sentence ‘Change is the same as stability’) while the second is true (because it is logically equivalent to the true sentence ‘Change is the same as something and stability is the same as something’).59 Given all of this, suppose that being and sameness were ‘a single thing’ (255B8). Given that two predicative expressions have the same meaning just in case they are correlated with the same kind, it would follow that ‘to be’ has the same meaning as ‘to be the same’, so that ‘Change and stability both are’ would have the same meaning and therefore the same truth value as ‘Change and stability are both the same’. But this is not the case: as we have seen, the first sentence is true whereas the second is false—indeed, what it says ‘is surely impossible’ (255C3). Therefore being is other than sameness.60

This interpretation is preferable to the others. One of its strengths is that it enables one to see that when earlier in the Sophist the ES said with regard to change and stability that ‘both and each of them are to the same extent [eiånai … o(moi¿wj … a)mfo/tera au)ta\ kaiì e(ka/teron]’ (250A11−12), the joint presence of ‘both’ and ‘each’ was not merely a redundancy motivated by style.61 A further strength of the interpretation at hand is that the argument it attributes to Plato has a parallel in the Greater Hippias (299B8−303D10). Hippias claims (300B6−301D4) that for every F, x and y are both F just in case each of x and y is F. He offers as examples several values of ‘F ’ for which this is true: ‘F ’ may for instance be replaced by ‘just’, ‘healthy’, and ‘made of gold’. But Socrates refutes Hippias’ claim (301D5−302B6) by offering as counterexamples the cases where ‘F ’ is replaced by ‘one’ and ‘two’: if each of x and y is one, it does not follow that x and y are both one, and if x and y are both two, it does not follow that each of x and y is two.62 In our

59 How is it that ‘Change and stability are both the same’ is logically equivalent to ‘Change is the same as stability’ whereas ‘Each of change and stability is the same’ is logically equivalent to‘Change is the same as something and stability is the same as something’? One possible explanation is that ‘the same’ is short for ‘the same as something’. Then it is trivial to see that ‘Each of change and stability is the same’, which abbreviates ‘Each of change and stability is the same as something’, is logically equivalent to‘Change is the same as something and stability is the same as something’ (the sentences have in fact the same meaning). On the other hand, ‘Change and stability are both the same’, being short for ‘Change and stability are both the same as something’, means the same as ‘Change and stability are the same as the same thing’, which in turn is logically equivalent to ‘Change is the same as stability’. The crucial point here is that ‘Change and stability are both the same as something’ means the same as ‘Change and stability are the same as the same thing’: consider that ‘Jim and Jane are both pulling a boat’ has the same meaning as ‘Jim and Jane are pulling the same boat’ (contrast ‘Each of Jim and Jane is pulling a boat’).

60 Cf. van Eck (2000), 67−8.

61 Cf. van Eck (2000), 68.

62 In the passage from the Greater Hippias Plato systematically uses ‘both’ (‘a)mfo/teroi’) to flag that an attribute holds collectively of two items: cf. 299C9, 300A10 (where ‘a)mfo/teroi’ is associated with ‘koinv=’), 300B1, 300B4, 300B8, 300D8, 300E4, 300E5, 300E9, 300E10, 301A3, 301A6, 301B8, 301C1−2, 301D8, 301E1, 301E3, 301E4, 301E5, 302A5, 302A7, 302A8, 302B2, 302B3, 302C1, 302C2, 302C3, 302C5, 302D1, 302D2, 302D4, 302E6, 302E8, 302E11, 303A2, 303A5, 303A6, 303A10, 303B2−3, , 303B4, 303B5, 303B6, 303C4, 303C5, 303D2. (At one point, at 303B8, Plato uses ‘sunamfo/teroi’ instead of ‘a)mfo/teroi’: a similar variation occurs in the Sophist, at 249D4 and 250C3.) It is also worth noting that the passage from the Greater Hippias employs the distinction between the collective and the distributive holding of attributes in order to

94

Sophist passage, ‘being’ behaves like ‘just’, ‘healthy’, and ‘made of gold’ whereas ‘the same’ behaves like ‘one’.63

When the sentence ‘Change and stability both are’ was accepted as true earlier in the dialogue, at 250B8−11, it was probably understood in accordance with its ‘generalizing’ reading, whereby it is logically equivalent to ‘Everything that partakes of change is and everything that partakes of stability is’.64 One might therefore be inclined to adopt the ‘generalizing’ reading also in interpreting the argument at 255B8−255C8. This inclination should however be resisted: the by far most natural way of understanding the sentence ‘Change and stability are both the same’ is the one given by its ‘ordinary’ reading, according to which the sentence is logically equivalent to ‘Change itself is a thing that is identical with stability itself ’. This requires that the ‘ordinary’ reading be adopted throughout the argument.

The argument that being is distinct from otherness (255C9−255E7). To distinguish being from otherness, Plato offers the following argument:

ES Ti¿ de/; to\ qa/teron aÅra h(miÍn lekte/on pe/mpton; hÄ tou=to kaiì to\ oÄn wj du/' aÃtta o)no/mata e)f' e(niì ge/nei dianoeiÍ- sqai deiÍ;

THT. Ta/x' aÃn.

255C9 C10

ES ¹All' oiåmai¿ se sugxwreiÍn tw½n oÃntwn ta\ me\n au)ta\ kaq' au(ta/, ta\ de\ pro\j aÃlla a)eiì le/gesqai.

THT. Ti¿ d' ouÃ;

C15 ES To\ de/ g' eÀteron a)eiì pro\j eÀteron: hÅ ga/r; THT. OuÀtwj.

255D1

ES Ou)k aÃn, eiã ge to\ oÄn kaiì to\ qa/teron mh\ pa/mpolu diefere/thn: a)ll' eiãper qa/teron a)mfoiÍn meteiÍxe toiÍn ei¹doiÍn wÐsper to\ oÃn, hÅn aÃn pote/ ti kaiì tw½n e(te/rwn eÀteron ou) pro\j

D5 eÀteron: nu=n de\ a)texnw½j h(miÍn oÀtiper aÄn eÀteron vÅ, sumbe/- bhken e)c a)na/gkhj e(te/rou tou=to oÀper e)stiìn eiånai.

THT. Le/geij kaqa/per eÃxei.

ES Pe/mpton dh\ th\n qate/rou fu/sin lekte/on e)n toiÍj eiãdesin ouÅsan, e)n oiâj proairou/meqa.

THT. Nai¿.

255E1

ES Kaiì dia\ pa/ntwn ge au)th\n au)tw½n fh/somen eiånai dielhluquiÍan: eÁn eÀkaston ga\r eÀteron eiånai tw½n aÃllwn ou)

dia\ th\n au(tou= fu/sin, a)lla\ dia\ to\ mete/xein th=j i¹de/aj th=j qate/rou.

THT. Komidv= me\n ouÅn.

E5

E7 ES What then? Must we speak of otherness as a fifth? Or must we regard this and being as two

names for a single kind? THT. Perhaps.

255C

show that certain attributes are reciprocally distinct: the interpretation of Sph. 255B11−255C3 we are now considering takes the argument there to be more or less along these lines.

63 Cf. 303B1−303D1; R. VII 524B10; Tht. 185B2; 203D7−8.

64 Cf. above, paragraph to n. 34.

95

ES However, I think you agree that, of beings, some are always spoken of in their own right, some relatively to other things.

THT. Of course.

ES But whatever is other is always spoken of relatively to an other: isn’t that so? THT. So it is.

255D

ES This would not be the case if being and otherness were not completely different: but if otherness had partaken of both forms, as being does, one of the things that are other would at some point have been other not relatively to an other. But now, in our view, whatever is other, it results of necessity that it is just what it is than an other.65

THT. It is just as you say.

ES Therefore the nature of the other must be said to be fifth among the forms we are selecting. THT. Yes.

255E

ES And we shall say that it runs through all of them: for each one is other than the others not by virtue of its own nature, but by virtue of partaking of the idea of otherness.

THT. By all means.

Five interpretations of this passage have been offered. According to the first exegesis, Plato is distinguishing a complete and an incomplete use of ‘to

be’. On the complete use of ‘to be’, a form of this verb is construed with no complement (not even implicit). From a semantic point of view, an occurrence of a form of ‘to be’ in its complete use is normally involved in making a statement of existence. On the incomplete use of ‘to be’, a form of the verb is construed with some complement (which however may remain implicit, in which case the incomplete use of ‘to be’ is elliptical). From a semantic point of view, a form of ‘to be’ in its incomplete use is often employed to make a statement of predication, whereby a certain thing is said to be in a certain way (it may also be employed to make a statement of identity, whereby a certain thing is said to be identical with a certain thing).66 The complete use of ‘to be’ and the incomplete one are closely connected from a semantic point of view: there is no change in the sense of the verb (the relation between ‘Jim is’, which exemplifies the complete use of ‘to be’, and ‘Jim is slow’, which exemplifies the predicative incomplete use of ‘to be’, is analogous to that between ‘Jim is running’ and ‘Jim is running slowly’—given that, as I said, the complete use of ‘to be’ is normally employed to make statements of existence, this implies an existential component of statements of predication and in general of all statements involving a form of ‘to be’ in its incomplete use). By contrast, ‘other’ has only an incomplete use: any form of the adjective must be construed with some complement (which however may remain implicit). For this reason, otherness is distinct from being.67

65 For the formulation here cf. Prm. 133C8, 133E3−4, 134A3−8.

66 The most frequently adopted variant of the first exegesis leaves statements of identity out of the picture.

67 Cf. Lacey (1959), 49; Moravcsik (1962), 48, 54; Bostock (1984), 92−4; Brown (1986/99), 462, 474−7; Moravcsik (1992), 185. The account offered in the main text above of the complete and the incomplete use of ‘to be’ differs from that adopted or presupposed by some of the commentators referred to at the beginning of the present footnote: according to these commentators, the semantic difference between the complete and the incomplete use of ‘to be’ is deeper and sharper than I have made it. My account is close to that of Lesley Brown. Ferg (1976), 337−9 identifies the incomplete use of ‘to be’ with the ‘converse’ of its predicative use, whereby it expresses a kind’s holding of a certain item (often an individual) (cf. below, text to n. 98).

96

According to the second exegesis, Plato is introducing an ontological classification whereby he distinguishes two kinds of beings: some beings are beings in their own right (e.g. fingers, men, horses), some are beings relatively to something (e.g. large things, fathers). By contrast, all things that are other are other relatively to something. To be more precise, consider first the class of beings. What is it that makes beings into beings? In the case of some beings, the answer to this question does not set the beings concerned in relation to something. For instance, what is it that makes a finger into a being? The appropriate answer is that what makes a finger into a being is its being a finger, an answer that does not set the being concerned in relation to something. In the case of other beings, the answer to the above question does set the beings concerned in relation to something. For instance, what is it that makes a large thing into a being? The appropriate answer is that what makes a large thing into a being is its being larger than things of the same sort that are of standard size, and this answer does set the being concerned in relation to something. Consider then the class of things that are other. What is it that makes a thing that is other other? In the case of all things that are other, the answer to this question sets things that are other in relation to something: if with regard to any thing that is other one asks what makes it other, the appropriate answer is that what makes this thing that is other other is its being other than something, an answer that sets the given thing that is other in relation to something. Since some beings are beings in their own right and some are beings relatively to something, and since all things that are other are other relatively to something, it follows that being is distinct from otherness.68

According to the third exegesis, Plato is distinguishing two kinds of incomplete use of ‘to be’. In the case of the first kind of incomplete use, ‘to be’ is employed to make statements of identity, and then beings are said to be ‘in their own right’ (because the only thing which any being is by being identical with it is that being itself ). In the case of the second kind of incomplete use, ‘to be’ is employed to make statements of predication, and then beings are said to be ‘relatively to others’ (because all, or perhaps most, things which any being is by partaking of them are other than that being). By contrast, whatever is other is said to be other relatively only to something other than it. For this reason otherness is distinct from being.69

According to the fourth exegesis, Plato is distinguishing two classes of beings on the basis of two ways in which the incomplete ‘to be’ can be used: on the one hand, there are forms, with regard to which the incomplete ‘to be’ can be used both to express what they are relatively to their own essence or parts thereof and to express their partaking of some attribute external to them; on the other hand, there are perceptible individuals, with regard to which the incomplete ‘to be’ can

68 Cf. Apelt (1897), 172; Cornford (1935), 282; Ross (1951), 113−14; Owen (1957), 107; Meinhardt (1968), 53−4; Vlastos (1970), 290; Seligman (1974), 60−3; Heinaman (1982−83), 186; Heinaman (1983), 14; Szaif (1998), 353−4; Dancy (1999), 59−70; Malcolm (2006b), 282−4. The second exegesis comes in many subtly different versions, which cannot be followed up here. What I have offered in the main text above is what seems to me the best possible argument that is recognizably close to those attributed to Plato by the commentators mentioned in the first part of this footnote.

69 Cf. Owen (1965), 71; Owen (1968a), 108; Owen (1971), 255−8; Wiggins (1971), 289−90; Reeve (1985), 54−5; Bordt (1991), 525−6; Fine (1993), 171−2.

97

only be used to express their partaking of some attribute external to them (this is because perceptible individuals do not have essences). Thus: in some cases, for x to be thus and so can be part of the very nature of x; in other cases, for x to be thus and so is not part of the nature of x, but something additional to x ’s nature. By contrast, for x to be other than something is never part of the nature of x, it is always something additional to x ’s nature. For this reason otherness is distinct from being.70

According to the fifth exegesis, Plato’s distinction between being and otherness is based on the logical structures of sentences in which these kinds may be involved. Being (always viewed as a vowel-kind linking further kinds) can be involved in sentences of two logical structures, which are displayed in ‘John and Mary are 25 years old’ and ‘John and Mary are married’: the first sentence says that John and Mary partake of being with respect to the kind 25-years-old independently of one another (this would be reflected in its being formalized by a conjunctive formula, e.g. ‘Oj & Om’), while the second sentence says that John and Mary partake of being with respect to the kind married relatively to one another (this would be reflected in its being formalized by an atomic formula with a two-place relation constant, e.g. ‘Vjm’). By contrast, otherness can be involved in sentences of only one logical structure, which is displayed in ‘John and Mary are different human beings’: this sentence says that John and Mary partake of otherness with respect to the kind human-being relatively to one another (this would be reflected in its being formalized by an atomic formula with a two-place relation constant, e.g. ‘Djm’). Otherness cannot be involved in sentences of such a logical structure that two items partake of it with respect to some kind but independently of one another.71

The fifth exegesis relies on the assumption that for Plato the otherness is always used relatively to a given kind: cf. ‘John and Mary are different human beings’. But Plato does not seem to conceive of otherness in these terms: he seems to regard it as linking items with no reference to a further form (e.g. ‘John is other than Mary’). This makes the fifth exegesis less plausible than the first and the second.

The third and the fourth exegesis share a presupposition: they both presuppose that when Plato says that ‘of beings, some are always spoken of in their own right [au)ta\ kaq' au(ta/], some relatively to other things [pro\j aÃlla]’ (255C13−14), he does not mean to distinguish an ‘absolute’ and a ‘relative’ use of ‘to be’, but two incomplete uses of ‘to be’: one in which the verb’s complement on the predicate side is supposed to mention the same item (or, in the case of the fourth exegesis, an essential part of the item) that is mentioned by the verb’s complement on the subject side; and one in which the verb’s complement on the predicate side is supposed to mention an item that is other than (or, in the case of the fourth exegesis, disjoint from) the one mentioned by the verb’s complement on the subject side. Both the third and the fourth exegesis therefore presuppose a strict understanding of the phrase ‘relatively to other things’. Such a strict understanding of this phrase is however rather implausible from a linguistic point of view. For we often use ‘other’ somewhat loosely, as when we use sentences like ‘one thing is the same as

70 Cf. Frede (1992), 400−2.

71 Cf. de Vries (1988), 390−2.

98

another’ without taking ourselves to be uttering a contradiction. More to the point, such a loose usage of ‘other’ occurs in ancient discussions of relatives. For instance, in the Philebus Plato contrasts things that are beautiful ‘relatively to something’ (‘pro/j ti’, 51C6) with things that are beautiful ‘in themselves’ (‘kaq' au(ta/’, 51C7) and shortly afterwards draws again the same contrast by opposing things that are beautiful ‘relatively to something else’ (‘pro\j eÀteron’, 51D7) to things that are beautiful in their own right (‘au)ta\j kaq' au(ta/j’, 51D8). Moreover, in a fragment of Plato’s pupil Hermodorus, recorded in Simplicius’ Commentary on Aristotle’s Physics, the phrase ‘relatively to other things’ is paired with the phrase ‘in their own right’ to introduce the contrast between relative and absolute beings: ‘Of beings, some are spoken of in their own right [kaq' au(ta/], like man and horse, some relatively to other things [pro\j eÀtera], and of these some as contraries, like good to bad, some instead as relatively to something [pro/j ti], and of these some as determinate, some instead as indeterminate’ (248, 2−5). A similar usage occurs in chapter 7 of Aristotle’s Categories: ‘Relative to something [pro/j ti] are called such things as those which are said to be just what they are of other things [e(te/rwn], or in whatever other way relatively to an other [pro\j eÀteron]’ (6a36−7, cf. 6, 5b17; 5b20; 5b28; 5b32; 7, 6a37; 6a38; 6b7; 8, 11a25; 11a27; Top. 6.8, 146b3; Ph. 1.7, 190a35; Metaph. ∆15, 1021a26−8).72 Since Hermodorus uses ‘in their own right’ and ‘relatively to other things’ to introduce the contrast between absolute beings and relative beings, an analogous usage may be safely assumed for our Sophist passage. An advocate of the third or of the fourth exegesis might retort that when he discusses the case of otherness, Plato says that ‘whatever is other is always spoken of relatively to an other [pro\j eÀteron]’ (255D1): here ‘other’ is to be taken strictly (i.e. as importing distinctness), hence it must be taken strictly also in the discussion of being, at 255C13−14. This retort is answered by pointing out that the first occurrence of ‘relatively to other things’ (‘pro\j aÃlla’, 255C13−14) is most naturally taken as meaning ‘relatively to something’ and that the second occurrence of ‘relatively to an other’ (‘pro\j eÀteron’, 255D1) may then be understood as providing a ground for the claim (which remains unformulated) that anything that is other is always spoken of relatively to something: anything that is other is always spoken of relatively to something because it is spoken of relatively to something that is in turn other than it. In conclusion, the third and the fourth exegesis rest on shaky ground.

On top of this, the third and the fourth exegesis face their own peculiar problems. The third exegesis seems to commit Plato to the claim that while the only thing which any being is by being identical with it is that being itself, everything which any being is by partaking of it is other than that being. But Plato should reject the second part of this claim: for he says that ‘everything partakes of the same’ (256A7−8), so that he is committed to concede that something (i.e. sameness) which some being (i.e. sameness) is by partaking of it is not other than that being. Plato could draw a correct distinction between the identity use of ‘to be’ and the predicative one by employing the conceptual apparatus displayed in the passage we are looking at: he could claim that while the only thing which any being is by being identical with it is that being itself, at least

72 Cf. Ferg (1976), 338; Heinaman (1983), 15; Brown (1986/99), 476; Szaif (1998), 354; Dancy (1999), 47−8, 59.

99

one thing which any being is by partaking of it is other than that being. But this claim can hardly be extracted from Plato’s words: ‘Of beings, some are always spoken of in their own right, some relatively to other things’ (255C13−14).

The fourth exegesis also has its own peculiar problems. For it requires that in the passage at hand ‘other’ should mean something like ‘external to the nature’. But this does not seem to be the way ‘other’ is used in the Sophist: otherness came to the fore as a kind to be discussed as a consequence of the fact that being, change, and stability are three kinds (cf. 254D12−254E3) and this seems to require that ‘other’ express nothing more than numerical non-identity.

In conclusion, the third, fourth, and fifth exegesis are rather implausible. It is difficult to decide between the survivors of this decimation, namely the first and the second exegesis. Two considerations tell against the second exegesis: first, it is somewhat elaborate; secondly, it relies on the principle that a thing’s essence is what makes it into a being, a view associated more with Aristotle (and his technical term for essence, ‘what it was to be’, ‘ti¿ hÅn eiånai’) than Plato. These considerations bear a particularly heavy weight in view of the fact that Theaetetus is supposed to be acquainted with the distinction concerning being: for the ES introduces this distinction with the words ‘I think you agree that …’ (255C13) (note that what is used is the present ‘you agree’, not the future ‘you will agree’).73

A difficulty faced by the first exegesis is that Plato, by saying that ‘of beings, some are always spoken of in their own right, some relatively to other things’ (255C13−14), seems to be introducing two disjoint classes of beings: the class whose members are the beings that ‘are always spoken of in their own right’ (255C13−14), on the one hand, and the class whose members are the beings that ‘are always spoken of relatively to other things’ (255C13−14), on the other. But the first exegesis does not introduce two disjoint classes of beings: it introduces two uses of ‘to be’ which are applicable to the same items (i.e. to everything).74 The first exegesis can overcome this objection only by denying the assumption on which it relies: by denying, in other words, that Plato is introducing two disjoint classes of beings. In order to do make this defense plausible, an advocate of the first exegesis must take it that the adverb ‘always’ in the sentence ‘Of beings, some are always spoken of in their own right, some relatively to other things’ (255C13−14) has a wide scope: the sentence used by Plato must be paraphrased by something like ‘Whenever we speak of beings as beings, we speak of them either in their own right or relatively to something’. This is surely not a natural paraphrase, but it is not impossible. Note that the second exegesis does not face this difficulty because it does take Plato to be introducing two disjoint classes of beings. On the other hand, the first exegesis is not affected by the problem that creates a major stumbling-block for the second: the complete and the incomplete uses of ‘to be’ have been on display in the immediately preceding lines (e.g. the sentence at 255B11−255C2 parades an example of the complete use of ‘to be’, i.e. the ‘eiånai’ at 255B12, alongside an example of the incomplete use of the verb, i.e. the ‘oÃnta’ at 255C1), so that the ES may perfectly reasonably introduce the distinction between these uses with the words ‘I think you agree that …’ (255C13). One

73 Cf. Owen (1971), 257.

74 Cf. Bluck (1963), 147; Kostman (1989), 352−3; Szaif (1998), 354.

100

consideration that supports the first exegesis is that at the end of the second aporetic argument about not being (238A1−238C12) Plato says that ‘one cannot correctly express nor say nor think of what is not in its own right [to\ mh\ oÄn au)to\ kaq' au(to/]’ (238C8−9).75 The use here of the formula ‘au)to\ kaq' au(to/’ in connection with a singular form of ‘not to be’ recalls the use of the formula ‘au)ta\ kaq' au(ta/’ in connection with a plural form of ‘to be’ at 255C13−14. But at 238C8−9 the formula ‘au)to\ kaq' au(to/’ seems to be indicating that a particular use of ‘not to be’ is at stake, in particular the complete use of the verb which may be deployed in denials of existence.76 This lends plausibility to the central claim made by the first exegesis: that at 255C13−14 the formula ‘au)ta\ kaq' au(ta/’ is involved in distinguishing the complete use of ‘to be’ from its incomplete use.

As I said, it is difficult to decide between the first and the second exegesis. I think that the considerations against the second exegesis outweigh those against the first and I therefore opt for the first exegesis. In particular, I opt for the version of the first exegesis according to which the incomplete use of ‘to be’ covers both the predicative use and the identity use. Accordingly, I take it that the identity use of ‘to be’ also involves the kind being functioning as a vowel-kind, but in combination with the kind sameness (so that whenever the incomplete ‘to be’ in its identity use is applied to x and y, what is said is that x partakes of being with respect to sameness with respect to y).77

Four apparently inconsistent pairs of sentences. The arguments showing that being, change, stability, sameness, and otherness are five pairwise distinct kinds were introduced almost as an aside with respect to the main project of this portion of the dialogue, which is to study the powers of reciprocal communion of the most important kinds (cf. 254B8−254D3). After this aside, the ES and Theaetetus return to the main project of discussing the powers of reciprocal communion of kinds: no longer of three kinds, as they had originally planned, but of five, since it has become clear that sameness and otherness are two more kinds over and above being, change, and stability. This leads to an examination of four apparently inconsistent pairs of sentences concerning change (255E8−256D10). These pairs, however, are not genuinely inconsistent because crucial expressions involved in them are used differently. In fact, in each pair both members are true. Here are the four pairs of sentences (where the reverse arrow ‘⇐’ is to be read as ‘is analyzed by’):

‘Change is not stability’ ⇐ ‘Change is other than stability’ (1) (255A11−256A2) ‘Change is’ ⇐ ‘Change partakes of being’

(2) (256A3−256B5) ‘Change is not the same’ ⇐ ‘Change is other than sameness’

75 Cf. Tht. 188D9−10; 189B1−2.

76 Cf. Thorp (1984), 89−90, who also convincingly argues that at 238C8−9 the formula ‘au)to\ kaq' au(to/’ cannot be plausibly taken to be indicating that the form of not being is being introduced.

77 Cf. Moravcsik (1962), 60; Malcolm (1967), 144−5. For the distinction between the predicative and the identity uses of ‘to be’ cf. below, text to n. 84.

101

‘Change is the same’ ⇐ ‘Change is the same as itself ’

‘Change is not other’ ⇐ ‘Change is other than otherness’ (3) (256C4−9) ‘Change is other’ ⇐ ‘Change is other than otherness’

‘Change is not being’ ⇐ ‘Change is other than being’ (4) (256C10−256D10) ‘Change is a being’ ⇐ ‘Change partakes of being’

Negating the incomplete ‘to be’ implies non-existence. Here is the discussion of pair (1):

ES âWde dh\ le/gwmen e)piì tw½n pe/nte kaq' eÁn a)nalamba/- nontej.

THT. Pw½j;

255E8

E10 ES Prw½ton me\n ki¿nhsin, wj eÃsti panta/pasin eÀteron

sta/sewj. hÄ pw½j le/gomen; THT. OuÀtwj.

ES Ou) sta/sij aÃr' e)sti¿n. THT. Ou)damw½j.

E15

ES ãEsti de/ ge dia\ to\ mete/xein tou= oÃntoj. THT. ãEstin.

256A1 A2

ES Let us then speak like this of the five [sc. kinds], taking them up one by one. THT. How?

255E

ES Let us first say that change is completely other than stability. Or how do we speak of it? THT. Like this.

ES Therefore it is not stability. THT. No way.

ES But it is because it partakes of being. THT. It is.

256A

The second member of pair (1), the affirmative sentence ‘Change is’ of 256A1, was endorsed earlier in the dialogue. In one of its earlier occurrences (at 250B8−11) it was understood in accordance with its ‘generalizing’ reading, whereby it means ‘Everything that partakes of change is’.78 But in its last earlier occurrence (at 255B12−255C1) ‘Change is’ was probably understood in accordance with its ‘ordinary’ reading, whereby it means ‘Change itself is a thing that is’.79 Hence the ‘ordinary’ reading should probably be adopted also for the second member of pair (1), the affirmative sentence ‘Change is’ of 256A1, and therefore for the first member of pair (1), the negative sentence ‘Change is not stability’, which thereby turns out to mean ‘Change itself is a thing that is not stability’. In fact, this is also true (because change is a kind that is other than the kind stability). Since on earlier occasions the sentence ‘Change and stability are’ has been

78 Cf. above, paragraph to n. 34.

79 Cf. above, paragraph to n. 64.

102

understood as making the claim that change and stability exist, ‘Change is’ of 256A1 should probably also be understood as making an existential claim.80

We are expected to have some uneasiness at accepting both ‘Change is’, i.e. ‘Change itself is a thing that exists’, and ‘Change is not stability’, i.e. ‘Change itself is a thing that is not stability’. Our expected uneasiness may be plausibly brought back to an inclination we are supposed to have to think that an incomplete ‘not to be’, as in ‘Change itself is a thing that is not stability’ (where ‘not to be’ is incomplete because of the complement ‘stability’), implies a denial of existence, as in ‘Change itself is a thing that does not exist’. As we shall see, the view that an incomplete ‘not to be’ implies a denial of existence will turn out to be one of the primary targets of Plato’s attack on the sophist’s position.

The ‘is’ of predication and the ‘is’ of identity. According to many logicians and philosophers of language, there are two importantly different incomplete uses of ‘to be’ and ‘not to be’: the predicative use and the identity use. On the predicative use, if a form of ‘to be’ is predicated of an object then what is said about this object is that it instantiates the attribute specified by the expression that complements the form of ‘to be’ (e.g. in ‘Socrates is astute’ Socrates is said to instantiate astuteness), while if a form of ‘not to be’ is predicated of an object then what is said about this object is that it does not instantiate the attribute specified by the expression that complements the form of ‘not to be’ (e.g. in ‘Theaetetus is not handsome’ Theaetetus is said not to instantiate handsomeness). On the identity use, if a form of ‘to be’ is predicated of an object then what is said about this object is that it is identical with whatever object is referred to by the expression that complements the form of ‘to be’ (e.g. in ‘Mark Twain is Samuel Clemens’ Mark Twain is said to be identical with Samuel Clemens), while if a form of ‘not to be’ is predicated of an object then what is said about this object is that it is other than whatever object is referred to by the expression that complements the form of ‘to be’ (e.g. in ‘Shakespeare is not Marlowe’ Shakespeare is said to be other than Marlowe). In the case of the identity use, often an occurrence of a form of ‘to be’ can be replaced with one of the corresponding form of ‘to be the same as’ without change in meaning, and similarly with ‘not to be’ and ‘to be other than’. Some cases are borderline: should the ‘is’ in ‘Socrates is a man’ be regarded as a form of ‘to be’ in its predicative or its identity use? My instinct is to consider it as a form of ‘to be’ in its predicative use, but one might also argue for the other alternative.81

In at least three of Plato’s four pairs of apparently inconsistent sentences it seems to be the case that the first member (the negative one) involves a form of ‘not to be’ in its identity use whereas the second member (the affirmative one) involves a form of ‘to be’ in its predicative use.

Nowhere in the Sophist does Plato use a sentence like ‘The verb “to be” is used in two ways: sometimes it is used to say that a certain object partakes of a certain form, on other occasions to say that a certain object is the same as a certain object’. Nevertheless, there are reasons for

80 For the existential use of ‘to be’ in ‘Change is’ at 256A1, cf. Rosen (1983), 274−6; Brown (1986/99), 471−3; O’Brien (1995), 63; Szaif (1998), 352−3; O’Brien (1999), 30−1.

81 Cf. Frege (1892), 193−5; Russell (1903), 64; Wittgenstein (1922), 3.323.

103

thinking that Plato in the Sophist is aware of the distinction between the ‘is’ of predication and the ‘is’ of identity. Here is the key passage (256A3−256B5), which constitutes Plato’s discussion of pair (2):

ES AuÅqij dh\ pa/lin h( ki¿nhsij eÀteron tau)tou= e)stin. THT. Sxedo/n.

256A3

ES Ou) tau)to\n aÃra e)sti¿n. THT. Ou) ga\r ouÅn.

A5

ES ¹Alla\ mh\n au(tv= g' hÅn tau)to\n dia\ to\ mete/xein auÅ pa=n tau)tou=.82

THT. Kaiì ma/la.

ES Th\n ki¿nhsin dh\ tau)to/n t' eiånai kaiì mh\ tau)to\n o(mologhte/on kaiì ou) dusxerante/on. ou) ga\r oÀtan eiãpwmen

A10

au)th\n tau)to\n kaiì mh\ tau)to/n, o(moi¿wj ei¹rh/kamen, a)ll' o(po/tan me\n tau)to/n, dia\ th\n me/qecin tau)tou= pro\j e(auth\n ouÀtw le/gomen, oÀtan de\ mh\ tau)to/n, dia\ th\n koinwni¿an auÅ

256B1

qate/rou, di' hÁn a)poxwrizome/nh tau)tou= ge/gonen ou)k e)keiÍno a)ll' eÀteron, wÐste o)rqw½j auÅ le/getai pa/lin ou) tau)to/n.

THT. Pa/nu me\n ouÅn.

B5 ES Again, change is other than the same. THT. Pretty much.

256A

ES Hence it is not the same. THT. Surely not.

ES But it was83 the same as itself because everything partakes of the same. THT. Definitely.

ES Hence we must agree that change is both the same and not the same and we must not be afraid. For, when we call it the same and not the same, we are not speaking likewise. But, when we call it the same, we speak thus because of its participation in the same with respect to itself, whereas when we call it not the same, we speak thus because of its communion with otherness, whereby, being separated from the same, it becomes not it but other, so that again it is correctly called not the same.

THT. Definitely.

256B

This passage consists of four remarks by the ES and Theaetetus’ responses to them. It is worth noting that the expression ‘the same’ (‘tau)to/n’) can function both as a proper name of the kind sameness and as an adjective used to say about something that it is the same as something.

Particularly important is the ES’s fourth remark. Before commenting on it as a whole, one must interpret the final part of it. Earlier on the ES had declared that there might be some kinds which are ‘causes of the division’ (253C3) between kinds (vowel-kinds that are causes of division). Now (256B2−3) he says that change is separated from the same, i.e. from sameness, because of its communion with otherness. This suggests that otherness is a kind causing division or separation (which, I take it, are the same thing) and that, in general, if x partakes of otherness with respect to y then x is separated from y. In particular, change is separated from the same (i.e. sameness)

82 The reading ‘au(tv= g' hÅn tau)to\n dia\ to\ mete/xein auÅ pa=n tau)tou=’ is an emendation by Madvig, adopted by Robinson. The main MSS have ‘auÀth g' hÅn tau)to\n dia\ to\ mete/xein auÅ pa=nt' au)tou=’.

83 The past tense ‘was’ probably implies a cross-reference to an earlier passage: cf. 254D14−15.

104

because it partakes of otherness with respect to the same (i.e. sameness), namely because it is other than the same (i.e. sameness). Thus, the final part of the ES’s fourth remark says that change is not the same (i.e. sameness) because it is other than the same (i.e. sameness). This is confirmed by the first and the second of the ES’s remarks, where we find that ‘change is other than the same’ (256A3) and therefore ‘is not the same’ (256A5). A further confirmation comes from 256C5−10 (cf. below), where from the premiss that change is other than otherness Plato infers that change is not other.

We may now come to the interpretation of the ES’s fourth remark as a whole. It consists of four sentences. In the first (‘We must agree without any qualms that change is both the same and not the same’, 256A10−11) we are told that change is both the same and not the same: this seems a contradiction, but we are invited to accept it without quarrelling. The second sentence (‘For, when we call it the same and not the same, we are not speaking likewise’, 256A11−12) provides an indication of why we must accept that change is both the same and not the same: it is because different ways of speaking are involved. The third sentence (‘When we call it the same, we speak thus because of its participation in the same with respect to itself ’, 256A12−256B2) explains the way of speaking on the basis of which we may accept that change is the same: we may accept that change is the same because change partakes of the same with respect to itself, i.e. because change is the same as itself. Finally, the fourth sentence (‘But when we call it not the same, we speak thus because of its communion with otherness, whereby, being separated from the same, it becomes not it but other, so that again it is correctly called not the same’, 256B2−4) explains the way of speaking on the basis of which we may accept that change is not the same: we may accept that change is not the same because change partakes of otherness with respect to the same (i.e. sameness), i.e. because change is other than the same (i.e. sameness).

By saying that ‘we are not speaking likewise’ (256A11−12), Plato recognizes that the sentences where ‘the same’ and ‘not the same’ are attributed involve different linguistic uses: since the first sentence, ‘Change is the same’, seems to involve a form of ‘to be’ in its predicative use whereas the second, ‘Change is not the same’, seems to involve a form of ‘to be’ in its identity use, it is natural to suggest that the different linguistic uses Plato has in mind are the predicative and the identity uses of ‘to be’. This distinction is precisely the one that was needed to solve the paradox of the late-learners (although Plato himself does not draw an explicit connection between the distinction and the paradox).84

Some commentators object that Plato cannot be distinguishing uses of ‘to be’ because he does not employ the verb where he ought to if he were distinguishing uses of it. For consider some of the formulations he adopts: ‘When we call it the same and not the same …’ (256A11−12), ‘When

84 Cf. Ackrill (1957), 82−4; Ryle (1960), 446; Moravcsik (1962), 51; Crombie (1963), 400, 499; Malcolm (1967), 145; Vlastos (1970), 288; Owen (1971), 256−7; von Weizsäcker (1973), 234−5; Gómez-Lobo (1977), 39−40; Ray (1984), 66−8; Crivelli (1990), 40, 59; Bordt (1991), 526−7; Fronterotta (2007), 436−7.

105

we call it the same …’ (256B1), ‘When we call it not the same …’ (256B2). The verb ‘to be’ is not in sight.85

There are two possible replies to this objection. First, one may accept the point made by the objection and modify accordingly one’s interpretation of what Plato is doing. Specifically, one may claim that what Plato is distinguishing is not the predicative use of ‘to be’ from its identity use, but statements of predication (taken as wholes) from statements of identification (also taken as wholes): the statements in question do employ forms of ‘to be’, but the responsibility for their difference is not laid upon ‘to be’. Plato might be merely pointing out that there is a difference between a statement of predication and one of identification without thinking of any component or factor as primarily responsible for this difference. Alternatively, he might think of a certain component or factor as primarily responsible for the difference between statements of identity and statements of predication, one however that is not the verb ‘to be’: in particular, he might think that in a statement of identity what follows the occurrence of a form of ‘to be’ names something whereas in a statement of predication what follows such an occurrence describes something.86

The second possible reply to the objection is based on noting two facts: first, the verb ‘to be’ occurs in the initial part of the passage, where the ES says about change that ‘it is not the same’ (256A5), that ‘it was the same as itself ’ (256A7), and that it ‘is both the same and not the same’ (256A10); secondly, Plato often uses forms of instances of ‘to call ϕ ψ’ or ‘to speak of ϕ as ψ’ (‘le/gein ϕ ψ’) in circumstances where he could as well have used the corresponding forms of instances of ‘to say that ϕ is ψ’ (cf. 222B9−10; 228B8−9; 229E3; 244A1−2; 244B6; 249A6; 250A8−9; 257D9−10).87 In view of these two facts it may be safely assumed that when in the second part of our passage he uses forms of ‘to call it the same’ and ‘to call it not the same’, Plato intends them to be stylistic variants of the corresponding forms of ‘to say that it is the same’ and ‘to say that it is not the same’. One might still wonder why Plato omits the verb ‘to be’ in the second part of the passage: would his point not have been clearer if he had always expressed the verb? There are two possible answers to this question. First, in the second part of the passage Plato intends to focus on the opposed complements of ‘to be’, namely ‘the same’ and ‘not the same’: what constitutes a puzzle is that what change is called is ‘the same and not the same’ (256A12), and the fact that this is done by using ‘to be’ becomes provisionally secondary (although, in the end, it will be the distinction of uses of ‘to be’ that will solve the puzzle). Secondly, Plato has a reason for withholding the forms of ‘to be’ in the second part of our passage. Our passage as a whole, 256A3−256B5, belongs to the initial stage of a long argument that runs from 255E8 to 257A12. The conclusion of this long argument, reached at 257A1−12, is that being is not (for it is not indefinitely many kinds because it is other than them). Since this

85 Cf. Owen (1971), 258; Gosling (1973), 219; Lewis (1975), 141; Ketchum (1978), 46; Prior (1980), 200−1; Bostock (1984), 95−6; Brown (1986/99), 471.

86 Cf. Owen (1971), 251, 257−8; Brown (1986/99), 471.

87 Cf. van Eck (2000), 71.

106

conclusion will be hard to swallow for an Eleatic, in the initial stage of the long argument leading to it Plato avoids negating ‘to be’ in a way that an Eleatic might find objectionable. So, in this initial stage of the long argument, Plato avoids using forms of instances of ‘not to be ϕ’ with ‘not’ governing ‘to be’. In their place he either employs forms of instances of ‘to be not ϕ’, where ‘not’ governs what replaces ‘ϕ’ (cf. 255E14, 256A5, 256A10, and 256C7), or (as in the second half of our passage) resorts to locutions constructed around forms of instances of ‘to call not ϕ’, thereby omitting the verb ‘to be’ altogether. In a later stage of the argument (at 256C10−256D10) Plato will come to use forms of that instance of ‘to be not ϕ’ which is ‘to be (a) not being’, and after this he will feel authorized to use forms of instances of ‘not to be ϕ’ with ‘not’ governing ‘to be’.88 (In English, although one can distinguish ‘to be not ϕ’ from ‘not to be ϕ’, one cannot distinguish ‘is not ϕ’ from ‘not is ϕ’: ‘not’ must always be placed after ‘is’ because ‘Socrates not is beautiful’ is ungrammatical. In Greek, by contrast, the word corresponding to ‘not’, ‘ou)’ or ‘ou)k’, may be placed in front of the word corresponding to ‘is’, ‘e)sti¿’ or ‘e)sti¿n’: ‘Swkra/thj ou)k eÃsti kalo/j’ is grammatical.)89

It is also worth observing that shortly after the passage we are considering Plato discusses pair (3) by saying:

ES Le/gwmen dh\ pa/lin: h( ki¿nhsi¿j e)stin eÀteron tou= e(te/rou, kaqa/per tau)tou= te hÅn aÃllo kaiì th=j sta/sewj;

THT. ¹AnagkaiÍon.

256C4 C5

ES Ou)x eÀteron aÃr' e)sti¿ pv kaiì eÀteron kata\ to\n nundh\ lo/gon.

THT. ¹Alhqh=.

C9

ES Let us then say again: change is other than otherness, just as it was other than the same and stability.

THT. Necessarily.

256C

ES Hence, according to the present argument, it is somehow not other and other. THT. True.

The remarks on change and otherness here at 256C4−9 are parallel to those on change and sameness at 256A3−256B5. In the sentence ‘It [sc. change] is somehow [pv] not other and other’ (256C7), the adverb ‘somehow’ modifies ‘is’: Plato seems thereby to be signaling that different uses of ‘to be’ are involved.90 Interestingly enough, one and the same situation, i.e. the fact change is other than otherness, justifies both apparently inconsistent sentences, ‘Change is not other’ and ‘Change is other’.

Similar observations may be made with regard to Plato’s discussion of pair (4):

88 Cf. van Eck (2000), 72−4.

89 Cf. Owen (1971), 234.

90 In some of its occurrences in the Sophist the indefinite adverb ‘somehow’ (‘pv’) might be taken to mean ‘in some sense’: cf. 241D7; 256B6; 259C9.

107

ES Ti¿ ouÅn dh\ to\ meta\ tou=to; aÅr' auÅ tw½n me\n triw½n eÀteron au)th\n fh/somen eiånai, tou= de\ teta/rtou mh\ fw½men, o(mologh/santej au)ta\ eiånai pe/nte, periì wÒn kaiì e)n oiâj

256C10

256D1 prouqe/meqa skopeiÍn;

THT. Kaiì pw½j; a)du/naton ga\r sugxwreiÍn e)la/ttw to\n a)riqmo\n tou= nundh\ fane/ntoj.

ES ¹Adew½j aÃra th\n ki¿nhsin eÀteron eiånai tou= oÃntoj diamaxo/menoi le/gwmen;

THT. ¹Adee/stata me\n ouÅn.

D5

ES Ou)kou=n dh\ safw½j h( ki¿nhsij oÃntwj ou)k oÃn e)sti kaiì oÃn, e)pei¿per tou= oÃntoj mete/xei;

THT. Safe/stata/ ge.

D10

ES What about what comes after this? Shall we say that it [sc. change] is other than the three and deny that it is other than the fourth, having agreed that those about and on which we propose to inquire are five?

256C 256D

THT. How could we? For it is impossible to concede that they are fewer in number than has appeared just now.

ES Must we then fearlessly contend and say that change is other than being? THT. Most fearlessly.

ES Is it then clearly the case that change really is not being and being, since it partakes of being? THT. Most clearly.91

The important fact here is that Plato uses a form of ‘to be’ in expressing the claim that ‘change really is not being and being’ (256D8−9).

Both replies to the objection formulated five paragraphs back are acceptable. The considerations in support of the second solution seem to me marginally more weighty, so that I opt for retaining the view that Plato is drawing attention to the difference between the predicative and the identity uses of ‘to be’.

Both sentences involved in pair (2), ‘Change is not the same’ and ‘Change is the same’, are probably supposed to be understood in accordance with their ‘ordinary’ reading. For the first member of this pair, the sentence ‘Change is not the same’, is probably picking up a result which had been proved earlier (at 255A4−255B7). But the result established in that earlier passage was that change itself is a thing that is not sameness (not that everything that partakes of change is not sameness). This strongly suggest that when the sentence ‘Change is not the same’ is used to pick up this earlier result, it is to be understood in accordance with its ‘ordinary’ reading, whereby it means ‘Change itself is a thing that is not the same’. And from this it may be plausibly inferred that the second member of pair (2), namely the sentence ‘Change is the same’, is also supposed to be understood in accordance with its ‘ordinary’ reading, whereby it means ‘Change itself is a thing that is the same’, a conclusion which moreover fits well with the justification which Plato offers for it, namely an appeal to change’s ‘participation in the same with respect to itself ’ (256B1). Similar considerations apply to pairs (3) and (4): in their case also both sentences involved are probably supposed to be understood in accordance with their ‘ordinary’ reading.

91 Later (259A6−259B1) Plato puts forward similar similar considerations with otherness in place of change: otherness is because it partakes of being, but is a not being because it is other than being.

108

Is change stable? The following exchange between the ES and Theaetetus occurs at 256B6−256C3, between the discussions of pairs (2) and (3):

ES Ou)kou=n kaÄn eiã pv metela/mbanen au)th\ ki¿nhsij sta/sewj, ou)de\n aÄn aÃtopon hÅn sta/simon au)th\n prosago- reu/ein;

256B6

THT. ¹Orqo/tata/ ge, eiãper tw½n genw½n sugxwrhso/- meqa ta\ me\n a)llh/loij e)qe/lein mei¿gnusqai, ta\ de\ mh/.

B10

ES Kaiì mh\n e)pi¿ ge th\n tou/tou pro/teron a)po/deicin hÄ tw½n nu=n a)fiko/meqa, e)le/gxontej w¨j eÃsti kata\ fu/sin tau/tv.

THT. Pw½j ga\r ouÃ;

256C1

C3 ES So, even if change itself had somehow participated of stability, it would not have been absurd

to call it stable? 256B

THT. And most rightly, if we are to agree that some of the kinds consent to mix with one another and some do not.

ES And we arrived at the demonstration of this before the present matters, by offering a refutation to the effect that this is how they are by nature.

THT. Precisely.

256C

The first remark of the ES, ‘Even if change itself had somehow participated [kaÄn eiã pv metela/mbanen] of stability, it would not have been [ou)de\n aÄn … hÅn] absurd to call it stable’ (256B6−8), is a counterfactual conditional (‘ei¹’ with the indicative imperfect in the protasis combined with ‘aÃn’ with the indicative imperfect in the apodosis is one of the standard formulations of the counterfactual conditional).92 It therefore implies the denials of both its antecedent and its consequent. Hence the ES is committed to claim that ‘Change is stable’ is false.93

Why does the ES introduce the allegedly false ‘Change is stable’? In pairs (2), (3), and (4) the second member (‘Change is the same’, ‘Change is other’, and ‘Change is a being’) is an affirmative predicative sentence with the same subject expression and (roughly) the same predicative expression as the negative identity sentence that constitutes the first member (‘Change is not the same’, ‘Change is not other’, and ‘Change is not being’). Pair (1) is anomalous: its second member, ‘Change is’, is not an affirmative predicative sentence with the same subject expression and (roughly) the same predicative expression as the negative identity sentence that is the first member, ‘Change is not stability’. The reason for the anomaly is obvious. The ES wants both members of each of his apparently inconsistent pairs to be true, but the affirmative predicative sentence with the same subject expression and (roughly) the same predicative expression as the negative identity sentence ‘Change is not stability’ is ‘Change is stable’, a

92 Cf. Kühner and Gerth (1892−1904), II.II 469−70. For the combination of ‘kaÃn’ with ‘ei¹’ see LSJ s.v. ‘kaÃn’ I.

93 Cf. Malcolm (1985a), 85; Roberts (1986), 240; Meinhardt (1990), 238; Stough (1990), 365; O’Brien (1995), 108. Ritter (1910), 61 instead thinks that the counterfactual conditional should not be understood so strongly as to imply that ‘Change is stable’ is false.

109

sentence so far regarded as false.94 For this reason the ES replaces ‘Change is stable’ with ‘Change is’, a sentence that may be unproblematically regarded as true. Now, at 256B6−256C3, the ES considers the allegedly counterfactual situation where change is stable. In this allegedly counterfactual situation, ‘Change is stable’ would have been true, so that ‘it would not have been absurd to call it [sc. change] stable’ (256B7−8). Even in this allegedly counterfactual situation, one might have had some inclination to think that it was absurd to call change stable: for, even in this allegedly counterfactual situation, change would not have been stability (because it would still have been other than stability), and one might have taken this to exclude its being stable. The impression of absurdity would have dissolved once one had realized that change’s participation in stability was compatible with its participation in otherness than stability, and therefore with not being stability, just as there is nothing absurd in saying that change is the same despite its not being the same, because change’s participation in sameness is compatible with its participation in otherness than sameness, and therefore with not being sameness.95

Some commentators, feeling that the second part of our passage (256B9−256C3) fits awkwardly with the immediately preceding first part (256B6−9), postulate a lacuna between 256B8 and 256B9, where the two parts join.96 But the text handed down by the tradition may be plausibly preserved as it is. Earlier, at 252D2−11 (cf. 254D7−9), the claim that change is not stable was made in order to show that change cannot combine with stability, which result was in turn used to prove that it is not the case that all kinds mix. As we have seen, in the first part of our passage the ES makes a statement that presupposes that change is not stable. This presupposition is endorsed by Theaetetus at the beginning of the second part by recalling that some kinds are able to mix whereas others are not: the unmixing kinds he has in mind are change and stability.

Otherness and not being. After presenting and discussing the four pairs of apparently inconsistent sentences concerning change, Plato draws some conclusions bout not being and otherness (256D11−257A12). He begins by drawing the following conclusion:

ES ãEstin aÃra e)c a)na/gkhj to\ mh\ oÄn e)pi¿ te kinh/sewj Eiånai kaiì kata\ pa/nta ta\ ge/nh: kata\ pa/nta ga\r h( qate/rou

256D11

fu/sij eÀteron a)pergazome/nh tou= oÃntoj eÀkaston ou)k oÄn poieiÍ, kaiì su/mpanta dh\ kata\ tau)ta\ ouÀtwj ou)k oÃnta o)rqw½j

256E1

e)rou=men, kaiì pa/lin, oÀti mete/xei tou= oÃntoj, eiånai¿ te kaiì oÃnta.

THT. Kinduneu/ei. E5 ES Periì eÀkaston aÃra tw½n ei¹dw½n polu\ me/n e)sti to\ oÃn,

aÃpeiron de\ plh/qei to\ mh\ oÃn. THT. ãEoiken.

E8

94 Cf. 250B5−7; 252D6−11; 254D7−9 (where ‘fame/n’ at 254D7 refers back to the two earlier passages); 255A7−255B2.

95 Cf. Heinaman (1981), 65.

96 Cf. Ritter (1910), 61; Cornford (1935), 286−7.

110

ES It is therefore necessarily the case97 that not being be about change and with respect to all the

kinds: for, with respect to all of them, the nature of the other, by rendering each one other than what is, makes it not being, and thus we will correctly call them all together in the same way not beings, and again, since they partake of being, we will say they are and call them beings.

256D 256E

THT. Apparently. ES Therefore about each of the forms what is is a lot whereas what is not is of indefinite

multitude. THT. So it seems.

This passage contains two remarks by the ES (256D11−256E4 and 256E6−7) approved by Theaetetus (256E5 and 256E8). The first remark begins with the sentence: ‘It is […] necessarily the case that not being be about change and with respect to all the kinds’ (256D11−12). The first half of this initial sentence, ‘It is […] necessarily the case that not being be about change’ (256D11−12), is a repetition of the result established in the immediately preceding passage (256D5−10), that change is not being because it is other than being.

The first half of the initial sentence of the ES’s first remark, ‘It is […] necessarily the case that not being be about change’ (256D11−12), displays the converse use of ‘to be’,98 i.e. the use of ‘to be’ followed by ‘about’ (which translates either ‘e)pi¿’ with the genitive, as at 256D11, or ‘e)pi¿’ with the dative, as at 247D2,99 or ‘peri¿’ with the accusative, as at 248E5, 255A11, 256E6, 261B1−2, 263B12, and perhaps 249B12,100 or ‘peri¿’ with the genitive, as at 263B4−5 and 263B11, where however the near presence of ‘le/gein’ probably influences the construction). When the converse use of ‘to be’ is involved, any instance of ‘ϕ is about σ’ means the same as the corresponding instance of ‘σ is ϕ’ (where ‘ϕ’ and ‘σ’ are schematic letters to be replaced by syntactically appropriate expressions). An example of the converse of the predicative use of ‘to be’ probably occurs in the first half of the initial sentence of the ES’s first remark: ‘… that not being be about change’ probably means the same as ‘… that not being hold of change’. The converse of the predicative use of ‘to be’ will play an important role later in the dialogue, in the account of truth and falsehood.

The standard interpretation of 256D11−256E8. The second half of the initial sentence of the ES’s first remark, ‘It is […] necessarily the case that not being be […] with respect to all the kinds’

97 I take ‘eÃstin’ (256D11) as veridical: ‘it is the case that’ (cf. Stallbaum (1840), 197; Campbell (1867), Sph. 156). For ‘e)c a)na/gkhj’ modifying a veridical ‘eÃstin’, cf. 259A8−259B1; Arist. APr. 1.15, 34a17. There is no need to take ‘eÃstin’ to mean ‘it is possible’, yielding the cumbersome ‘It is therefore necessarily possible that …’.

98 Cf. Frede (1967), 53; Lewis (1976), 110.

99 At 252D7−8 ‘e)p' a)llh/loin’ could could be both a case of ‘e)pi¿’ with the genitive and one of ‘e)pi¿’ with the dative.

100 Cf. above, n. 50.

111

(256D11−12), expresses a claim concerning all kinds. According to its usual interpretation,101 which I shall call its ‘standard interpretation’, it is a generalization to all kinds, or rather to all kinds other than being, of the claim which has just been made about change: the claim just made about change is that not being is about it in that it is not being because it is other than being; the generalization is that not being is about every kind F other than being in that F is not being because it is other than being.

A difficulty for the standard interpretation of 256D11−256E8. The standard interpretation of 256D11−256E8 faces a difficulty. In his second remark, ‘Therefore about each of the forms what is is a lot whereas what is not is of indefinite multitude’ (256E6−7), the ES draws a conclusion based on what precedes (note the initial ‘therefore’). This conclusion has an affirmative and a negative component. Let us focus on the negative component, i.e. the claim that ‘about each of the forms […] what is not is of indefinite multitude’ (256E6−7). What is the argument leading to this claim? If one accepts the standard interpretation, one is inclined to think that the argument starts with the premiss that every kind apart from being is other than being and therefore is a non-being, but then one cannot see how the result that ‘about each of the forms […] what is not is of indefinite multitude’ (256E6−7) can be reached with even a semblance of validity.

At least one portion of the argument for the claim that ‘about each of the forms […] what is not is of indefinite multitude’ (256E6−7) can be plausibly reconstructed. For it may be plausibly assumed that the argument involves the following three steps: first, it claims that each kind is other than indefinitely many kinds (i.e. all kinds apart from itself ); secondly, it infers that each kind is not indefinitely many kinds; thirdly, it infers that ‘about each of the forms […] what is not is of indefinite multitude’ (256E6−7) (where the converse of the identity use of ‘not to be’ is brought in). The inference from the first to the second step, i.e. the inference from the claim that each kind is other than indefinitely many kinds to the claim that each kind is not indefinitely many kinds, relies on a use of ‘not to be’ with ‘not’ governing ‘to be’ according to which for every kind F and every kind K, ‘not to be’ is true of F and K just in case F is other than K (cf. 257A4−6 and 259B1−5). However, if the standard interpretation of the second half of the initial sentence of the ES’s first remark (256D11−12) is right, then nowhere in what precedes the second remark do we find such a use of ‘not to be’ introduced or explained. So, if the standard interpretation is right, then the ES has no warrant for inferring the conclusion that constitutes his second remark.102

The first defense of the standard interpretation of 256D11−256E8. Some commentators defend the standard interpretation of 256D11−256E8.103 Their defense is based on attributing to Plato a use of ‘not to be’ with ‘not’ governing ‘to be’ according to which for every kind F and every kind K, if

101 Cf. Campbell (1867), Sph. 156; Vlastos (1970), 289; Seligman (1974), 76; Lewis (1976), 91−2, 94, 109; Ray (1984), 68; Brown (1986/99), 473; Movia (1991), 376−7; O’Brien (1995), 50−3; Szaif (1998), 432.

102 Cf. Frede (1967), 85; van Eck (2002), 66−8.

103 Cf. Campbell (1867), Sph. 156−7; O’Brien (1995), 52; Szaif (1998), 450.

112

F is other than K then ‘not to be’ is true of F because F is a non-K and K is a being. On the basis of this use of ‘not to be’ and of the fact that every kind is other than indefinitely many kinds, one gets the desired result that ‘about each of the forms […] what is not is of indefinite multitude’ (256E6−7).

This defense of the standard interpretation is unsatisfactory. First, it has the drawback of reaching the desired result by the wrong route. The result that ‘about each of the forms […] what is not is of indefinite multitude’ (256E6−7) must be reached on the basis of the inference from the claim that each kind is other than indefinitely many kinds to the claim that each kind is not indefinitely many kinds. This inference, in turn, relies on a use of ‘not to be’ with ‘not’ governing ‘to be’ according to which for every kind F and every kind K, ‘not to be’ is true of F and K just in case F is other than K. But, according to the defense we are now considering, the desired result is not reached on the basis of this use of ‘not to be’ with ‘not’ governing ‘to be’, but on the basis of another such use: that according to which for every kind F and every kind K, if F is other than K then ‘not to be’ is true of F because F is a non-K and K is a being. The difference between the two uses of ‘not to be’ may be spotted by noting that the second use does not enable one to infer ‘Change is not stability’ with ‘not’ governing ‘is’ from ‘Change is other than stability’: it only enables one to infer either ‘Change is not’ with ‘not’ governing ‘is’ or ‘Change is a non-being’. The defense of the standard interpretation we are now considering has a second drawback: the use of ‘not to be’ with ‘not’ governing ‘to be’ it attributes to Plato is not explained or introduced in the text: it must be supplied by the reader. Such a demand is rather implausible. This defense must therefore be discarded.104

The second defense of the standard interpretation of 256D11−256E8. Other commentators propose a different defense of the standard interpretation.105 This second defense involves noting that the ES has found a justification for placing ‘not’ in such a way that it governs ‘to be’ in a description of all kinds other than being. Our passage, 256D11−256D8, belongs to a long argument running from 255E8 to 257A12. The conclusion of this argument, reached at 257A1−12, is that being is not (for it is not indefinitely many kinds because it is other than them). As I pointed out earlier,106 since this conclusion will be hard to accept for an Eleatic, in the initial stage of the argument (255E8−256C9) Plato avoids negating ‘to be’ in a way that an Eleatic might find objectionable: he avoids using forms of instances of ‘not to be ϕ’ with ‘not’ governing ‘to be’ and in their place he either employs forms of instances of ‘to be not ϕ’ with ‘not’ governing what replaces ‘ϕ’ (cf. 255E14, 256A5, 256A10, and 256C7) or resorts to locutions constructed around forms of instances of ‘to call not ϕ’, thereby omitting the verb ‘to be’ altogether (cf. 256A11−12 and 256B2). So in the first stage of the argument Plato infers instances of ‘Change is not ϕ’ with ‘not’ governing what replaces ‘ϕ’ from the corresponding instances of ‘Change is other than ϕ’

104 Further criticisms of the first defense of the standard interpretation may be found in van Eck (2002), 67−8.

105 Cf. McDowell (1982), 117−8.

106 Cf. above, text to n. 88.

113

(‘ϕ’ is replaced by ‘stability’, ‘the same’, and ‘otherness’). In the second stage of the argument (256C10−256D10) Plato comes to the case where ‘ϕ’ is replaced by ‘being’: he infers ‘Change is (a) not being’ with ‘not’ governing ‘being’ from ‘Change is other than being’. Plato thereby finds himself entitled to place ‘not’ in front of ‘to be’ in a description of change (cf. 256D8): he now has a use of ‘not to be’ with ‘not’ governing ‘to be’ according to which for every kind F, ‘not to be’ is true of F just in case F is other than being. In the third stage of the argument (256D11−256E5), Plato immediately employs this use of ‘not to be’ with ‘not’ governing ‘to be’ to describe all kinds other than being (cf. 256D11−12, 256E1, and 256E2). Once the ice has been thus broken, in the fourth stage of the argument (256E6−8) Plato feels entitled to introduce a new use of ‘not to be’ with ‘not’ governing ‘to be’, that according to which for every kind F and every kind K, ‘not to be’ is true of F and K just in case F is other than K. As I said earlier, such a use of ‘not to be’ is what is needed for establishing the result that ‘about each of the forms […] what is not is of indefinite multitude’ (256E6−7). From this in the fifth and final stage of the argument (257A1−12) Plato reaches his conclusion, namely that ‘being, however many the others are, with respect to as many is not’ (257A4−5).

This second defense of the standard interpretation is far more convincing than the first. It too, however, fails to be completely persuasive. The problem has to do with the passage from the third (256D11−256E5) to the fourth (256E6−8) stage of the argument: Plato can hardly be justified in using ‘therefore’ (256E6) to introduce a sentence where he places ‘not’ in such a way as to govern ‘to be’ in a situation that is profoundly different from the one in which he had shown such a construction to be justified.

The interpretation of 256D11−256E8 proposed by Job van Eck. Given that the standard interpretation of 256D11−256E8 faces such difficulties, there are reasons for looking for an alternative exegesis. Job van Eck,107 who has noticed the difficulties faced by the standard interpretation, urges to avoid regarding the second half of the initial sentence of the ES’s first remark as a generalization of what was said in the first half: when the ES says that ‘it is […] necessarily the case that not being be about change and with respect to all the kinds’ (256D11−12), it is not the case that he is first making a claim about the kind change (i.e. that it is not being because it is other than being) and then generalizing it to a claim about all kinds other than being (i.e. that each one of them is not being because it is other than being). Rather, by saying that ‘it is […] necessarily the case that not being be about change and with respect to all the kinds’ (256D11−12), the ES introduces two types of not being that are about change: on the one hand, the not being that is about change in that change is other than being; on the other hand, the not being with respect to all the kinds, or rather with respect to all kinds apart from change, which is also about change. The second kind of not being calls for an explanation, which is in fact offered by the ES in the continuation of his first remark: ‘For, with respect to all of them, the nature of the other, by rendering each one other than what is, makes it not being’ (256D12−256E2). According to van Eck, the ES’s explanation here is that for every kind F and

107 Cf. van Eck (2000), 73−4; van Eck (2002), 69−70.

114

every kind K other than F, F is not being with respect to K in that F is other than being with respect to K (e.g. change is other than stability so that change is not being with respect to stability in that it is other than being with respect to stability, i.e. than being stable, and change is other than sameness so that change is not being with respect to sameness in that it is other than being with respect to sameness, i.e. than being the same, and so on for all kinds other than change). This interpretation avoids the difficulty faced by the standard interpretation because it warrants the negative component of the conclusion inferred by the ES: since every kind F is other than being with respect to K for every kind K other than F, it follows that every kind F is not being with respect to K for every kind K other than F, and this makes true the claim that constitutes the negative component of the inferred conclusion, i.e. the claim that ‘about each of the forms […] what is not is of indefinite multitude’ (256E6−7).

Should we endorse van Eck’s exegesis? As I pointed out earlier, what a satisfactory interpretation must give to Plato is an explanation of a use of ‘not to be’ with ‘not’ governing ‘to be’ according to which for every kind F and every kind K, ‘not to be’ is true of F and K just in case F is other than K. As we have seen, van Eck’s interpretation does give Plato an explanation of a use of ‘not to be’ with ‘not’ governing ‘to be’ according to which for every kind F and every kind K, if F is other than K then ‘not to be’ is true of F and K (because if F is other than K then F is other than being with respect to K ). But it remains unclear whether the argument given to Plato by van Eck warrants a use of ‘not to be’ with ‘not’ governing ‘to be’ which supports the converse implication, i.e. that for every kind F and every kind K, if ‘not to be’ is true of F and K then F is other than K. More precisely, it remains unclear whether the use of ‘not to be’ with ‘not’ governing ‘to be’ warranted by the argument given to Plato by van Eck requires that for every kind F, no form of ‘not to be’ on this use is true of F and F itself: this ought of course to be required (cf. 257A4−5 and 259B1−5). Suppose that every kind F is identical with being with respect to F: then, for any kind F, since F is identical with being with respect to F, it follows that F is not other than being with respect to F, so that no form of ‘not to be’ on the use of the expression warranted by the argument given to Plato by van Eck is true of F and F itself. Thus, if every kind F is identical with being with respect to F, the desired result holds. Suppose instead that it is not the case that every kind F is identical with being with respect to F, so that some kind F is other than being with respect to F: then, for this kind F, forms of ‘not to be’ on the use of the expression warranted by the argument given to Plato by van Eck are true of F and F itself. Thus, if it is not the case that every kind F is identical with being with respect to F, the desired result fails. So, whether the desired result holds depends on whether every kind F is identical with being with respect to F. But this is controversial. One might argue that every kind F is identical with being with respect to F because there is no recognizable difference between a kind F and being with respect to F: what is the difference between beauty and being beautiful? One might however argue for the contrary claim that every kind F is other than being with respect to F: for F is other than its combination with any kind, even with the kind being. Of course, it is not always the case that an item is other than its ‘combination’ with any item: for instance, every set is identical with its union with the empty set. But it seems unlikely that a notion of combination analogous to that of set-theoretical union is what Plato has in mind with regard to kinds. In fact, the analogy drawn by Plato (at 252E9−253A12 and 253B9−253C5) between kinds and letters suggests that

115

just as the syllable ‘at’ is different from the letter ‘t’, so also the combination of the vowel-kind being with any kind F is different from F.108 I think that an interpretation that attributes to Plato an argument whose validity depends on an assumption as controversial as the one to the effect that every kind F is identical with being with respect to F should be avoided. In conclusion, we should reject van Eck’s exegesis.109 We must search further.

An alternative interpretation of 256D11−256E8. Several points of van Eck’s interpretation of 256D11−256E8 may be plausibly retained. First, it stands to reason that the second half of the initial sentence of the ES’s first remark should not be a generalization of what is said in its first half. Secondly, it stands to reason that by saying that ‘it is […] necessarily the case that not being be about change and with respect to all the kinds’ (256D11−12), the ES should be introducing two types of not being that are about change: the not being that amounts to otherness than being, on the one hand, and some sort of not being with respect to all kinds, or rather with respect to all kinds apart from change, on the other. Thirdly, it is reasonable to expect an explanation of the second type of not being and to spot it in the ES’s continuation of his first remark: ‘For, with respect to all of them, the nature of the other, by rendering each one other than what is, makes it not being’ (256D12−256E2). But at this juncture one should part company with van Eck. As an alternative to his exegesis, I make two suggestions about the occurrence of the expression ‘what is’ in the ES’s explanation, i.e. in the sentence ‘With respect to all of them, the nature of the other, by rendering each one other than what is, makes it not being’ (256D12−256E2): first, this occurrence of ‘what is’ is not one of a name or a of definite description that picks out the kind being, but of a universal quantifier (‘everything that is’); secondly, this occurrence of ‘what is’ involves the identity use of ‘to be’. If these suggestions are correct, then the explanation provided by the ES is that every kind F is not with respect to every other kind K in that F is other than what is K, i.e. than everything that is K by being the same as K (whereby ‘what is’ is treated as a universal quantifier and as involving the identity use of ‘to be’). As I pointed out earlier, what Plato needs for his argument to go through is an explanation of a use of ‘not to be’ with ‘not’ governing ‘to be’ according to which for every kind F and every kind K, ‘not to be’ is true of F and K just in case F is other than K. The interpretation I am now proposing gives Plato such an explanation. This may be seen by considering that the present interpretation gives Plato an explanation of a use of ‘not to be’ with ‘not’ governing ‘to be’ that supports the following chain of

108 Even if Ryle (1960), 434−6 were right when he claims that an important aspect of Plato’s analogy between kinds and letters is that consonants (or at least some consonants) cannot be pronounced without an accompanying vowel, it would still be the case that the analogy requires that the combination of the kind being with any kind F be different from F: just as a consonant is distinct from its combination with any specific vowel (otherwise it could be pronounced in combination with only one vowel), so every kind F is distinct from its combination with the kind being (otherwise the kind being would be the only vowel- kind with which it can be combined, contrary to Plato’s indication at 253C1−3).

109 Matters are even worse for van Eck’s exegesis if for some kind K, being with respect to K is other than K and is a kind. Since being with respect to K is surely not other than being with respect to K, it follows that the principle on which van Eck’s exegesis relies is false: it is not the case that every kind F is other than being with respect to K for every kind K other than F.

116

equivalence, for every kind F and every kind K: ‘not to be’ is true of F and K just in case F is other than everything that is K by being identical with K just in case F is other than K.

It follows that change is not all other kinds, so that not being with respect to all other kinds holds of change, i.e. not being with respect to all other kinds is about change (converse of the predicative use of ‘to be’): the second half of the initial sentence of the ES’s first remark, ‘It is […] necessarily the case that not being be about change […] with respect to all the kinds’ (256D11−12), is justified.

It also follows that the ES is justified in inferring the general conclusion about being and not being in connection with all kinds which constitutes his second remark: ‘Therefore about each of the forms what is is a lot whereas what is not is of indefinite multitude’ (256E6−7). On the one hand, every kind partakes of many kinds and therefore is many kinds, so that about it ‘what is is a lot’. On the other hand, every kind is other than indefinitely many kinds and therefore is not indefinitely many kinds, so that about it ‘what is not is of indefinite multitude’. In both these parallel inferences the last step involves the switch from the direct to the converse use of ‘to be’ and ‘not to be’, whereby any instance of ‘ϕ is about σ’ means the same as the corresponding instance of ‘σ is ϕ’ and any instance of ‘ϕ is not about σ’ means the same as the corresponding instance of ‘σ is not ϕ’. Specifically, in the affirmative component of the inferred conclusion the converse of the predicative use of ‘to be’ may be seen in the claim that ‘about each of the forms what is is a lot’, whereas in the negative component the converse of the identity use of ‘not to be’ may be seen in the claim that ‘about each of the forms […] what is not is of indefinite multitude’.110

Three things are worth noting about my interpretation of 256D11−256E8. (1) According to my interpretation, in the Greek sentence ‘kata\ pa/nta […] h( qate/rou

fu/sij eÀteron a)pergazome/nh tou= oÃntoj eÀkaston ou)k oÄn poieiÍ ’ (256D12−256E2), ‘kata\ pa/nta’ is to be construed with the predicate position of both ‘oÃntoj’ and ‘ou)k oÄn’. Accordingly, in the English translation the phrase ‘with respect to all of them’ introduces a quantification on the predicate position of the following occurrences of ‘what is’ and ‘not being’: ‘With respect to all of them, the nature of the other, by rendering each one other than what is <them>, makes it not being <them>’ (256D12−256E2). The construction of ‘kata\ pa/nta’ with the predicate position of ‘oÃntoj’ and ‘ou)k oÄn’ at 256D12−256E2 is confirmed by two facts. First, shortly after our passage (58 words down the text) the ES uses the sentence ‘kata\ tosau=ta ou)k eÃstin’ (257A4−5), to be rendered by ‘with respect to as many is not’: here ‘kata\ tosau=ta’ is unequivocally construed with the predicate position of ‘ou)k eÃstin’.111 Second, the construction of ‘kata\ pa/nta’ with the predicate position of ‘oÃntoj’ and ‘ou)k oÄn’ at 256D12−256E2 avoids the inelegant repetition of the universal quantification (‘kata\ pa/nta’ at 256D12 and ‘eÀkaston’

110 My interpretation is close to that of Frede (1967), 85.

111 Cf. Owen (1971), 233−4; van Eck (2000), 73; van Eck (2002), 70.

117

at 256E1) which is required by other constructions (including the one postulated by the standard interpretation).112

(2) According to my interpretation, when ‘kata\ pa/nta’ is construed with the predicate position of ‘oÃntoj’ and ‘ou)k oÄn’, these last two expressions are to be taken to involve the identity use of ‘eiånai’ and ‘ou)k eiånai’. This is matched by the fact that the parallel sentence ‘kata\ tosau=ta ou)k eÃstin’ (257A4−5) involves the identity use of ‘ou)k eiånai’.

(3) My interpretation brings it about that at 256D12−256E2 Plato puts forward for the first time a principle which will play an important role in the dialogue’s sequel: for something not to be ϕ is for it to be other than everything that is ϕ (where ‘ϕ’ is a schematic letter to be replaced by a syntactically appropriate expression and the occurrence of ‘not to be’ and the occurrence of ‘is’ in penultimate position are occurrences of forms of ‘not to be’ and ‘to be’ in their identity use).113 So, my interpretation fosters continuity in the conceptual apparatus deployed by Plato in these pages of the Sophist.

‘Being is not’. Finally, at 257A1−12, the ES draws his conclusion about the not being of being:

ES Ou)kou=n kaiì to\ oÄn au)to\ tw½n aÃllwn eÀteron eiånai lekte/on.

THT. ¹Ana/gkh.

257A1

ES Kaiì to\ oÄn aÃr' h(miÍn, oÀsape/r e)sti ta\ aÃlla, kata\ tosau=ta ou)k eÃstin: e)keiÍna ga\r ou)k oÄn eÁn me\n au)to/ e)stin, a)pe/ranta de\ to\n a)riqmo\n taÅlla ou)k eÃstin auÅ.

THT. Sxedo\n ouÀtwj.

A5

ES Ou)kou=n dh\ kaiì tau=ta ou) dusxerante/on, e)pei¿per eÃxei koinwni¿an a)llh/loij h( tw½n genw½n fu/sij. Ei¹ de/ tij tau=ta

mh\ sugxwreiÍ, pei¿saj h(mw½n tou\j eÃmprosqen lo/gouj ouÀtw peiqe/tw ta\ meta\ tau=ta.

THT. Dikaio/tata eiãrhkaj.

A10

A12 ES Therefore being itself must also be said to be other than the others. THT. Necessarily.

257A

ES Hence, in our view, being, however many the others are, with respect to as many is not: for, not being them, it is itself one, but it is not in turn the others, which are indefinite in number.

THT. Probably so.

ES So we must not be afraid of this either,114 if really the nature of the kinds has reciprocal communion. And if someone does not agree to this, it is after having prevailed on our earlier accounts that he must prevail on what follows.

THT. Absolutely right.

This is a straightforward application of the result reached in the immediately preceding passage: since each kind is not indefinitely many kinds because it is other than them, it follows that the

112 Cf. van Eck (2000), 73; van Eck (2002), 66.

113 Later (cf. below, n. 176 and text thereto) we shall encounter a variation of this principle where the relevant occurrences of ‘not to be’ and ‘is’ are occurrences of forms of ‘not to be’ and ‘to be’ in their predicative use.

114 Cf. 256A10−11.

118

kind being also is not indefinitely many kinds because it is other than them. This yields the conclusion that in a sense ‘being is not’.

Negative predicative expressions involving forms of ‘to be’. At Sophist 257B1−257C4 the ES and Theaetetus have the following exchange:

ES ãIdwmen dh\ kaiì to/de. THT. To\ poiÍon;

257B1

ES ¸Opo/tan to\ mh\ oÄn le/gwmen, w¨j eÃoiken, ou)k e)nanti¿on ti le/gomen tou= oÃntoj a)ll' eÀteron mo/non.

THT. Pw½j; B5 ES Oiâon oÀtan eiãpwme/n ti mh\ me/ga, to/te ma=llo/n ti¿

soi faino/meqa to\ smikro\n hÄ to\ iãson dhlou=n t%½ r(h/mati;

THT. Kaiì pw½j; ES Ou)k aÃr', e)nanti¿on oÀtan a)po/fasij le/ghtai shmai¿-

Nein, sugxwrhso/meqa, tosou=ton de\ mo/non, oÀti tw½n aÃllwn

B10 ti mhnu/ei to\ mh\ kaiì to\ ouÄ protiqe/mena tw½n e)pio/ntwn 257C1 o)noma/twn, ma=llon de\ tw½n pragma/twn periì aÀtt' aÄn ke/htai ta\ e)pifqeggo/mena uÀsteron th=j a)pofa/sewj o)no/mata.

THT. Panta/pasi me\n ouÅn.

C4 ES Let us consider also this. THT. What?

257B

ES Whenever we say ‘not being’, it seems that we do not call something contrary to what is, but only other.

THT. How so?

ES For instance, when we call something ‘not large’ do we then appear to you to indicate the small by that expression any more115 than the equal?116

THT. How on earth?117

ES So, when the negative118 is said to mean a contrary, we shall not agree, but admit no more than this, that the ‘not’119 prefixed to the names that follow indicates one of the things other than them,120 or, rather,121 than whatever objects the names uttered after the negative are given to.

THT. By all means.

257C

115 ‘ti’ modifies ‘ma=llon’ (cf. LSJ s.v. ‘ma/la’ II 1).

116 Cf. Prm. 150D4−150E1; 161C7−161E1. There are two ways in which the attributes largeness, smallness, and equality may be considered as reciprocally coordinate. First, one may regard all three attributes as purely relational: largeness is largeness-with-respect-to, smallness is smallness-with-respect-to, and equality is equality-to. Secondly, one may treat all three attributes as non-relational, their relativity being eliminated by reference to some standard (which remains understood): largeness is largeness-with-respect-to-the-standard, smallness is smallness-with-respect-to-the-standard, and equality is equality-to-the-standard. On the second way of understanding it, equality, being equality-to-the-standard, coincides with middling size. In the Phaedo (102B3−102D2) largeness and smallness are presented as the attributes whose presence in Simmias contributes to explaining why he is larger than Socrates and smaller than Phaedo. This favors the first of the two accounts of how largeness, smallness, and equality may be considered as coordinate, i.e. the one according to which they are regarded as purely relational.

117 For ‘Kaiì pw½j;’ meaning ‘How on earth?’ cf. 238A10; 249A5; 250B7; 256D3; 264A3.

118 Pace Dixsaut (1991), 175, here at 257B9 and at 257C3 ‘a)po/fasij’ means ‘negative particle’ (cf. Campbell (1867), Sph. 158). I therefore translate it by ‘negative’. At 263E12 ‘a)po/fasij’ means ‘denial’.

119

This passage contains four remarks by the ES and Theaetetus’ reactions to them. I shall discuss these four remarks one by one.

The ES’s first remark. The ES’s first remark (‘Let us consider also this’, 257B1) hints that a new topic is broached.122 The general drift of the passage shows that the new topic has something to do with the word ‘not’, which is discussed in each of the following three remarks of the ES (cf. 257B3, 257B6, and 257C1).

This result may be plausibly sharpened. For the ES’s third remark is concerned with our calling something ‘not large’, i.e. with our predicating the negative expression ‘not large’ of something. This fact makes it plausible to assume that the new topic has something to do with the predication of negative expressions.

This result may be plausibly sharpened even further, however. Since in the preceding portion of text Plato has offered an account of negative sentences involving the identity use of ‘to be’, and since he has also distinguished the identity use of ‘to be’ from the predicative use of that verb, one expects him to say something about negative sentences involving the predicative use of ‘to be’: were he not to do this, his theory would be defective. These data generate the presumption that the new topic heralded by ES’s first remark is that of negative sentences involving the predicative use of ‘to be’, or, equivalently, sentences involving the predicative use of ‘not to be’. However, since several commentators deny that in 257B1−257C4 Plato is concerned with sentences involving the predicative use of ‘not to be’,123 this presumption must be checked on the basis of a careful textual analysis.

The ES’s second remark. The ES’s second remark begins with the words ‘Whenever we say “not being”, …’ (‘o(po/tan to\ mh\ oÄn le/gwmen, …’, 257B3). In Greek, a neuter article in front of a word or a phrase occasionally indicates that the word or phrase is being mentioned. Sometimes the neuter article does not match the word or phrase it precedes (cf. 252C2−4 and 257C1): on

119 The Greek language has two negative particles, ‘mh/’ and ‘ouÃ’, and Plato mentions them both. Since both ‘mh/’ and ‘ouÃ’ are normally rendered in English by ‘not’, in my translation I have not tried to have two negative particles of English.

120 The genitive ‘tw½n e)pio/ntwn o)noma/twn’ (257C1−2) is to be construed in common with ‘protiqe/mena’ (257C1) and ‘aÃllwn’ (257B10): cf. Jowett (1892), IV 391−2; Fowler (1921), 417; Vitali (1992), 127; Cordero (1993), 180. Some commentators construe ‘tw½n e)pio/ntwn o)noma/twn’ only with ‘protiqe/mena’: cf. Ficino (1484), 361; Cornarius (1561), 156; Diès (1925), 371−2; Zadro (1961), 143; Zadro (1971), 244. Others construe it only with ‘aÃllwn’: cf. Stallbaum (1840), 198−9; Wagner (1856), 139−41; Campbell (1867), Sph. 158; Cornford (1935), 290; Benardete (1984), 55. Here ‘aÃllwn’ is a stylistic variant of ‘e(te/rwn’: cf. 256C4−5 (which provides also an example of ‘aÃlloj’ construed with the genitive). Van Eck (1997) has a different interpretation of the grammatical structure of 257B9−257C4, but, so far as I can see, his exegesis does not yield an importantly different sense of the passage.

121 For ‘ma=llon de/’ meaning ‘or rather’ cf. Tht. 170A7.

122 Cf. Szaif (1998), 434−5.

123 Cf. Malcolm (1967), 145; Ray (1984), 69−72; Roberts (1986), 238−9; van Eck (1995), 25−35.

120

such occasions, it is obvious that the article is indicating that the word or phrase is being mentioned. Sometimes however the neuter article does match the word or phrase it precedes: on some such occasions it is hard to tell whether the article is indicating that the word or phrase is being mentioned (an example of a neuter article that matches the following word or phrase and does indicate that it is being mentioned occurs at 237C2). I take it that the article ‘to/’ in front of ‘mh\ oÃn’ at 257B3 is providing an indication of this sort, i.e. is marking that the phrase ‘mh\ oÃn’ is being mentioned (hence the quotation marks around ‘not being’ in my English translation). The ES is addressing the issue of what we say when we use the expression ‘not being’, which probably stands in for any form of ‘not to be’.

A different interpretation of the initial words of the ES’s second remark is possible: the ES might be (not mentioning but) using the expression ‘not being’. In this case, he would be addressing the question of what we say when we speak of what is not. Consider, however, what comes immediately after the ES’s second remark. We find Theaetetus asking the question ‘Pw½j;’ (257B5), which could be translated either by ‘What do you mean?’ or by ‘How so?’.124 Theaetetus is therefore asking either for a clarification or for a justification of the ES’s second remark. Since the immediately following third remark of the ES is either an answer or the beginning of an answer to this question, it is closely linked to his second remark. Since in his third remark the ES makes it clear that he is dealing with the negative linguistic expression ‘not large’ (cf. ‘by that expression’, ‘t%½ r(h/mati’, at 257B7), it may be reasonably inferred that in his second remark he is dealing with the negative linguistic expression ‘not being’.125

On my translation, after the initial words ‘Whenever we say “not being”, …’ (257B3), the ES’s second remark continues with the words ‘… it seems that we do not call something contrary to what is, but only other’ (257B3−4). On this translation, ‘something’ is the direct object of ‘we do not call’ and ‘contrary to what is’ indicates what we do not call this something as. Translators normally adopt a different solution: they take ‘something’ to form a single phrase together with ‘contrary to what is’, in which case the continuation of the ES’s second remark should be rendered by ‘… it seems that we are not speaking of something contrary to what is, but only of something other’ (257B3−4).126 Since the ES’s third remark is closely linked to his second remark, and since the ES’s third remark contains the sentence ‘We call something “not large”’ (257B6), where ‘something’ is the direct object of ‘we call’ and ‘“not large”’ refers to the expression we use

124 Cf. below, paragraph to n. 128.

125 Cf. van Eck (1995), 30.

126 Cf. Ficino (1484), 361; Cornarius (1561), 156; Jowett (1892), IV 391; Apelt (1914), 106; Fowler (1921), 415; Diès (1925), 371; Fraccaroli (1934), 192; Cornford (1935), 290; Arangio-Ruiz (1951), 196; Taylor (1961), 164; Warrington (1961), 209; Matthews (1972), 250; Cavini (1982), 133; Benardete (1984), 53; Meinhardt (1990), 153; Mazzarelli (1991), 300; Vitali (1992), 127; Cordero (1993), 179; N.P. White (1993), 51; Brann, Kalkavage, and Salem (1996), 68; Duerlinger (2005), 127; Ambuel (2007), 232.

121

to express what we call this something as, it may be reasonably inferred that in his second remark also the ES is using a similar construction. This speaks for the translation I offered above.127

A first plausible conclusion may now be drawn regarding the interpretation of the ES’s second remark: the new topic that is being broached is that of what it is that we are saying when we use the expression ‘not being’ with regard to something. The point made by the ES in his second remark is that when we use ‘not being’ with regard to a thing, we are not saying about this thing that it is contrary to what is, but that it is other than what is. This is only a first approach to the interpretation of the ES’s second remark. A full interpretation must wait after the discussion of the ES’s third and fourth remarks.

The ES’s third remark. To understand the ES’s third remark, two problems of literal interpretation must be solved.

(1) The first problem concerns the question that immediately precedes and introduces the ES’s third remark: the question ‘Pw½j;’ (257B5). What is Theaetetus asking for? Two solutions are possible. (1.1) By asking the question ‘Pw½j;’ Theaetetus might be calling for a clarification (i.e. an explanation of the meaning) of the ES’s second remark. In this case ‘Pw½j;’ would have to be translated by ‘What do you mean?’.128 Such a use of the isolated ‘Pw½j;’ has at least one parallel in the Sophist.129 (1.2) By asking the question ‘Pw½j;’ Theaetetus might be calling for a justification of the ES’s second remark. In this case ‘Pw½j;’ would have to be translated by ‘How so?’, ‘How come?’, or even ‘Why?’.130 Such a use of the isolated ‘Pw½j;’ has many parallels within the Sophist.131

(2) The second problem of literal interpretation raised by the ES’s third remark has to do with its first word: ‘oiâon’ (257B6). Two interpretations of it are possible. (2.1) The word ‘oiâon’ might introduce an analogy. In this case it would have to be translated by ‘as’ or ‘in the same way as’.132 (2.2) The word ‘oiâon’ might introduce an example: in this case it is to be translated by ‘for instance’ or ‘for example’.133 If ‘oiâon’ introduces an example, one may ask what it is an example

127 Cf. Cri. 48A6; Sph. 256A11−12; 257D9−10; LSJ s.v. ‘le/gw’ (B) III 3; Crivelli (1990), 65. The parallel with 257B6 is more compelling than the one with 258E8, which favours taking ‘something’ to form a single phrase together with ‘contrary to what is’ at 257B4.

128 Cf. Jowett (1892), IV 391; Fowler (1921), 415.

129 Cf. 261E7.

130 Cf. Taylor (1961), 164; White (1993), 51; Brann, Kalkavage, and Salem (1996), 68; Ambuel (2007), 233.

131 Cf. 219A9; 239E4; 244C7; 245A7; 249B11; 252C1; 254A3; 255A6; 258C12; 262B4; 263D9; LSJ s.v. ‘pw½j’ I 1 b.

132 Cf. LSJ s.v. ‘oiâoj’ V 2 a; Sph. 228A1; 253A1. Cornford (1935), 290 translates the occurrence of ‘oiâon’ at 257B6 by ‘in the same way that’.

133 Cf. LSJ s.v. ‘oiâoj’ V 2 b; Sph. 218E3. For the translation of the occurrence of ‘oiâon’ at 257B6 by ‘for example’ cf. Fowler (1921), 417; Diès (1925), 371; Taylor (1961), 164; Brann, Kalkavage, and Salem (1996), 68; Ambuel (2007), 233.

122

of. Two solutions are possible. (2.2.1) The example introduced by ‘oiâon’ might be backwards-looking: in this case its purpose would be to clarify or justify the ES’s second remark.134 (2.2.2) The example introduced by ‘oiâon’ might be forward-looking: in this case its purpose would be to justify what follows, in particular to provide support for at least part of the general claim made by the ES’s fourth remark: after all, the ES’s fourth remark contains the word ‘so’ (‘aÃr' ’, 257B9), which indicates that a conclusion is being drawn.

I begin by addressing problem (2). Solution (2.2.1) faces a difficulty. For suppose that, as solution (2.2.1) assumes, ‘oiâon’ were to introduce an example that is backwards-looking, i.e. an example whose intent is to clarify or justify the ES’s second remark, according to which ‘whenever we say “not being”, it seems that we do not call something contrary to what is, but only other’ (257B3−4). Then what we say about a thing when we apply ‘not large’ to it would be introduced as an example of what we say about a thing when we use ‘not being’ with regard to it. But the expression ‘not large’ contains no form of the verb ‘to be’, so that what we say about a thing when we call it ‘not large’ can hardly count as an example of what we say about a thing when we use ‘not being’ with regard to it. This difficulty might be answered by pointing out that a thing can be called ‘not large’ only in the context of a sentence that contains some form of ‘to be’, i.e. in the context of a sentence involving a phrase like ‘is not large’. So—the retort goes—what we say about a thing when we apply ‘not large’ to it is after all an example of what we say about a thing when we use ‘not being’ with regard to it. This answer however faces a further difficulty. For the English phrase ‘is not large’ can correspond to two different Greek phrases. On the one hand, ‘is not large’ can correspond to ‘eÃstin ou) me/ga’ or ‘eÃsti mh\ me/ga’, where the word corresponding to ‘is’ (‘eÃstin’ or ‘eÃsti’) is immediately followed by the one corresponding to ‘not’ (‘ou)’ or ‘mh/’), which in turn is immediately followed by the one corresponding to ‘large’ (‘me/ga’). On the other hand, ‘is not large’ can correspond to ‘ou)k eÃsti me/ga’ or ‘mh\ eÃsti me/ga’, where the word corresponding to ‘not’ (‘ou)k’ or ‘mh/’) is immediately followed by the one corresponding to ‘is’ (‘eÃsti’), which in turn is immediately followed by the one corresponding to ‘large’ (‘me/ga).135 If the Greek expression corresponding to ‘not large’ (‘ou) me/ga’ or ‘mh\ me/ga’) is applied to a thing, the sentence in the context of which this happens must be one involving a phrase like ‘eÃstin ou) me/ga’ or ‘eÃsti mh\ me/ga’, not one involving a phrase like ‘ou)k eÃsti me/ga’ or ‘mh\ eÃsti me/ga’: but it should be sentences of the second type, not of the first, that the ES should get us to consider if he was providing an example of what we say about a thing when we use ‘not being’ with regard to it. Moreover, it can hardly be the case that the ES is blind to the difference between phrases of the form ‘eÃstin ou) me/ga’ or ‘eÃsti mh\ me/ga’ and phrases of the form ‘ou)k eÃsti me/ga’ or ‘mh\ eÃsti me/ga’: for the argument at 255E8−257A12, immediately before the portion of text we are dealing with, turned on a gradual and careful transposition of the negative particle from the position between a form of ‘eiånai’ and a complement to the

134 Cf. Kostman (1973), 203; Marcos de Pinotti (1994), 155−6.

135 Cf. above, text to n. 89.

123

position before ‘eiånai’ followed by the complement.136 In view of this difficulty, solution (2.2.1) ought to be discarded. This leaves solutions (2.1) and (2.2.2). It is hard to choose between them. I have some uneasiness with solution (2.1), according to which ‘oiâon’ introduces an analogy: introducing an analogy by means of a question (albeit a rhetorical one) is awkward. I therefore opt for solution (2.2.2): ‘oiâon’ introduces an example that is forward-looking, i.e. aimed at providing support for at least part of the general claim made by the ES’s fourth remark.

Let us now go back to problem (1). Given that the correct solution of problem (2) is (2.2.2), the ES’s third remark is the beginning of an argument. Since arguments are more naturally taken to provide justifications for claims than to provide clarifications for obscure formulations, the most natural solution of problem (1) is (1.2), according to which when Theaetetus asks the question ‘Pw½j;’ (257B5) what he is calling for is a justification of the ES’s second remark.

Let me now summarize the results of my reflections on the ES’s third remark. With his second remark (257B3−4) the ES claims that when we use ‘not being’ with regard to a thing we are not calling this thing contrary to what is but only other than it. By uttering the expression ‘How so?’ (257B5) Theaetetus calls for a justification of this claim. At least part of the reason why he requires such a justification is that earlier in the dialogue (at 240B5 and 240D6) he seemed to assume that what is not is the contrary of what is. The justification is reached by looking at the semantic behavior of ‘not’ in expressions which do not contain a form of ‘to be’ (‘not large’ is one such expression). These expressions are employed as examples (hence the use of ‘oiâon’ at 257B6) on the basis of which a general conclusion is drawn about the semantic behavior of ‘not’. This general conclusion may then be applied back to the original problem of justifying the claim that when we use ‘not being’ with regard to a thing we are not calling this thing contrary to what is but only other than it. The argument in this section therefore turns out to be very similar to the usual ‘Socratic inductive arguments’: to solve a difficult problem, consider first an easy similar case, then infer a general rule covering all cases, and finally apply this rule to the difficult original case. Arguments of this sort occur in several dialogues.137

The ES’s second remark (again). In his third remark, the ES addresses the question of what we say ‘when we call something “not large”’ (257B6). The question seems to be that of what it is that we are saying about a thing when we describe it by using the phrase ‘not large’, or, equivalently, what it is that we are saying about a thing when we predicate ‘not large’ of it. As we have seen in the last subsection, with his third remark the ES begins an analogical argument whose purpose is to justify the claim he made with his second remark. Moreover, as I pointed out at the end of the penultimate subsection, the point made by the ES in his second remark is that when we use ‘not being’ with regard to a thing, we are not saying about this thing that it is contrary to what is, but that it is other than what is. In view of these results, we may plausibly sharpen our account of the point made the ES in his second remark: it is that when we predicate ‘not being’ of a thing, we are not saying about this thing that it is contrary to what is, but that it is other than what is. This

136 Cf. above, text to n. 88.

137 Cf. Euthphr. 10B1−10C12; Grg. 460B1−7; 476B3−476E3; Lewis (1976), 109; Gómez-Lobo (1977), 36.

124

is a step forward in the interpretation of the ES’s second remark, but is not yet the full exegesis. This can only be given after the discussion of the ES’s fourth remark.

The ES’s fourth remark. The ES’s fourth remark (257B9−257C3) consists of a negative and a positive component. The negative component is a general point about what the negative particle ‘not’ does not mean: it does not mean contrariety. I shall return to the precise interpretation of this negative component after addressing the affirmative one. In the positive component, the ES introduces a general rule concerning the meaning of ‘not’. These lines have been variously interpreted. There are four main exegeses.138

(1) According to the Oxford interpretation, Plato’s thesis is that if a negative predicative expression consisting of a negative particle followed by a predicative expression is predicated of an object then what is said about this object is that every kind that holds of it is other than the kind signified by the predicative expression that follows the negative particle. For instance, if the negative predicative expression ‘not flying’ is predicated of Theaetetus then what is said about him is that every kind that holds of him is other than flying, the kind signified by the negated predicative expression ‘flying’.139

(2) According to the incompatibility interpretation, Plato’s thesis is that if a negative predicative expression consisting of a negative particle followed by a predicative expression is predicated of an object then what is said about this object is that some kind that holds of it is incompatible with the kind signified by the predicative expression that follows the negative particle (where incompatibility is a relation which obtains between kinds that cannot both hold together of any object). For instance, if the negative predicative expression ‘not flying’ is predicated of Theaetetus then what is said about him is that some kind that holds of him is incompatible with flying, the kind signified by the negated predicative expression ‘flying’.140

(3) According to the quasi-incompatibility interpretation, Plato’s thesis is that if a negative predicative expression consisting of a negative particle followed by a predicative expression is predicated of an object then what is said about this object is that some kind that holds of it is

138 The exegeses of Plato’s account of negation in terms of otherness are recorded by Pelletier (1983), 38−54; Crivelli (1990), 41−58; Pelletier (1990), 45−93; Dixsaut (1991), 176−7.

139 Cf. Ackrill (1964), 613; von Weizsäcker (1973), 235; Charlton (1995), 114−16. According to Wiggins (1971), 288−94, the account of statements of negative predication favoured by the

Oxford interpretation is ‘the most natural amendment’ of what Plato actually says and perhaps ‘is really what Plato meant all along’.

Bordt (1991), 527 favors a ‘linguistic variant’ of the Oxford interpretation: if a negative predicative expression consisting of a negative particle followed by a predicative expression is predicated of an object then what is said about this object is that every predicative expression that is true of it is other than the predicative expression that follows the negative particle.

Van Eck (1995), 43−4 attributes to Plato an account of statements of negative predication close to (perhaps identical with) that favoured by the Oxford interpretation, but he emphatically denies that this account is to be found in the ES’s words at 257B9−257C3. According to van Eck, Plato intends the account to be inferred from the analysis of false statement at 262E11−263D5 and from the obvious point that a negative statement is true just in case the corresponding affirmative statement is false.

140 Cf. below, nn. 149−152.

125

other than, but in the same incompatibility range as, the kind signified by the predicative expression that follows the negative particle (where an incompatibility range is an exhaustive set of incompatible kinds). For instance, if the negative predicative expression ‘not flying’ is predicated of Theaetetus then what is said about him is that some kind that holds of him is other than, but in the same incompatibility range as, flying, the kind signified by the negated predicative expression ‘flying’.141

(4) According to the extensional interpretation, Plato’s thesis is that if a negative predicative expression consisting of a negative particle followed by a predicative expression is predicated of an object then what is said about this object is that it is other than everything the predicative expression that follows the negative particle is true of. For instance, if the negative predicative expression ‘not flying’ is predicated of Theaetetus then what is said about him is that he is other than everything the negated predicative expression ‘flying’ is true of.142

The Oxford interpretation. The main reason why the Oxford interpretation is favored by some commentators is that it fits well with the account of falsehood offered by Plato later in the dialogue (at 262E11−263D5). However, the Oxford interpretation cannot be plausibly read into the ES’s fourth remark at 257B9−257C3. For, at the point of the text where the Oxford interpretation requires a universal quantification on the kinds that hold of an object, there is only a particular quantifier: ‘The “not” prefixed to the names that follow indicates one of the things other than […] whatever objects the names uttered after the negative are given to’ (257B10−257C3).143

Four attempts might be made to reconcile 257B10−257C3 with the Oxford interpretation: (1) The phrase ‘tw½n aÃllwn ti’ (257B10−257C1) is to be rendered (not by ‘one of the things

other’, but) by ‘something other’. The point made by the ES’s fourth remark at 257B9−257C3 is: the negative particle ‘not’ indicates that the object x which is being spoken about is (i.e. is identical with) something other than the kind P signified by the words that follow ‘not’ and the otherness of x with respect to P amounts to each of the kinds that hold of x being other than P. The crucial assumption on which this solution relies is implausible: why should the otherness of x with respect to P amount to each of the kinds that hold of x being other than P?

(2) The phrase ‘tw½n aÃllwn ti’ (257B10−257C1) is to be rendered by ‘something other’ (as above). The point made by the ES’s fourth remark at 257B9−257C3 is: the negative particle ‘not’ indicates that the object x which is being spoken about is (i.e. partakes of ) something other than the kind P signified by the words that follow ‘not’ and what x partakes of and is other than P is the collection of all the kinds that hold of x. Such a reading is evidently forced.

141 Cf. Ryle (1967), 329; Philip (1968), 317−9; Pelletier (1990), 66, 77.

142 Cf. Taylor (1926), 389; Dürr (1945), 187−8; Taylor (1961), 165; Toms (1962), 87; Frede (1967), 78−80; Bostock (1984), 115; Mignucci (1989), 273−4; Crivelli (1990), 44−8, 59−62; Denyer (1991), 137−9; Crivelli (1993), 72−3.

143 Cf. Bostock (1984), 113; Ray (1984), 70; Crivelli (1990), 52; van Eck (1995), 26.

126

(3) The phrase ‘tw½n aÃllwn ti’ (257B10−257C1) is to be rendered by ‘something other’ (as above). The point made by the ES’s fourth remark at 257B9−257C3 is: the negative particle ‘not’ indicates that the object which is being spoken about is (i.e. partakes of ) <only> something other than the kind signified by the words that follow ‘not’. The problem is that ‘only’ is not in the text and cannot be plausibly supplied.

(4) The phrases ‘tw½n e)pio/ntwn o)noma/twn’ (257C1−2) and ‘tw½n pragma/twn’ (257C2) are to be construed exclusively with ‘protiqe/mena’ (257C1).144 The point made by the ES’s fourth remark at 257B9−257C3 is: the negative particle ‘not’ prefixed to the following words, or rather to the objects (sc. kinds) signified by those words, indicates that those objects are among the things other <than all the kinds that hold of the object which is being spoken about>. The problem with this solution is that the thesis that characterizes the Oxford interpretation is supposed to be supplied almost completely by the reader.

The incompatibility interpretation. The incompatibility interpretation has always attracted commentators, for at least two reasons. First, the incompatibility interpretation has a philosophical charm: it seems to answer the problem of the meaning of negation by invoking only positive kinds. The meaning of negation is a problem. It is perhaps relatively easy to explain the meaning of an affirmative expression by saying that it means a certain kind, e.g. to explain the meaning of the affirmative expression ‘white’ by saying that ‘white’ means the kind white. (Answers of this sort may be criticized from several points of view, but let us provisionally waive these criticisms.) It is much more difficult to explain the meaning of a negative expression by saying that it means a (negative) kind, e.g. to explain the meaning of the negative expression ‘not white’ by saying that it means the (negative) kind non-white. For one finds oneself presupposing that there are negative kinds—a very dubious presupposition. Are there negative kinds like non-white, non-square, non-man, etc.? What common characteristics do my right thumb, the number two, St Paul’s Cathedral, and generosity have? Still, they should all fall under non-white, if any such kind exists. Moreover, we think we can explain what white by pointing to an example of a white object. But we cannot explain what non-white is by pointing to an example of an object that is not white: any non-white object is first and primarily an example of a different color-form (if it is red, it is more readily an example of red than of non-white). The incompatibility interpretation seems to answer the problem of the meaning of negation by saying e.g. that the negative predicative expression ‘not white’ means the belonging of at least one of the positive kinds that are incompatible with white, i.e. the belonging of at least one of the positive kinds black, gray, red, etc.

The solution of the problem of the meaning of negation based on the notion of incompatibility among kinds is open to at least three objections. (1) The notion of incompatibility seems to presuppose negation: were one to define the relation of incompatibility, one of the things one would say is that two kinds are incompatible only if they cannot hold

144 Cf. above, n. 120.

127

together of the same object—and this definition involves negation.145 One might however answer this objection by taking the relation of incompatibility as an undefined primitive. (2) Another objection might be leveled at the definition of incompatibility as impossibility of joint instantiation. The objection is that the definition fails to specify the nature of the impossibility it invokes: is it ‘logical’ impossibility, ‘physical’ impossibility, ‘analytic’ impossibility, or what? It remains unclear whether this objection may be answered by appealing simply to the view that the relation of incompatibility is an undefined primitive: even if one denies that incompatibility is defined as impossibility of joint instantiation, one may hardly avoid granting that incompatibility is logically equivalent to the impossibility of joint instantiation, so that the question about the nature of the impossibility involved returns. (3) The account of the meaning of negation based on the notion of incompatibility among kinds does not seem to be universally applicable. For instance, although the negative predicative expression ‘not white’ appears to be true of courage, it is hard to see which kind incompatible with white could be holding of courage.146 An advocate of the analysis of negation based on incompatibility could reply either that the attribution of ‘not white’ to courage is meaningless, or that the kind incompatible with white that holds of courage is the kind colorless (in general, there are more kinds than one would expect). Neither reply is promising. The first seems to make a move that is both ad hoc and, arguably, false: if in a conversation with you I use the sentence ‘Courage is not yellow’, you may at first be a little surprised, but you understand perfectly well what I am saying. The inclination to describe my use of the sentence ‘Courage is not yellow’ as meaningless might be due to a confusion about the adjective ‘meaningless’, which can indicate both lack of purpose and lack of sense. The second reply runs the risk of reintroducing negative kinds in order to supply enough incompatible kinds.

A second reason why the incompatibility interpretation attracts commentators is that it seems to be confirmed by a passage of the text, in the ES’s third remark (at 257B6−7). Since smallness and equality may be regarded as the kinds that are incompatible with largeness, Theaetetus’ negative answer to the ES’s third remark, whereby it is denied that ‘when we call something “not large” … we appear … to indicate the small by that expression any more than the equal’ (257B6−7), might be taken to be saying that when we call something ‘not large’ we mean that it is either small or equal. Such a claim is one that Plato might be expected to make if he were proposing the theory attributed to him by the incompatibility interpretation.

An analysis of negation grounded in incompatibility was proposed by Hegel. Hegel’s analysis was taken over by Bernard Bosanquet, who was quoted by J. Burnet with regard to the Sophist, and Burnet’s interpretation was cited by A. E. Taylor.147 An interpretative tradition Hegel-Bosanquet-Burnet-Taylor may therefore be plausibly identified.

Three variants have been suggested for the incompatibility interpretation, according to the kind of incompatibility involved. Three kinds of incompatibility can be distinguished:

145 Cf. Wiggins (1971), 291; Ray (1984), 70.

146 Cf. Ray (1984), 71.

147 Cf. Bosanquet (1888/1911), I 289; Burnet (1914), 278; Taylor (1961), 63−4.

128

(1) A kind F is incompatible in an absolutely complementary way with a kind K just in case (i) F is other than K and (ii) for every individual x, F and K cannot both hold together of x and (iii) for every individual y, necessarily either F or K holds of y at any given time.148 For instance, white and non-white are incompatible in an absolutely complementary way. Commentators who favor this kind of incompatibility are not worried by the problems of negative kinds I outlined earlier. According to this interpretation, Plato’s view is that if a negative predicative expression consisting of the negative particle ‘not’ followed by a predicative expression is predicated of an object then what is said about this object is that some kind that holds of it is incompatible in an absolutely complementary way with the kind signified by the predicative expression that follows the negative particle ‘not’.149

(2) A kind F is incompatible in a categorically complementary way with a kind K just in case (i) F is other than K and (ii) for every individual x, F and K cannot both hold together of x and (iii) for every individual y belonging to the appropriate category, necessarily either F or K holds of y at any given time.150 For instance, the kinds white and of a color other than white are incompatible in a categorically complementary way: they cannot both hold together of any individual, but at any time at least one of them holds of any individual of the appropriate category, i.e. of any colored individual. They are not incompatible in an absolutely complementary way because the number two (which does not belong to the appropriate category) enjoys neither the kind white nor the kind of a color other than white at any time. According to this interpretation, Plato’s view is that if a negative predicative expression consisting of the negative particle ‘not’ followed by a predicative expression is predicated of an object then what is said about this object is that some kind that holds of it is incompatible in a categorically complementary way with the kind signified by the predicative expression that follows the negative particle ‘not’.151

(3) A kind F is non-exhaustively incompatible with a kind K just in case (i) F is other than K and (ii) for every individual x, F and K cannot both hold together of x. For instance, the kinds white and red are incompatible in a non-exhaustive way. They are not incompatible either in an absolutely complementary way or in a categorically complementary way because neither white nor red holds of a yellow crayon (an individual which does belong to the appropriate category). According to this interpretation, Plato’s view is that if a negative predicative expression consisting of the negative particle ‘not’ followed by a predicative expression is predicated of an object then

148 If there are individuals then condition (i) is redundant because it is entailed by (ii) and (iii).

149 Cf. Moravcsik (1962), 68−77; Szaif (1998), 439−41.

150 If there are individuals belonging to the appropriate category then condition (i) is redundant because it is entailed by (ii) and (iii).

151 Cf. Kostman (1973), 201−6; Sayre (1976), 584−6; Sayre (1983), 230−4.

129

what is said about this object is that some kind that holds of it is non-exhaustively incompatible with the kind signified by the predicative expression that follows the negative particle ‘not’.152

The incompatibility interpretation faces a serious difficulty: it implausibly presupposes that at some points in the Sophist the Greek word ‘eÀteron’ expresses (not otherness, as it does elsewhere in the dialogue, but) incompatibility.153

An advocate of the incompatibility interpretation might try to defend it by appealing to the ‘generalizing’ reading of sentences containing abstract nouns:154 when the ES says that ‘the “not” prefixed to the names that follow indicates one of the things other than […] whatever objects the names uttered after the negative are given to’ (257B10−257C3), he means that anyone making a statement of negative predication states that the item referred to has some kind other than the kind signified by the expression that follows the negative particle ‘not’, i.e. has some kind whose every instance is other than every instance of the kind signified by the expression that follows the negative particle ‘not’.155 Such a defense, however, amounts to watering down the incompatibility interpretation in such a way as to make it coincide with the extensional interpretation. So, despite its attractiveness, the incompatibility interpretation must be rejected.

The quasi-incompatibility interpretation. To avoid the difficulties of the incompatibility interpretation without losing its attractiveness, some commentators propose the quasi-incompatibility interpretation, which seems logically equivalent to the incompatibility interpretation but is supposed to fit the text better: for, according to the quasi-incompatibility interpretation, Plato analyzes negation by appealing to kinds other than, but in the same incompatibility range as, the kind signified by the words following the ‘not’ (so the Greek word ‘eÀteron’ continues to express otherness and does not shift to expressing incompatibility).

This solution however incurs a cost: the Sophist does not explicitly mention incompatibility ranges, which play a pivotal role in the quasi-incompatibility interpretation. Therefore the quasi-incompatibility interpretation also fails to match the text.156 An advocate of the quasi-incompatibility interpretation might say that although the Sophist does not explicitly mention incompatibility ranges, it displays them: it does so, for instance, in the ES’s third remark at 257B6−7, where the incompatibility range consisting of largeness, equality, and smallness is on show. This reply is not convincing: if the Sophist is to offer a good account of negative predication along the lines envisaged by the quasi-incompatibility interpretation, it can hardly get away with merely displaying incompatibility ranges, which have so central a role in the theory it is supposed to be putting forward.

152 Cf. Hamlyn (1955), 292 (not quite a formulation of the incompatibility interpretation); Gosling (1973), 226−7; Pelletier (1975), 143−6.

153 Cf. Owen (1971), 232, 238; Wiggins (1971), 291; Lewis (1976), 101; Szaif (1998), 490−1.

154 Cf. above, n. 30 and text thereto.

155 Cf. Bostock (1984), 117.

156 Cf. van Eck (1995), 26−7.

130

The account of negation attributed to Plato by the quasi-incompatibility interpretation also faces one of the difficulties confronted by the incompatibility interpretation, but with less of an opportunity of a solution of it. Earlier157 I pointed out that although the negative predicative expression ‘not white’ appears to be true of courage, it is hard to identify a kind that both holds of courage and is incompatible with white. The best solution to this difficulty available to an advocate of the account of negation based on incompatibility was to claim that the kind that holds of courage and is incompatible with white is the kind colorless. This solution is no longer affordable in the case of the account of negation attributed to Plato by the quasi-incompatibility interpretation. For the kind colorless is not a member of the incompatibility range of which white is a member: this incompatibility range comprises only the kinds black, gray, yellow, white, etc., i.e. only the color-forms. In general, the class whose elements are the other members of the incompatibility range of which a kind F is a member is a proper subclass of the class whose elements are the kinds incompatible with F (because all elements of the former class are kinds on the same ‘level’ as F, while the latter class has among its elements kinds on a different level with respect to F ). So, the account of negation attributed to Plato by the quasi-incompatibility interpretation is too narrow.158

The extensional interpretation. The extensional interpretation is close to Plato’s words in the passage we are looking at: for Plato’s sentence ‘The “not” prefixed to the names that follow indicates one of the things other than […] whatever objects the names uttered after the negative are given to’ (257B10−257C3) can be plausibly paraphrased by the sentence ‘If a negative predicative expression consisting of the negative particle “not” followed by a predicative expression (consisting of one or more words) is predicated of an object then what is said about this object is that it is one of the objects other than all the objects denoted by the predicative expression that follows the negative particle “not”’. (I use ‘to denote’ and ‘denotation’ for the relation that obtains between a predicative expression P and an object x just in case P is true of x.)159

An objection might be raised against the extensional interpretation. When ‘to be given’ (‘keiÍsqai’) is used with regard to names and objects, it often expresses the original imposition of names to objects whereby the names’ signification is established. This suggest that the ‘objects the names uttered after the negative are given to’ (257C2−3) are not the objects denoted by the names uttered after the negative particle, but the kind or kinds signified by those names. Since the extensional interpretation requires that the ‘objects the names uttered after the negative are given to’ (257C2−3) should be the objects denoted by those names, the extensional interpretation ought to be abandoned.160

157 Cf. above, n. 146 and text thereto.

158 Cf. Keyt (1994), 118.

159 Cf. Quine (1950), 80.

160 Cf. Ray (1984), 71; Dixsaut (1991), 191; Charlton (1995), 115.

131

This objection is answered by highlighting two linguistic features of our passage which suggest that the ‘objects the names uttered after the negative are given to’ (257C2−3) are indeed the objects denoted by those names—as befits the extensional interpretation. The first noteworthy linguistic feature of our passage is its expressing the relation of names to objects by means of the preposition ‘peri¿’ construed with the accusative of the expression referring to the objects (257C2). When ‘peri¿’ is construed with the accusative in formulations of this sort, it may be rendered with ‘about’. This suggests that the semantic relation of names to objects involved in our passage is that of denotation: a predicative expression denotes what it is ‘about’. This chimes with the fact that in the Sophist the preposition ‘peri¿’ construed with the accusative is also used for the relation of a kind to the items it holds of,161 a relation that constitutes the ontological counterpart of denotation.

The second linguistic feature of our passage that deserves highlighting is its use of the indefinite relative pronoun ‘aÀtta’ followed by ‘aÃn’ construed with the subjunctive (257C2), a formulation that expresses indeterminacy162 and generality163 (in my English translation I tried to bring this out by means of the locution ‘whatever objects …’). This suggests that in our passage the ES is speaking of all of the possibly many objects to which the names uttered after a negative particle bear a certain semantic relation. And this, in turn, fits well with the semantic relation in question being that of denotation.

The objector could retort that the ES introduces not only many objects, but also many predicative expressions (consider the plurals ‘o)noma/twn’, 257C2, and ‘o)no/mata’, 257C3). Specifically, the objector could retort that the ES means that negative predicative expressions (‘not beautiful’, ‘not just’, etc.) consist of the negative particle ‘not’ followed by predicative expressions (‘beautiful’, ‘just’, etc.) and signify otherness with respect to the objects (beauty, justice, etc.) to which the predicative expressions following the negative particle ‘not’ (‘beautiful’, ‘just’, etc.) bear a certain semantic relation, each one of these predicative expressions (‘beautiful’, ‘just’, etc.) bearing this semantic relation to exactly one of these objects (‘beautiful’ to beauty, ‘just’ to justice, etc.): in which case the semantic relation in question would surely not be that of denotation (‘beautiful’ does not denote beauty but beautiful things, ‘just’ does not denote justice but just things, etc.). Finally, the objector could note that the indeterminacy expressed by the indefinite relative pronoun ‘aÀtta’ followed by ‘aÃn’ construed with the subjunctive (257C2) is compatible with the presence of a definite description, which in the case at hand would be ‘the objects to which the names uttered after the negative are given’: ‘The tallest philosopher, whoever that is, is over seven feet tall’ is a perfectly sensible sentence. (The fact that ‘the objects to which the names uttered after the negative are given’ is a plural definite description while ‘the tallest philosopher’ is a singular definite description makes no difference with respect to the point at hand.)

161 Cf. above, text to n. 98.

162 Cf. Kühner and Gerth (1892−1904), II.II 424−6 (cf. Cri. 50E6; Prt. 348B8; Grg. 452E4; 525D2).

163 Cf. Lg. V 736E6.

132

This retort is answered by pointing out that it sits uneasily with the singular ‘one of the things other’ (‘tw½n aÃllwn ti’, 257B10−257C1) in the phrase ‘one of the things other than them [sc. the following names], or, rather, than whatever objects the names uttered after the negative are given to’ (257B10−257C3). For, were the ES making the point which the retort takes him to be making, he should use the plural ‘things other’ (‘aÃlla’) (or, perhaps, ‘things from among those that are other’, ‘tw½n aÃllwn aÃtta’), in a phrase like ‘things other than them [sc. the following names], or, rather, than whatever objects the names uttered after the negative are given to’: different ‘things other’ would be surely involved in the case of different negative predicative expressions (‘not beautiful’, ‘not just’, etc.).

In my translation of 257B1−257C4, I rendered the form of ‘keiÍsqai’ occurring at 257C2 with a form of ‘to be given’. This rendering is warranted by the fact that ‘keiÍsqai’ is used as a passive for ‘tiqe/nai’ (‘to posit’, ‘to lay down’),164 a fact reflected by Plato’s treatment of ‘ti¿qesqai’ (the passive of ‘tiqe/nai’) and ‘keiÍsqai’ as synonyms.165 If a name is given to an object, the event is either the original imposition of that name to that object or the attribution of that name to that object on the occasion of a statement. In fact, Plato often uses ‘keiÍsqai’ and ‘tiqe/nai’ to express the original imposition of a name to an object.166 But he also employs them to express the attribution of a name to an object on the occasion of a statement. For instance, in the Symposium (189E2−5) he uses ‘keiÍsqai’ to express the attribution of the name ‘androgynous’ intended as an insult. Again, in the Cratylus (431B6) he employs ‘tiqe/nai’ to express the attribution of names and verbs to objects on the occasion of statements (recall that ‘keiÍsqai’ is used as a passive for ‘tiqe/nai’). Perhaps in our Sophist passage the use of ‘to be given’ (‘keiÍsqai’) in connection with names and objects does not express the original imposition of names to objects whereby the signification of the names is established, but the correct attribution of names to whatever objects they denote: ‘whatever objects the names uttered after the negative are given to’ (257C2−3) are all the objects to which the predicative expression (consisting of one or more names) uttered after the negative particle is correctly attributed. The slip from attribution (which, of course, may be incorrect as well as correct) to correct attribution is easy and natural: consider Aristotle’s use of ‘le/gesqai’ (‘to be said’) and ‘kathgoreiÍsqai’ (‘to be predicated’) to mean ‘to hold of ’.167

The ES does have a reason for mentioning the possibility that what follows the negative particle ‘not’ could consist of more than one name: the formulations he is interested in, as we shall soon see, include negative predicative expressions consisting of the negative particle ‘not’ followed by at least two words, i.e. a form of ‘to be’ and some predicative complement (e.g. ‘is not large’—recall that in Greek what corresponds to ‘not’ is followed by what corresponds to ‘is’ which in turn is followed by what corresponds to ‘large’).

164 Cf. LSJ s.v. ‘keiÍmai’ I 1.

165 In a passage of the Cratylus Plato replaces ‘keiÍtai’ (397B2) with ‘ti¿qentai’ (397B4).

166 For ‘keiÍsqai’ cf. Cra. 395C2; for ‘tiqe/nai’ cf. Cra. 416B3.

167 Cf. Arist. Cat. 3, 1b11; 1b13; 1b14; 5, 3a26; 3a27; etc.

133

Given that the extensional interpretation is correct as an exegesis of Plato’s words, is Plato’s explanation of the meaning of negation adequate? One might object to it on the grounds that the concept of otherness is a ‘negative’ concept: for, it might be claimed, ‘other than’ is to be analyzed as meaning ‘not the same as’. This objection may be answered by denying the assumption on which it relies: the concept of otherness is not a ‘negative’ concept, but a perfectly ‘positive’ or ‘affirmative’ one. In particular, one may deny that ‘other than’ is to be analyzed as meaning ‘not the same as’ and claim instead that it is a primitive expression that cannot be analyzed. From the point of view of extension, Plato’s explanation of the meaning of negation (on the assumption that the extensional interpretation is the correct exegesis of Plato’s words) is perfectly adequate, at least in so far as the negation of predicative expressions is concerned: the objects which a negative predicative expression consisting of the negative particle ‘not’ followed by a predicative expression is true of are all and only those other than everything which the predicative expression following the negative particle ‘not’ is true of.168 Plato, and for that matter Aristotle, never seems to have considered the idea that ‘not’ might govern (not a predicative expression, but) a whole sentence.

The ES’s second remark (yet again). Let us go back to the ES’s second remark: ‘Whenever we say “not being”, it seems that we do not call something contrary to what is, but only other’ (257B3−4). In our earlier discussion we reached the plausible result that the point made by the ES is that when we predicate ‘not being’ of an object, we are not saying about this object that it is contrary to what is, but that it is other than what is. On the assumption that the extensional interpretation of Plato’s account of ‘not’ is correct, this result may be sharpened: we may plausibly infer that the point made by the ES in his second remark is that when we predicate ‘not being’ of an object, we do not say about this object that it is contrary to everything that is, but that it is other than everything that is.

There remains one last, difficult question about the ES’s second remark: does it involve forms of the complete or the incomplete use of ‘not to be’ and ‘to be’? Perhaps the most plausible answer is that both uses are involved. For at 255C9−255E7, not long before the point of the text we are looking at, Plato distinguishes the complete and the incomplete use of ‘to be’.169 It is therefore likely that in presenting the view that when we predicate ‘not being’ of an object we do not say about this object that it is contrary to everything that is but that it is other than everything that is, Plato intends to involve both the complete and the incomplete use of ‘not to be’ and ‘to be’ (had he intended to restrict himself to either, he would have given some hint). One might still be worried that there is no explicit warning that an incomplete use of ‘not to be’ and ‘to be’ is part of what is being introduced. But such an unannounced introduction has an important parallel in at least one other passage, from Republic V (478E1−479B9):

168 Cf. proposition [18] of the Appendix (which however states not what it is for a negative predicative expression to be true of an object, but necessary and sufficient conditions for a kind not to hold of an object).

169 Cf. above, text to n. 67.

134

Apparently, then, it only remains for us to find what participates in both being and not being [tou= eiånai¿ te kaiì mh\ eiånai] and cannot correctly be called purely one or the other, in order that, if there is such a thing, we can rightly call it the believable, thereby setting the extremes over the extremes and the intermediate over the intermediate. Isn’t that so?

It is.

478E

Now that these points have been established, I want to address a question to our friend who does not believe in the beautiful itself or any form of the beautiful itself that remains always the same in all respects, but who does believe in the many beautiful things—the lover of sights who wouldn’t allow anyone to say that the beautiful itself is one or that the just is one or any of the rest: ‘My dear fellow’, we shall say, ‘of all the many beautiful things, is there one that will not also appear ugly? Or is there one of those just things that will not also appear unjust? Or one of those pious things that will not also appear impious?’

479A

There isn’t one, he said, for it is necessary that they appear beautiful in a way and also ugly, and the same with the other things you asked about.

What about the many doubles? Do they appear any the less halves than doubles?

479B

Not one. So with large and small things, and light and heavy ones: will any one of them be any more

described as whatever we call it than as its opposite?

No, he said, each of them will always participate in both. Is each of the many whatever one says it is, then, any more than it is not that? [po/teron ouÅn eÃsti

ma=llon hÄ ou)k eÃstin eÀkaston tw½n pollw½n tou=to oÁ aÃn tij fv= au)to\ eiånai;]

In this Republic passage Plato sets out looking for something that both is and is not and declares himself satisfied when he finds something that both is thus and so (e.g. beautiful) and is not thus and so (e.g. beautiful). Apparently, the occurrences of ‘to be’ and ‘not to be’ were to be understood as incomplete (in fact, predicative) from the start, but no warning was given that they were to be thus understood.170

The ES’s second remark may therefore be taken to make two points: (1) when we predicate ‘not being’ of an object, we are not saying about this object that it is contrary to everything that exists, but that it is other than everything that exists; (2) when we predicate ‘not being ϕ’ of an object, we are not saying about this object that it is contrary to everything that is ϕ, but that it is other than everything that is ϕ (where ‘ϕ’ is a schematic letter to be replaced by syntactically appropriate expressions).171

There are two incomplete uses of ‘not to be’ and ‘to be’: the identity use and the predicative use.172 Which of these uses is involved in the ES’s second remark? As before, since earlier in the dialogue Plato distinguished these two uses (cf. 256A3−256B5) but here provides no indication that either of them is in play, the most plausible answer is that both uses are involved. If this is correct, two points about ‘not being ϕ’ are covered: (2.1) when we predicate ‘not being ϕ’ of an object in accordance with its identity use, we are not saying about this object that it is contrary to

170 Cf. Brown (1994), 222; Malcolm (2006a), 286−7, 292.

171 The first occurrence of ‘is’ in ‘is contrary to everything that is ϕ’ and that in ‘is other than everything that is ϕ’ are occurrences of the ‘is’ of predication: this should assuage the worry, voiced by Charlton (1995), 115−16, that the extensional interpretation is committed to take the sentence ‘… we do not call something contrary to what is, but only other’ (257B3−4) as pairing a predication with an identification.

172 Cf. above, text to n. 81.

135

everything that is ϕ by being identical with ϕ, but that it is other than everything that is ϕ by being identical with ϕ; (2.2) when we predicate ‘not being ϕ’ of an object in accordance with its predicative use, we are not saying about this object that it is contrary to everything that is ϕ by partaking of ϕness, but that it is other than everything that is ϕ by partaking of ϕness. However, in view of the fact that the identity use of ‘not being ϕ’ has already been discussed in earlier pages of the dialogue (cf. 256D11−256E8), and in view also of the circumstance that predication is prominent in the immediate surroundings of the ES’s second remark (for the ES’s third remark, which begins an analogical argument to justify the second, is concerned with our calling something ‘not large’, i.e. with our predicating the negative expression ‘not large’ of something), it may be plausibly assumed that the ES is mainly thinking of the predicative use of ‘not being ϕ’. Accordingly, henceforth I shall often leave the identity use of ‘not being ϕ’ out of the picture.

The ES’s third remark (again). Having established the most plausible interpretation of the ES’s second and fourth remark, we are in a position to offer a detailed exegesis of his third remark: ‘For instance, when we call something “not large” do we then appear to you indicate the small by that expression any more than the equal?’ (257B6−7)

I take the claim made by the ES’s third remark and Theaetetus’ reply to be purely negative. We are not told that ‘when we call something “not large”, […] by that expression’ we are indicating either the small or the equal and thereby ranking the object we are talking about as being either small or equal. Again, we are not told that ‘when we call something “not large”, […] by that expression’ we are indicating either the small or the equal and thereby either ranking the object we are talking about as being small or ranking it as being equal. Rather, we are told that ‘when we call something “not large”, […] by that expression’ we are indicating neither the small nor the equal and therefore we are neither ranking the object we are talking about as small nor ranking it as equal.173 More precisely, by saying that ‘when we call something “not large”’ we do not indicate ‘the small by that expression any more than the equal’, Plato is appealing to the fact that when we call something ‘not large’ we obviously do not indicate the equal by that expression and he is inviting us to infer that since we do not indicate ‘the small by that expression any more than the equal’, we do not indicate the small either.

This purely negative interpretation of the ES’s third remark and Theaetetus’ reply fits well with the extensional interpretation of Plato’s account of ‘not’. For, if this interpretation is correct, the ES is committed to grant that ‘when we call something “not large”, […] by that expression’ we are ranking the object referred to as other than all the objects which ‘large’ is true of—and, by doing this, we are neither ranking the object in question as small nor ranking it as equal.

This purely negative claim contributes to support the negative component of the ES’s fourth remark, which is in fact presented as an inference from what precedes: ‘So, when the negative is said to mean a contrary, we shall not agree, …’ (257B9−10). Were we to agree ‘when the negative is said to mean a contrary’, we would after all be committed to grant that ‘when we call something “not large”, […] by that expression’ we are indicating the small and therefore we are

173 Cf. Owen (1971), 232, 234; Lee (1972), 287−8; Johnson (1978), 153, 156. I have changed my mind with respect to the ES’s third remark: for my earlier exegesis, see Crivelli (1990), 69−71.

136

ranking the object referred to as small. We are now in a position to see that the negative component of the ES’s fourth remark may be taken to amount to the claim that if a negative predicative expression consisting of ‘not’ followed by a predicative expression is predicated of an object then what is said about this object is not that it is contrary to every object which the predicative expression following the ‘not’ is true of.

Note that earlier in the Sophist negative predicative expressions had been treated in the way which is being now chastised. For, at 240B5−6 the ES and Theaetetus had agreed that what is not genuine is the contrary of what is genuine. Again, at 240D6−10 the ES and Theaetetus had agreed that ‘a false belief … believes the contraries of the things which are’ and therefore ‘believes the things which are not’. A similar treatment of negative predicative expressions can also be found in other dialogues: for instance, in the Protagoras (331A7−331B1) Plato goes from ‘… is not just’ to ‘… is unjust’ and from ‘… is not pious’ to ‘… is impious’. Moreover, in the Symposium (201E6−202B5), Socrates passes from the claim that love is neither beautiful nor good to the claim that it is ugly and bad and he is criticized by Diotima for not realizing that what is neither beautiful nor good can be in a condition which is intermediate between those of beauty and ugliness and those of goodness and badness. Diotima’s criticism resembles the one voiced in the third and the fourth remark of the ES in our Sophist passage.174

One aspect of the ES’s third remark that remains unclear is how the expressions ‘the small’ and ‘the equal’ are used: they might be used either in their ‘naming’ role or in their ‘generalizing’ role. If they are used in their ‘naming’ role, then by saying that ‘when we call something “not large”’ we do not indicate ‘the small by that expression any more than the equal’, Plato means that ‘when we call something “not large”’ we do not indicate smallness by that expression any more than equality. If they are used in their ‘generalizing’ role, then by saying that ‘when we call something “not large”’ we do not indicate ‘the small by that expression any more than the equal’, Plato means that ‘when we call something “not large” ’ we do not indicate small things by that expression any more than equal things. There is no need to choose one of these readings: either will provide the point that ‘when we call something “not large”’, we are neither ranking the object we are talking about as small nor ranking it as equal.

The outcome of Plato’s account of negation. We are now in a position to evaluate the outcome of Plato’s argument at 257B1−257C4. Plato may be regarded as committed to the following principles:

NCBC It is not the case that for something not to be is for it to be contrary to everything that is (where the occurrence of ‘not to be’ and the final occurrence of ‘is’ are occurrences of forms of ‘not to be’ and ‘to be’ in their existential use).

NCBO For something not to be is for it to be other than everything that is (where the occurrence of ‘not to be’ and the final occurrence of ‘is’ are occurrences of forms of ‘not to be’ and ‘to be’ in their existential use).

174 Cf. Ly. 216D5−7; Prm. 150D4−150E1; 161C7−161E1.

137

NIBC It is not the case that for something not to be ϕ is for it to be contrary to everything that is ϕ (where the occurrence of ‘not to be’ and the occurrence of ‘is’ in penultimate position are either occurrences of forms of ‘not to be’ and ‘to be’ in their predicative use or occurrences of forms of ‘not to be’ and ‘to be’ in their identity use).

NIBO 175 For something not to be ϕ is for it to be other than everything that is ϕ (where the occurrence of ‘not to be’ and the occurrence of ‘is’ in penultimate position are either occurrences of forms of ‘not to be’ and ‘to be’ in their predicative use or occurrences of forms of ‘not to be’ and ‘to be’ in their identity use).176

What relevance do these theses have for the problems addressed in the Sophist? The most plausible answer to this question relies on the assumption that in Plato’s view ‘to be’ in its incomplete use has an existential component: part of what one says about an object by saying that it is ϕ (where the occurrence of ‘is’ is one of a form of ‘to be’ either in its predicative use or in its identity use) is that it exists. In other words:

PE To be ϕ is to exist while partaking of ϕness (where the occurrence of ‘to be’ is an occurrence of a form of ‘to be’ in its predicative use).

IE 177 To be ϕ is to exist while being identical with ϕ (where the occurrence of ‘to be’ is an occurrence of a form of ‘to be’ in its identity use).

Assume PE and the position rejected by NIBC with respect to the predicative use of ‘not to be’ and ‘to be’, i.e. that for something not to be ϕ is for it to be contrary to everything that is ϕ (where the occurrence of ‘not to be’ and the occurrence of ‘is’ in penultimate position are occurrences of forms of ‘not to be’ and ‘to be’ in their predicative use). (Henceforth I forgo specifying that the predicative use of ‘not to be’ and ‘to be’ is involved. Moreover I avoid offering an explicit treatment of the case of the identity use, which however may be easily worked out by parallel reasoning.) Suppose that a is not ϕ. Then, by the position rejected by NIBC, a is contrary to everything that is ϕ. Then a is in the condition that is contrary to the condition shared by everything that is ϕ, so that a is in the condition that is contrary to the condition of being ϕ, whence (by PE ) a is in the condition that is contrary to the condition of existing while partaking of ϕness. Then a does not exist. This last step is based on the intuition that for a condition C to be contrary to a condition S is for C to be at the opposite end of the spectrum with respect to S,

175 ‘NCBC ’, ‘NCBO ’, ‘NIBC ’, and ‘NIBO ’ are acronyms for (respectively) ‘Negation of Complete Being and Contrariety’, ‘Negation of Complete Being and Otherness’, ‘Negation of Incomplete Being and Contrariety’, and ‘Negation of Incomplete Being and Otherness’.

176 NIBO, in the version according to which the occurrence of ‘not to be’ and the occurrence of ‘is’ in penultimate position are occurrences of forms of ‘not to be’ and ‘to be’ in their identity use, already appeared in my final exegesis of 256D12−256E2: cf. above, text to n. 113.

177 ‘PE ’ and ‘IE ’ are acronyms for (respectively) ‘Predication and Existence’ and ‘Identity and Existence’.

138

i.e. for C to be as far as possible removed from and opposed to S, so that the condition that is contrary to the state of existing while partaking of ϕness is a condition of non-existence ( just as the condition that is contrary to the condition of being colored while partaking of squareness is a condition of lacking color). But a was arbitrary, so whatever is not ϕ does not exist. Thus, if the position rejected by NIBC were true then it would be the case that whatever is not ϕ does not exist. So, the contribution of NIBC to the problems addressed in the Sophist is to help to block the inference from something’s not being ϕ to its non-existence.

The position formulated by NIBO does not warrant this unwelcome inference: given that for something not to be ϕ is for it to be other than everything that is ϕ, one cannot infer that whatever is not ϕ does not exist. It is instructive to see where the inference breaks down. Suppose that a is not ϕ. Then, by NIBO, a is other than everything that is ϕ, so that (by PE ) a is other than everything that exists while partaking of ϕness. From this, however, one cannot conclude that a does not exist (although London is other than everything that exists while partaking of locatedness-in-France, London does exist).

Thus, the upshot of Plato’s account of negation is the principle:

NIBNE 178 If something is not ϕ it does not follow that it is not (where the first occurrence of ‘is not’ is an occurrence of a form of ‘not to be’ either in its predicative use or in its identity use while the second occurrence of ‘is not’ is an occurrence of a form of ‘not to be’ in its existential use).

Given the difficulties concerning non-existence addressed earlier in the Sophist, one sees that NIBNE is important. In particular, it turns out to be crucial in blocking the paradox about images, whose crucial but fallacious move was the passage from ‘An image of a cat is an image but is not a cat’ to ‘An image of a cat exists and does not exist’: NIBNE enables one to block the negative component of this inference (from ‘… is not a cat’ to ‘… does not exist’).179 Again, NIBNE enables one to block one of the moves which might be involved in the paradox of falsehood of 240C7−241B4: for, as I pointed out earlier, one of the ways in which this paradox exerts its deceptive power probably relies on the assumption that if something is not ϕ then it does not exist, which is precisely what NIBNE denies.180 Moreover, NIBNE will come of help later too,181 albeit implicitly.

The two theses concerning the complete use of ‘to be’, NCBC and NCBO, remain inert: in particular, there is no advantage in assuming NCBO instead of the thesis rejected by NCBC. These theses are covered by the ES’s second remark, but make no contribution to the dialogue’s argument. The expression ‘not being’ (in the sense of ‘not existent’) is perfectly meaningful and

178 ‘NIBNE ’ is an acronym for ‘Negation of Incomplete Being and Non-Existence’.

179 Cf. above, text to n. 44 of Ch. II.

180 Cf. above, text to n. 75 of Ch. II.

181 Cf. below, text to n. 61 of Ch. V.

139

its meaning is spelled out in the section of the Sophist we are now addressing. Nevertheless, one cannot say what is not (in the sense of ‘what does not exist’) in the way one ought to in order to say falsehoods (on a conception of what is involved in saying falsehoods that will soon come under attack): this much we were told by the first aporetic argument about not being at 237B7−237E7 and the point remains unchallenged. Again, one cannot say what is not (in the sense of ‘what does not exist’) in the way one ought to in order to apply predicates to anything that is not: this much we were told by the second aporetic argument about not being at 238A1−238C12 and this point also remains unchallenged. Note that the aporetic arguments about not being could not even have been formulated or understood if ‘not being’ (in the sense of ‘not existent’) had been meaningless: the problems raised by the expression ‘what does not exist’ do not concern its meaningfulness.

Predication, identity, and existence in Plato. My interpretation relies on the assumption that Plato accepts PE, the principle that to be ϕ is to exist while partaking of ϕness (where the occurrence of ‘to be’ is one of a form of ‘to be’ in its predicative use), and IE, the principle that to be ϕ is to exist while being identical with ϕ (where the occurrence of ‘to be’ is one of a form of ‘to be’ in its identity use). Earlier182 I argued that at 255C14−15 Plato distinguishes a complete and an incomplete use of ‘to be’. I claimed that a form of ‘to be’ in its complete use is one which has no complement (not even implicit) and that from a semantic point of view an occurrence of a form of ‘to be’ in its complete use is normally employed to make a statement of existence. I also claimed that a form of ‘to be’ in its incomplete use is one which has some complement (which however can be implicit) and that from a semantic point of view an occurrence of a form of ‘to be’ in its incomplete use is normally employed to make a statement either of predication or of identity. Finally, I claimed that the complete use of ‘to be’ and the predicative incomplete one are closely connected from a semantic point of view in that there is no change in the sense of the verb: the relation between ‘Jim is’, exemplifying the complete use of ‘to be’, and ‘Jim is slow’, exemplifying the predicative incomplete use of ‘to be’, is analogous to that between ‘Jim is running’ and ‘Jim is running slowly’. I also pointed out that this position implies that statements of predication have an existential component: part of the claim made by ‘Jim is slow’ is the claim made by ‘Jim is’, which in turn is the claim made by ‘Jim exists’. This is, basically, what PE says. Similar considerations apply for the identity use of ‘to be’ and IE.

The attribution of PE to Plato is in substantial agreement with results of recent research on the relation between predication and existence in the Greek language and in ancient philosophical reflections thereupon. Lesley Brown has argued that the relation between ‘Jim is slow’ and ‘Jim is’ is analogous to that between ‘Jane is teaching French’ and ‘Jane is teaching’,183 and this view is now accepted by many scholars.184 One consideration supporting these results is that it is often

182 Cf. above, paragraph to n. 67.

183 Cf. Brown (1986/99), 458−62; (1994), 226.

184 Cf. Prior (1980), 209; Szaif (1998), 346−56; Burnyeat (2003), 9−13; Kahn (2004), 383−4, 393. Robinson (2001), 443 is unhappy with Brown’s analogy because ‘is teaching’ in ‘Jane is teaching’ is dependent or

140

hard to decide whether an occurrence of a form of ‘eiånai’ is a case of the incomplete use of the verb (linking a subject expression with a complement) or of its complete use (applied to a complex subject expression but having no complement, not even implicit): should ‘eÃsti di¿kaioj aÃnqrwpoj’ (cf. Arist. Int. 10, 19b27) be rendered by ‘A man is just’ or ‘A just man is’? In view of the fact that word order has no syntactic significance in Greek, the doubt about translation becomes even more pressing.185 Another consideration in support of the results just mentioned is that sometimes Plato places a form of ‘eiånai’ near the beginning of a sentence in a way which is clearly intended to make an existential claim, but then adds predicative expressions without repeating the verb, so that the original occurrence of the form of ‘eiånai’ is to be supplied with a predicative value:

… du/' oÃntej … oi¸ qeou\j me\n a)mfo/teroi o(mologou=ntej eiånai, paraithtou\j de\ aÀteroj, o( de\ a)meleiÍj tw½n smikrw½n. (Lg. X 901C8−901D2)

… being two … those who both admit that gods are, the one claming that they are venal, the other that they are negligent about small things.186

Examples like the above suggest that for Plato the use of ‘eiånai’ whereby it makes an existential claim is close to that whereby it makes a predication.

Negative kinds. In 257C5−257D13 Plato introduces a theory of negative kinds. Here is the relevant passage:

ES To/de de\ dianohqw½men, ei¹ kaiì soiì sundokeiÍ. THT. To\ poiÍon;

257C5

ES ¸H qate/rou moi fu/sij fai¿netai katakekermati¿- sqai kaqa/per e)pisth/mh.

THT. Pw½j;

ES Mi¿a me/n e)sti¿ pou kaiì e)kei¿nh, to\ d' e)pi¿ t% gigno/- menon me/roj au)th=j eÀkaston a)forisqe\n e)pwnumi¿an iãsxei tina\ e(auth=j i¹di¿an: dio\ pollaiì te/xnai t' ei¹siì lego/menai kaiì e)pisth=mai.

C10

257D1

THT. Pa/nu me\n ouÅn. ES Ou)kou=n kaiì ta\ th=j qate/rou fu/sewj mo/ria mia=j

ouÃshj tau)to\n pe/ponqe tou=to. THT. Ta/x' aÃn: a)ll' oÀpv dh\ le/gwmen;

D5

ES ãEsti t%½ kal%½ ti qate/rou mo/rion a)ntitiqe/menon; THT. ãEstin.

ES Tou=t' ouÅn a)nwnumon e)rou=men hà tin' eÃxon e)pwnu- mi¿an;

D10

parasitic on ‘is teaching’ in ‘Jane is teaching French’ (or ‘Jane is teaching some subject or other’) in a way in which ‘is’ in ‘Jim is’ is not with respect to ‘is’ in ‘Jim is slow’ (or ‘Jim is something or other’). I share Robinson’s unhappiness and for this reason I modified the analogy into one with ‘Jim is running’ and ‘Jim is running slowly’.

185 Cf. Kahn (2004), 383.

186 This example is due to Lesley Brown, who communicated it by private correspondence both to Myles Burnyeat, who then discussed it in Burnyeat (2003), 13, and to Charles Kahn, who discussed it in Kahn (2004), 385. Burnyeat (2003), 13 adds Lg. X 887B7−8.

141

THT. ãExon: oÁ ga\r mh\ kalo\n e(ka/stote fqeggo/- meqa, tou=to ou)k aÃllou tino\j eÀtero/n e)stin hÄ th=j tou= kalou= fu/sewj.

D13 ES And here, if you agree, is a point for us to consider. THT. Namely?

257C

ES It seems to me that the nature of the other is parceled out in the same way as knowledge. THT. How so?

ES Knowledge is also surely one, but every part of it that comes to be in respect of something, being marked off, is given a name proper to itself. Hence we speak of many arts and knowledges.

THT. Certainly.

257D

ES And the same happens to the parts of the nature of the other, which is one. THT. Perhaps. But could we say how?

ES Is there a part of otherness that is contrasted with the beautiful? THT. There is.

ES Shall we say it is nameless, or that it has a denomination? THT. It has one: for what on any occasion we call ‘not beautiful’, this is other than nothing else but

the nature of the beautiful.

The ES’s first remark (‘And here, if you agree, is a point for us to consider’) indicates that a new topic is coming up for discussion. The new topic is that of negative kinds. For present purposes, let a negative kind be the negative counterpart of a given kind, namely a kind which holds of all and only the items of which the given kind does not hold. If K is a kind that matches a predicative expression, the negative counterpart of K matches the negation of that negative expression (i.e. the predicative expression obtained by prefixing ‘not’ to the original predicative expression) (where a kind matches a predicative expression just in case it holds of all and only the items which the predicative expression is true of ). Let it be noted that by attributing to Plato the view that certain negative kinds match certain negative predicative expressions I am not committing him to the view that for every negative predicative expression there is a matching kind.

In order to establish that there are negative kinds, Plato describes them as bearing a certain relation to the kind otherness. And to explain the relation that negative kinds bear to the kind otherness, he draws an analogy between otherness and knowledge. There are two relevant aspects to this analogy.

The first aspect concerns the ontological and logical character of the relationship of negative attributes to otherness.187 Just as particular knowledges are marked off from knowledge and come to be by virtue of the relationship of knowledge to all the objects that fall under a certain kind (e.g. grammar is marked off from knowledge and comes to be by virtue of the relationship of knowledge to all the objects that fall under the kind letter—for grammar is knowledge of all letters, cf. 252E9−253A12),188 so also negative attributes are marked off from otherness and come to be by virtue of the relationship of otherness to all the objects that fall under a certain kind (e.g.

187 For the fragmentation of knowledge, cf. Ion 537C5−538B1; R. IV 438C6−438D10; V 477C1−477E1.

188 Also note the plural ‘e(te/rwn pragma/twn’ at Ion 537D5.

142

the negative attribute non-beauty is marked off from otherness and comes to be thanks to the relationship of otherness to all the objects that fall under the kind beauty—for the negative attribute non-beauty is otherness than all beautiful objects).189

The logical mechanism involved in Plato’s analogy is close to what a modern logician would describe as the saturation of one of the free variables of an open formula by means of a quantifier. Take a formula A(x, y) containing free exactly the two variables displayed. This formula A(x, y) expresses a two-place relation (like knowledge and otherness). Take a further formula B( y) containing free only the variable displayed. This formula B( y) expresses an attribute (i.e. a one-place relation, like health and beauty). Now construct a new formula, ∀y (B( y) → A(x, y)). This new formula contains only one free variable and therefore represents an attribute (i.e. a one-place relation): it expresses the attribute of bearing the two-place relation expressed by A(x, y) to everything that has the attribute expressed by B( y). For instance, if A(x, y) expresses knowledge and B( y) expresses the attribute of being a letter, the formula ∀y (B( y) → A(x, y)) expresses the attribute of knowledge of everything that falls under the attribute of being a letter; if instead A(x, y) expresses otherness and B( y) expresses beauty, the formula ∀y (B( y) → A(x, y)) expresses the attribute of otherness than everything that falls under beauty.

There is a hint that matters were viewed in the Academy in a way not too different from the one sketched in the last paragraph. For in the Categories, probably an early work that still echoes views circulating in the Academy, Aristotle discusses the relationship of the genus knowledge to its species, specific knowledges like grammar and music. He claims (8, 11a20−36) that while the genus knowledge is a relative (because knowledge is called knowledge of something), its species are not relatives: for grammar is not called grammar of something, nor is music called music of something.190

The second aspect of Plato’s analogy concerns the linguistic formulation of negative attributes. Just as the particular knowledges marked off from knowledge have special names derived from those of the kinds to which they are related (cf. ‘gra/mma’-‘grammatikh/’, ‘po/lij’-‘politikh/’), so also the negative attributes marked off from otherness have special names derived from those of the kinds to which they are related (cf. ‘beauty’-‘non-beauty’).191

Some aspects of Plato’s language in 257C5−257D13. Here are some remarks on the terminology of 257C5−257D13.

(1) At 257D11−13 Plato uses the sentence: ‘What on any occasion we call “not beautiful”, this is other than nothing else but the nature of the beautiful [th=j tou= kalou= fu/sewj]’. Given that the extensional interpretation of Plato’s account of negation is correct, the claim which Plato ought to make at this point is that what on any occasion we call ‘not beautiful’ is other than all objects denoted by ‘beautiful’, or (equivalently) that what on any occasion we call ‘not beautiful’

189 Cf. Lee (1972), 269−76, 278−9; Szaif (1998), 438.

190 Cf. Lee (1972), 272.

191 Cf. Lewis (1976), 114.

143

is other than all beautiful things. We may identify the claim that Plato ought to make with the one he does make by using the sentence at 257D11−13 if we recall the fact that in Greek, and especially in Plato’s Greek, expressions constructed around ‘h( fu/sij’ governing a noun-phrase or an adjectival phrase in the genitive (singular or plural) are often periphrastic: they often mean the same as the result obtained by deleting ‘h( fu/sij’ (and making the further adjustments required by grammaticality).192 For, according to this interpretation, the sentence uttered by Theaetetus at 257D11−13, ‘What on any occasion we call “not beautiful”, this is other than nothing else but the nature of the beautiful’, means the same as ‘What on any occasion we call “not beautiful”, this is other than nothing else but the beautiful’, which in turn is to be understood as meaning the same as ‘What on any occasion we call “not beautiful”, this is other than nothing else but all beautiful things’.193

It might be objected that this interpretation of the sentence containing the expression ‘the nature of the beautiful’ is arbitrary and forced. But there are several parallels in Plato’s works:

(α) ‘Shan’t we also say that what exceeds the nature of the moderate [th\n tou= metri¿ou fu/sin] and what is exceeded by it really come to be either in speeches or in actions?’ (Plt. 283E3−5) Here ‘the nature of the moderate’ means ‘moderate things’.

(β) ‘The most extreme case of all is when the nature of the marrow [h( tou= muelou= fu/sij] becomes diseased because of some deficiency or excess’ (Ti. 84C3−5). Here ‘the nature of the marrow’ means ‘instances of marrow’ or ‘bits of marrow’.

(γ) ‘… the nature of the wing [h( tou= pterou= fu/sij], by which the soul is lifted, is nourished by this’ (Phdr. 248C1−2). Here ‘the nature of the wing’ means ‘the wing’ or ‘wings’.

(δ) ‘… the nature of the kinds [h( tw½n genw½n fu/sij] has reciprocal communion’ (Sph. 257A9). Here ‘the nature of the kinds’ means ‘kinds’.194

Some commentators195 take issue with the exegesis developed in the preceding paragraphs. They attack it by mentioning several points in the Sophist where Plato uses sentences that contain expressions constructed around ‘h( fu/sij’ governing a noun-phrase or an adjectival phrase in the genitive and understands them as referring to kinds without bringing in what partakes of them (255D9, 256D12−256E1, 257C7, 257D4, 258A7−8, 258A11, 258D7). This attack misfires. There is no denial that sentences containing expressions constructed around ‘h( fu/sij’ governing a noun-phrase or an adjectival phrase in the genitive may be understood as referring to kinds without bringing in what partakes of them. Evidence that such a usage is present in Plato does nothing to disprove the existence of a different usage, which is also well attested.

192 Cf. Ast (1835−38), s.v. ‘fu/sij’ (III 522−3); LSJ s.v. ‘fu/sij’ II 5; des Places (1970), s.v. ‘fu/sij’ 4o (II 559).

193 Cf. Frede (1967), 88; Owen (1971), 238.

194 Further examples: Smp. 186B4; 191A5; Phd. 79B9−10; R. IV 429D6; IX 589B4; Phdr. 251B2−3; Plt. 267B5; 283E3; Ti. 55B5; 62C3−4; 75D3−4; 84C6; Phlb. 25E7−8; 26E6. This should answer the doubt raised by Lewis (1976), 114 by saying that he has found ‘no parallel in which sense requires that “h( tou= X fu/sij” (with singular “tou= X ”) must be reconstructed unambiguously as a universally quantified plural’.

195 Cf. Kostman (1973), 200; Ketchum (1978), 60; van Eck (1995), 27; Szaif (1998), 439−40.

144

One might still ask why at 257D12−13 Plato chooses to employ the noun-phrase ‘the nature of the beautiful’ in a sentence which he wants to be understood as saying something about beautiful things, if the claim he is making could have been made less ambiguously by using ‘beautiful things’. The most plausible answer appeals to the fact that at 257D11−13 Plato is still in the initial stage of an argument leading to the conclusion that the objects denoted by a negative predicative expression like ‘not beautiful’ and ‘not being’ are unified by a single nature: it is not a matter of chance that he later speaks of ‘the nature of a part of otherness’ (258A11) and that he boasts that ‘not being is firmly a thing that has its own nature’ (258B11). This conclusion, which amounts to the claim that there are negative attributes corresponding to negative predicative expressions, is far from trivial: it is surely not obvious that the objects denoted by a negative predicative expression are unified by a single nature. In particular, this is not obvious at the beginning of Plato’s argument, which is the stage to which 257D11−13 belongs. At this initial stage, it is worth while to hint at a contrast between the objects denoted by ‘not beautiful’, of which it is still unclear whether they are unified by a single nature, and the objects denoted by ‘beautiful’, which are obviously unified by a single nature. Such a hint Plato makes by using the expression ‘the nature of the beautiful’: the sentence does quantify on all beautiful things, but it alludes to their being unified by a single nature.196

(2) The sentence ‘What on any occasion we call “not beautiful”, this is other than nothing else but [ou)k aÃllou tino\j … hÄ] the nature of the beautiful’ (257D11−13) raises a further difficulty. Some commentators197 take the phrase ‘nothing else but’ to commit Theaetetus to the claim that what we call ‘not beautiful’ is other than the nature of the beautiful and only from it. If these commentators are right and, as the extensional interpretation of Plato’s account of negation requires, the noun-phrase ‘the nature of the beautiful’ at 257D12−13 is taken to mean ‘all beautiful things’, then Theaetetus would be committed to the claim that what we call ‘not beautiful’ is other than all beautiful things and only from them, which in turn entails the obvious falsehood that ‘not beautiful’ is true of at most one thing. Such a result could be regarded a reductio ad absurdum of the extensional interpretation of Plato’s account of negation. However, the phrase ‘nothing else but’ need not be taken to commit Theaetetus to the claim that what we call ‘not beautiful’ is other than the nature of the beautiful and only from it: it is most plausibly understood as focusing the reader’s attention on the nature of the beautiful, i.e. beautiful things.198

(3) Plato uses the expression ‘contrasted’ (‘a)ntitiqe/menon’, 257D7, cf. 257E3, 257E6, 258B1, and Plt. 263D6) to describe the relation between the part of otherness which is non-beauty and beauty. This contrast probably consists in the fact that non-beauty and beauty are necessarily disjoint: it cannot be the case that any object falls under both non-beauty and beauty (such an

196 This paragraph is indebted to a discussion with Stefan Koller.

197 Cf. Kostman (1973), 198; Ketchum (1978), 45; van Eck (2000), 64.

198 Cf. Chrm. 167B11−167C1; 168D3−4; Men. 84C11−84D1; Phd. 64C4−5; Sph. 247E4; Frede (1967), 88; Lee (1972), 275; Lewis (1976), 113; Crivelli (1990), 48.

145

object would have to fall under beauty while being other than everything that falls under beauty, so that it would have to be other than itself—which is absurd).

The alleged clash with the Statesman. There is a tension between what Plato appears to say in this part of the Sophist, namely that there are negative kinds corresponding to negative predicative expressions, and what he appears to say in the Statesman (262A3−264B6), namely that there are no negative kinds corresponding to negative predicative expressions. Some commentators assume that Plato changed his mind between writing the Sophist and the Statesman.199 Others assume that this section of the Sophist does not commit Plato to negative kinds corresponding to negative predicative expressions:200 this route however is not promising because Plato does explicitly affirm that non-being, the part of otherness contrasted with being, is a form (cf. 258C2−4, 258D5−7, 260B7−8) and it is hard to see how he could coherently refrain from making the same claim about the parts of otherness contrasted with kinds other than being.

However, a reconciliation of the views presented in the Sophist and the Statesman is perhaps possible. For in the Statesman Plato is not concerned with the question whether there are negative kinds corresponding to negative predicates, but with the question of how divisions should be carried out. Specifically, he prescribes that in divisions one should not separate off a ‘small’ species of a kind by contrasting it with the remainder of that kind (for instance, one should avoid dividing the kind animal in such a way as to obtain the kind man contrasted with beast). The problem Plato has in mind might be that the remainder of the separation of a ‘small’ species from a kind straddles across the kind in a way that is incompatible with a correct classification generated by the process of division. Graphically:

footed-domestic-animal

biped quadruped

man bird cat dog

man footed-domestic-animal-other-than-man

The first three rows of this table display a classification generated by dividing the kind footed-domestic-animal.201 This kind is divided into the species biped and quadruped, each of which is therefore immediately subordinate to it. Each of these species is then divided again: the species biped is divided into the further species man and bird, each of which is immediately subordinate to it, and the species quadruped is divided into the further species cat and dog, each of which is immediately subordinate to it. The fourth row of the table shows the result of separating off the ‘small’ species man from the kind footed-domestic-animal: the remainder of this separation, namely

199 Cf. Frede (1967), 93−4.

200 Cf. Dixsaut (1991), 196−7, 198; Fine (1993), 115.

201 Cf. above, text to n. 7 of Ch. I.

146

footed-domestic-animal-other-than-man, straddles across the kind footed-domestic-animal in a way that makes it impossible for it to be a kind belonging to the classification displayed in the first three rows. To see why this is impossible, consider that within a classification, if one kind is properly extensionally included in another, then it is subordinate to it: if footed-domestic-animal-other-than-man were a kind belonging to the classification displayed in the first three rows, then, since the kind quadruped would be properly extensionally included in the kind footed-domestic-animal-other-than-man, which in turn would be properly extensionally included in the kind footed-domestic-animal, it would follow that the kind quadruped is subordinate to the kind footed-domestic-animal-other-than-man which in turn is subordinate to the kind footed-domestic-animal, which contradicts the earlier result that the kind quadruped is immediately subordinate to the kind footed-domestic-animal. For this reason footed-domestic-animal-other-than-man does not constitute a species of the kind footed-domestic-animal, but is merely a part of it. For all we are told in the Statesman, the remainder of the separation of a small species from a kind may well constitute a single kind, a kind that is a part but not a species of the genus. So, if this interpretation of the passage from the Statesman is correct, then the passage does not clash with the Sophist’s acceptance of negative kinds corresponding to negative predicative expressions.202 One advantage of this interpretation of the passage from the Statesman is its ability to account for certain features of the text: first, for Plato’s insistence that one should avoid separating ‘small’ species within ‘large’ kinds (cf. Plt. 262A9, 262B6); secondly, for his claim that every species of something is a part of it (cf. Plt. 263B8−9), a claim which suggests that the problem addressed is not that of whether there are kinds corresponding to certain groups of individuals, but that of whether certain groups of individuals constitute species of given genera.

Aristotle (Metaph. Α 9, 990b13−14; Μ 4, 1079a9−10; Alex. Aphr. in Metaph. 80, 15−81, 7, a fragment of the now lost On Ideas) argues that the Platonic ‘One Over Many’ argument for forms would, if sound, establish the existence of forms corresponding to negative expressions and then makes the case that such a result should be unpalatable for the Platonists. However, nothing guarantees that Aristotle’s views as to what should or should not be acceptable to a Platonist were shared by Plato himself.

The being of not being. In 257D14−258C6 Plato finally offers his proof that not being is. Here is the passage:

ES ãIqi nun to/de moi le/ge. THT. To\ poiÍon;

257D14 257E1

ES ãAllo ti tw½n oÃntwn tino\j e(no\j ge/nouj me/roj a)fo- risqe/n, kaiì pro/j ti tw½n oÃntwn auÅ pa/lin a)ntiteqe/n, ouÀtw

sumbe/bhken eiånai to\ mh\ kalo/n; THT. OuÀtwj.

E5

ES ãOntoj dh\ pro\j oÄn a)nti¿qesij, w¨j eÃoik', eiånai¿ tij sumbai¿nei to\ mh\ kalo/n.

THT. ¹Orqo/tata.

202 Cf. Szaif (1998), 441.

147

ES Ti¿ ouÅn; kata\ tou=ton to\n lo/gon aÅra ma=llon me\n to\ kalo\n h(miÍn e)sti tw½n oÃntwn, hÂtton de\ to\ mh\ kalo/n;

THT. Ou)de/n.

E10

ES ¸Omoi¿wj aÃra to\ mh\ me/ga kaiì to\ me/ga au)to\ eiånai lekte/on;

THT. ¸Omoi¿wj.

258A1

ES Ou)kou=n kaiì to\ mh\ di¿kaion t%½ dikai¿% kata\ tau)ta\ qete/on pro\j to\ mhde/n ti ma=llon eiånai qa/teron qate/rou;

THT. Ti¿ mh/n;

A5

ES Kaiì taÅlla dh\ tau/tv le/comen, e)pei¿per h( qate/rou fu/sij e)fa/nh tw½n oÃntwn ouÅsa, e)kei¿nhj de\ ouÃshj a)na/gkh dh\

kaiì ta\ mo/ria au)th=j mhdeno\j hÂtton oÃnta tiqe/nai. THT. Pw½j ga\r ouÃ;

A10

ES Ou)kou=n, wj eÃoiken, h( th=j qate/rou mori¿ou fu/sewj kaiì th=j tou= oÃntoj pro\j aÃllhla a)ntikeime/nwn a)nti¿qesij

258B1

ou)de\n hÂtton, ei¹ qe/mij ei¹peiÍn, au)tou= tou= oÃntoj ou)si¿a e)sti¿n, ou)k e)nanti¿on e)kei¿n% shmai¿nousa a)lla\ tosou=ton mo/non,

eÀteron e)kei¿nou. THT. Safe/stata/ ge.

B5

ES Ti¿n' ouÅn au)th\n prosei¿pwmen; THT. Dh=lon oÀti to\ mh\ oÃn, oÁ dia\ to\n sofisth\n e)zh-

tou=men, au)to/ e)sti tou=to.

ES Po/teron ouÅn, wÐsper eiåpej, eÃstin ou)deno\j tw½n aÃllwn ou)si¿aj e)lleipo/menon, kaiì deiÍ qarrou=nta hÃdh le/gein

B10

oÀti to\ mh\ oÄn bebai¿wj e)stiì th\n au(tou= fu/sin eÃxon, wÐsper to\ me/ga hÅn me/ga kaiì to\ kalo\n hÅn kalo\n kaiì to\ mh\ me/ga <mh\

258C1

me/ga> kaiì to\ mh\ kalo\n <mh\ kalo/n>, ouÀtw de\ kaiì to\ mh\ oÄn kata\ tau)to\n hÅn te kaiì eÃsti mh\ oÃn, e)na/riqmon tw½n pollw½n

oÃntwn eiådoj eÀn; hà tina eÃti pro\j au)to/, wÕ Qeai¿thte, a)pisti¿an eÃxomen;

THT. Ou)demi¿an.

C5 C6

ES Then tell me this. THT. What?

257D 257E

ES Is it not the case that203 by being a part marked off from a certain single kind among those which are, and besides by being contrasted with one of those which are—in this way the not beautiful turns out to be?

THT. Yes. ES So, as it seems, the not beautiful is a contrast of a being with respect to a being. THT. Right.

ES What then? According to this argument, shall we hold that the beautiful is to a higher degree one of the beings whereas the not beautiful is one to a lesser degree?

THT. Not at all. ES Therefore we must say that the not large and the large itself are to the same degree? THT. To the same degree.

258A

ES And we must also put the not just on the same footing as the just with respect to the fact that neither of them is to a higher degree than the other?

THT. Certainly.

ES And we shall say the same of all the rest, because it turned out that the nature of the other is to be ranked among beings and, since it is, we must necessarily hold that its parts also are no less than anything else.

THT. Of course.

203 Taking ‘aÃllo ti’ as the Latin ‘nonne’: cf. Lee (1972), 278. For a different rendering cf. Owen (1971), 239.

148

ES So, it seems, the contrast of the nature of a part of otherness and that of being, which are mutually opposed, is, if it be permissible to say so, a being no less than being itself and it does not mean contrariety to it, but only otherness than it.

THT. That is quite clear. 258B ES How shall we call it? THT. Clearly this is precisely not being, which we were seeking for the sake of the sophist.

ES Therefore, as you say, it is not inferior to any of the rest in being? May we now be bold to say that not being is firmly a thing that has its own nature—just as the large was large and the beautiful was beautiful, so too with the not large and the not beautiful, and not being also in the same way was and is not being, a single kind to be reckoned among the many which are? Or have we any further doubts with regard to it, Theaetetus?

THT. None at all.

258C

The ES’s initial remark (‘Then tell me this’, 257D14) indicates that a new topic is going to be discussed. The new topic is the being of negative attributes, or negative kinds. This portion of text represents the culmination of Plato’s argument: the claim that not being is.

Being as non-emptiness. What does Plato mean when he describes negative kinds as being? He might mean that they are instantiated: according to this suggestion, for a kind to be is for it to be non-empty, i.e. to hold of something or other. This is a perfectly possible way of understanding the jargon of the being of kinds: one can say ‘Nastiness is’ to mean that there are nasty people, i.e. that nastiness is a non-empty, or instantiated, attribute.204

Despite its attractiveness, this interpretation of the jargon of the being of kinds must be rejected because it saddles Plato with a disastrously fallacious argument. For, suppose the interpretation under examination were correct. Then, at 258A7−10, Plato would be arguing that since otherness is, i.e. is non-empty, then all its parts also are, i.e. are non-empty. This argument is invalid. The non-emptiness of otherness is a necessary, but not a sufficient condition for the non-emptiness of its parts. To make it into a sufficient condition, one additional requirement must be made: that the kind with which the part of otherness is contrasted should not be universal (i.e. it must not be the case that it holds of everything). For suppose that a kind K is universal: K holds of everything. Consider the part of otherness contrasted with K: otherness than everything of which K holds. Suppose that this part of otherness were to hold of something, say a. Then a would be other than everything of which K holds. But, since it holds of everything, K would hold also of a, so that a would be other than itself—which is absurd. We may therefore conclude that the part of otherness in question holds of nothing, i.e. fails to be non-empty, so that it is not. Therefore, if a kind is universal, the part of otherness contrasted with it is not. Thus, from the fact that otherness is it does not follow all its parts also are.

Being as definability. A more plausible understanding of what Plato means when he describes negative kinds as being is given by the assumption that the being of a kind consists in its having a definite nature, its occupying a determinate position within the world of forms.205 This

204 I defended this exegesis in Crivelli (1990), 75.

205 Cf. Lee (1972), 276−7; Roberts (1986), 234; Szaif (1998), 442.

149

interpretation gives Plato a better argument for the conclusion that negative kinds are beings. The process of generation of the parts of otherness, whereby the part of otherness corresponding to a given kind K is otherness than everything of which K holds, ensures that the parts of otherness have a perfectly definite nature which they inherit from otherness itself and from the kind with respect to which the part is isolated. Since the negative counterpart of any kind K is identified with the part of otherness corresponding to K, it follows that negative kinds have definite natures and therefore are beings.

Part of the reason why Plato speaks of a kind having a definite nature as its being is probably linked to the close connection between the existential and predicative uses of ‘to be’: as I pointed out earlier, Plato is probably committed to PE, the principle that to be ϕ is to exist while partaking of ϕness (where the occurrence of ‘to be’ is one of a kind of ‘to be’ in its predicative use). Given that to be thus and so by partaking of a genuine kind is a way of existing, kinds may be viewed as endowers of being. This function of endowing with being might justify the talk of kinds as beings.

After considering the case of non-beauty and non-justice, namely the negative kinds contrasted with beauty and justice, Plato comes to consider the case of non-being, namely the negative kind contrasted with being. According to the by now well rehearsed pattern of argument, non-being amounts to otherness than everything that partakes of being. Since there are a complete as well as an incomplete use of ‘to be’, one wonders whether non-being is non-existence (corresponding to the complete use of ‘not to be’) or non-being ϕ (corresponding to the incomplete use of ‘not to be’). Given that no restriction is indicated in the text, the most likely solution is that both kinds of non-being are covered: non-existence, which amounts to otherness than everything that partakes of being, i.e. otherness than everything that exists, and non-being ϕ, which amounts to otherness than everything that partakes of being with respect to ϕness, i.e. otherness than everything that is ϕ.

Even non-existence is granted citizenship in Plato’s ontology. This does not clash with the results of the aporetic arguments about not being, which continue to hold good and remain as puzzling as before. For in those aporetic arguments Plato did not raise doubts about the legitimacy of the negative attribute of non-existence. Rather, he argued that it is impossible to say what does not exist and that what does not exist cannot be referred to in mental or uttered discourse. Neither of these results entails that the negative attribute of non-existence should be banned.206

206 This view is in sharp contrast with the one adopted by Moravcsik (1962), 26−7.

CHAPTER V FALSE STATEMENT AND FALSE BELIEF

The need for a discussion of statement. After developing his account of not being, Plato goes on to address some problems that may be broadly described as pertaining to philosophy of language (259D9−264B10). The ES declares that he and Theaetetus must ‘agree what statement is’ (260A7−8)—that is, define statement (cf. 263C2). Theaetetus wonders why this is needed (260B3−4). The ES explains (260B5−261C10) that since to state, or believe, a falsehood is to state, or believe, what is not, the sophist could still adopt a last defense based on denying that not being combines with statement and belief: only by defining statement and belief will it be possible to show that not being combines with them. At first glance, this last defense of the sophist seems a silly and desperate move. An account of not being in terms of otherness has been offered; we have been hearing all along that to state, or believe, a falsehood is to state, or believe, what is not; why on earth should we doubt that not being combines with statement and belief ?

The subsequent discussion divides into three parts: a definition of statement (261D1−262E10), a proof that statements can be false (262E11−263D5), and a definition of belief on the basis of which it can be easily established that beliefs can be false (263D6−264B5).

Words, names, and verbs. When it comes to defining statement, the ES and Theaetetus do not apply the method of division. Instead, they describe statements of the simplest kind. Does this suffice to yield a definition? The Socrates of Plato’s early dialogues would probably have denied it.

The ES distinguishes (261E4−262A8) two kinds of words (o)no/mata, 261D2), or vocal indicators (tv= fwnv= dhlw/mata, 261E5): verbs (r(h/mata), which signify actions (pra/ceij), and names (o)no/mata), which signify objects (pra/gmata, 262E13), i.e. ‘those who perform the actions’ (262B10).1 When words of only one kind are combined, the resulting string signifies nothing: only if words of one kind are combined with words of the other is the resulting string endowed with signification (261D9−261E2). More specifically, when words of only one kind are combined, the resulting string does not constitute a statement (lo/goj) (262A9−11): if only verbs are combined, the resulting string fails to be a statement (e.g. the string of verbs ‘walks runs sleeps’ is not a statement) (262B2−8); if only names are combined, the resulting string again fails to be a statement (e.g. the string of names ‘lion stag horse’ is not a statement) (262B9−262C2). Only if verbs are combined with names does the resulting string constitute a statement (262C4−6). For instance, ‘Man understands’ is a statement (262C9−262D1) and it is obtained by combining the verb ‘understands’ with the name ‘man’.

1 The expressions used to describe the relation of words to what they stand for are the nouns ‘indicator’ (‘dh/lwma’, 261E5, 262A3) and ‘sign’ (‘shmeiÍon’, 262A6, 262D9) and the verb ‘to signify’ (‘shmai¿nein’, 262B6).

151

Let me pause for a few remarks. (1) The noun ‘oÃnoma’ is used in two ways, a narrow and a broad one. On its narrow usage, on which it is best rendered by ‘name’, ‘oÃnoma’ denotes the vocal indicators that signify objects. On its broad usage, on which it might be translated by ‘word’, it denotes all vocal indicators (including those that signify actions as well as those that signify objects).2 (2) The distinction between actions and objects is unclear. Is it an exhaustive ontological classification, so that every being is an action just in case it is not an object? (3) The distinction between verbs and names also is unclear. Is it contrasting the grammatical categories of verbs and names, or the syntactic categories of predicate expressions and subject expressions? Both alternatives face difficulties. For, arguably, not every member of the grammatical category of verbs signifies an action (consider ‘is carried’, to which in Greek corresponds a single word, ‘fe/retai’). Analogously, it looks as if not every predicate expression signifies an action, unless the category of actions is larger than expected (consider ‘is tall’).3 (4) Although verbs signify actions, the contribution made by a verb to a statement of which it is a component cannot be identified with its signifying an action: otherwise the statement ‘Man understands’ would mean the same as the string of names ‘man, understanding’.4 However, although the contribution made by a verb to a statement of which it is a component cannot be identified with its signifying an action, it is nevertheless coherent to claim that the verb’s contribution involves signifying an action.5

Naming and stating. The ES and Theaetetus agree that one name and one verb make up a statement that is shortest (smikro/tatoj, e)la/xistoj, braxu/tatoj) and primary (prw½toj) (262C5−262D1, 263C1−4). This presupposes that there are statements of other kinds, in particular, longer and non-primary statements that do not consist of merely one name and one verb. These other kinds of statement are not described. The use of ‘primary’ suggests that statements of other kinds are composed out of primary ones (whose components are not statements, but names and verbs), much in the same way as, according to the Cratylus (422A1−422E1), derivative names are composed out of primary ones (whose components are not names, but syllables and, ultimately, letters). Statements can concern not only the present, but also the past and the future (262D2−3).

The ES remarks that when a string composed only of names, or one composed only of verbs, is pronounced, ‘either way the utterance reveals no action nor inaction nor being of what is or of what is not’ (262C2−4). The part of this remark about the failure to reveal the ‘being of what is or of what is not’ is obscure. It probably involves the predicative elliptical use of ‘to be’ (whereby ‘to be’ is employed as a copula to be completed with a predicative expression which however is suppressed and remains understood): the point made is probably that a string of words of the sort

2 Cf. Stough (1990), 370.

3 Cf. Xenakis (1957), 168; Galligan (1983), 272; Stough (1990), 370.

4 Cf. Denyer (1991), 164−7.

5 Cf. Davies (1981), 108−9.

152

described fails to signify the being thus and so of what either is thus and so or is not thus and so. This sounds like an anticipation of the account of truth and falsehood given later.

The ES says that a speaker6 producing a primary statement, i.e. a statement composed of one name and one verb, ‘does not only name something [o)noma/zei], but accomplishes something [ti perai¿nei]’ (262D3−4):7 a speech-act is brought to completion.8 He adds that in producing a primary statement, a speaker ‘does not only name something [o)noma/zein], but also states something [le/gein]’ (262D5). This remark presupposes that naming and stating are different. No explanation of what they are is offered. The ES and Theaetetus then agree (262E4−8) that every statement must be ‘of ’, or ‘about’, something. Their later observations (262E13−263A11, 263C1−12) on the primary statements introduced as examples, ‘Theaetetus is sitting’ and ‘Theaetetus is flying’, show that the item a primary statement is about is the object signified by its name. In the face of these data, let me indulge in some speculation.

(1) What is it that a speaker producing a primary statement names? Does such a speaker name both the object signified by the primary statement’s name and the action signified by its verb?9 Or only the object?10 The ES and Theaetetus do not address this problem, but the etymological link between ‘o)noma/zein’ (‘to name’) and ‘oÃnoma’ (‘name’) suggests that the last alternative is the right one: a speaker producing a primary statement names only the object signified by its name.11

(2) Granted that this result is correct, a further point may be plausibly inferred: in a primary statement, the name is what mainly contributes to the speaker’s performing the speech-act of naming whereas the verb is what mainly contributes to the speaker’s performing the speech-act of stating.12

6 Taking ‘tij’ at 262C9 as the grammatical subject of ‘ti perai¿nei’ at 262D4 (cf. Nuchelmans (1973), 15).

7 Cf. Cra. 425A2−3. For the phrase ‘ti perai¿nein’, cf. Grg. 472B8; Smp. 217C1−2; R. IV 426A2; Tht. 180A6−7.

8 Cf. Xenakis (1957), 168−9; Nuchelmans (1973), 15−17; Swiggers (1984), 16; Stough (1990), 371, 379. Other commentators (e.g. Derbolav (1972), 174; Galligan (1983), 270; Rudebusch (1990), 601−2) take the ES to be claiming that a speaker producing a primary statement ti perai¿nei in the sense of limiting something: such a speaker limits both the object signified by the primary statement’s name (by specifying what action it is performing) and the action signified by the primary statement’s verb (by specifying which object is performing it).

9 Cf. Galligan (1983), 270, 278; Swiggers (1984), 16.

10 Cf. Owen (1971), 263.

11 It is less likely that ‘o)noma/zein’ should be connected to ‘oÃnoma’ in its broad usage (in which case a speaker producing a primary statement would probably name both the object signified by the primary statement’s name and the action signified by its verb): for the wide usage of ‘oÃnoma’ appears only at the beginning of the linguistic section (261D2, 261D4) and is then superseded by the narrow usage.

12 Cf. Frede (1992), 413−14. By contrast, Stough (1990), 370−1 denies that in Plato’s view there is a difference between the semantic functions of names and verbs (apart from that consisting in their signifying items of different sorts).

153

(3) Given that a speaker producing a primary statement names only the object signified by its name, and given that the item a primary statement is about is the object signified by its name, it follows that a speaker producing a primary statement names only the item the primary statement is about. On the plausible assumption that a speaker producing a primary statement refers only to the item the primary statement is about, a further inference can be plausibly drawn: for a speaker who produces a primary statement to name an item is to refer to it.

(4) One reason for insisting that every statement must be about something is probably the need to do away with an assumption made by one version of the falsehood paradox—that is, the assumption that a false statement is about nothing (because it is about what is not). Note that two different uses of ‘about’ are around: that whereby ‘about’ expresses the relation of a statement to its referent (in this use, ‘Theaetetus is flying’ is about Theaetetus) and that whereby ‘about’ expresses the relation of a statement to what is stated (in this use, ‘Theaetetus is flying’ is about Theaetetus’ flying).

(5) Plato is probably committed to the claim that strings of words which one might be inclined to describe as singular predicative statements with empty subject expressions (e.g. ‘Pegasus is flying’) are not genuine statements. Some modern philosophers of language (e.g. Frege) explicitly endorse this claim.

(6) When he says that a speaker makes a primary statement ‘by putting an object together with an action [sunqeiìj pra=gma pra/cei] by means of a name and a verb’ (262E13−14), the ES does not commit himself to the view that by making a primary statement a speaker states, or bears some other propositional attitude to, a single composite item which he or she puts together from the object and the action signified by (respectively) the primary statement’s name and verb. In other words, what the ES says does not commit him to the view that speakers bear propositional attitudes to something like propositions or Fregean Gedanken: such a view would be exposed to difficulties similar to those brought to light by the aporetic arguments about not being of 237B7−239C8. What the ES says is consistent with a weaker view, namely that the production of a primary statement by a speaker coincides with an event in which the object and the action signified by (respectively) the primary statement’s name and verb are both involved.13

(7) In the Cratylus, in connection with a version of the falsehood paradox (Cra. 429C6−430A5), Socrates and Cratylus agree (430A6−431C3) both that one can assign14 names to objects and that such an assignment15 can be carried out correctly, and therefore truly, as well as incorrectly, and therefore falsely. One is tempted to modify this Cratylus account to fit the situation of the Sophist by substituting actions for names. The result is: when in producing a primary statement a speaker brings about an event (a speech-act of stating) in which the object and the action signified by (respectively) the primary statement’s name and verb are both

13 Cf. above, text to n. 33 of Ch. II.

14 The verbs used are ‘diane/mein’ (430A7−8, 430E1), ‘prosfe/rein’ (430A8), and ‘a)podido/nai’ (431B4).

15 The nouns used are ‘dianomh/’ (430D3, 431B1), ‘do/sij’ (430D6), and ‘e)pifora/’ (430D6).

154

involved, the event in question is a case of assigning the action signified by the primary statement’s verb to the object signified by the primary statement’s name.

True and false statements. The ES and Theaetetus agree that a statement must be ‘of a certain quality’: either true or false (Sph. 262E9−10, 263A12−263B3, cf. Phlb. 37B10−37C2). Plato occasionally contrasts the qualities of a thing with what it is, i.e. its essence (cf. Men. 71A1−71B8; 86D8−86E1; 87B3; Grg. 448E6−7; Tht. 152D3−4; 152D6). Therefore, by saying that truth and falsehood are qualities of statements, the ES and Theaetetus are probably hinting that neither truth nor falsehood is essential to statements as such (some statements are true and not false, others false and not true).16

Two examples are brought in: the true statement ‘Theaetetus is sitting’ and the false statement ‘Theaetetus is flying’.17 Here is the relevant passage:

ES Prose/xwmen dh\ to\n nou=n h(miÍn au)toiÍj. THT. DeiÍ gou=n.

262E11

ES Le/cw toi¿nun soi lo/gon sunqeiìj pra=gma pra/cei di' o)no/matoj kaiì r(h/matoj: oÀtou d' aÄn o( lo/goj vÅ, su/ moi fra/zein.

E15 THT. Tau=t' eÃstai kata\ du/namin. ES Qeai¿thtoj ka/qhtai. mw½n mh\ makro\j o(

lo/goj;

263A1

THT. OuÃk, a)lla\ me/trioj. ES So\n eÃrgon dh\ fra/zein periì ou t' e)stiì kaiì oÀtou.

A5

THT. Dh=lon oÀti periì e)mou= te kaiì e)mo/j. ES Ti¿ de\ oÀd' auÅ;

THT. PoiÍoj; ES Qeai¿thtoj—%Ò nu=n e)gwÜ diale/gomai—pe/tetai.

THT. Kaiì tou=ton ou)d' aÄn eiâj aÃllwj eiãpoi plh\n e)mo/n te kaiì periì e)mou=.

A10

ES Poio\n de/ ge/ tina/ famen a)nagkaiÍon eÀkaston eiånai tw½n lo/gwn.

THT. Nai¿. ES Tou/twn dh\ poiÍo/n tina e(ka/teron fate/on eiånai; THT. To\n me\n yeudh= pou, to\n de\ a)lhqh=.

263B1

ES Le/gei de\ au)tw½n o( me\n a)lhqh\j ta\ oÃnta w¨j eÃstin periì sou=.

THT. Ti¿ mh/n;

B5

ES ¸O de\ dh\ yeudh\j eÀtera tw½n oÃntwn. THT. Nai¿.

ES Ta\ mh\ oÃnt' aÃra w¨j oÃnta le/gei. THT. Sxedo/n.

B10

ES ãOntwn 18 de/ ge oÃnta eÀtera periì sou=. polla\ me\n ga\r eÃfamen oÃnta periì eÀkaston eiånai¿ pou, polla\ de\ ou)k oÃnta.

THT. Komidv= me\n ouÅn.

16 Cf. Frede (1992), 417.

17 The English phrases ‘is sitting’ and ‘is flying’ render the Greek words ‘ka/qhtai’ (263A2) and ‘pe/tetai’ (263A9) (the English words ‘sits’ and ‘flies’ convey the wrong sense).

155

ES áOn uÀsteron dh\ lo/gon eiãrhka periì sou=, prw½ton me/n, e)c wÒn w¨risa/meqa ti¿ pot' eÃsti lo/goj, a)nagkaio/taton au)to\n eÀna tw½n braxuta/twn eiånai.

263C1

THT. Nundh\ gou=n tau/tv sunwmologh/samen. ES ãEpeita de/ ge tino/j. THT. OuÀtwj.

C5

ES Ei¹ de\ mh\ eÃstin so/j, ou)k aÃllou ge ou)deno/j. THT. Pw½j ga/r;

ES Mhdeno\j <de/> ge wÔn ou)d' aÄn lo/goj eiãh to\ para/pan: a)pefh/namen ga\r oÀti tw½n a)duna/twn hÅn lo/gon oÃnta mhdeno\j eiånai lo/gon.

C10

THT. ¹Orqo/tata. ES Periì dh\ sou= lego/mena, <lego/mena> me/ntoi qa/tera

263D1

wj ta\ au)ta\ kaiì mh\ oÃnta wj oÃnta, panta/pasin eÃoiken h( toiau/th su/nqesij eÃk te r(hma/twn gignome/nh kaiì o)noma/twn oÃntwj te kaiì a)lhqw½j gi¿gnesqai lo/goj yeudh/j.

THT. ¹Alhqe/stata me\n ouÅn.

D5

ES Now let us fix our attention on ourselves. THT. We will.

262E

ES I shall make a statement to you by putting an object together with an action by means of a name and a verb. You are to tell me what the statement belongs to.

THT. I shall do my best. ES ‘Theaetetus is sitting’—not a lengthy statement, is it?

263A

THT. No, of a just length. ES Now it is for you to say what it is about and what it belongs to.

THT. Clearly it is about me and mine. ES What about this one?

THT. Which one? ES ‘Theaetetus—with whom I am now speaking—is flying’.19

THT. This one also can only be described as mine and about me. ES Besides we say that it is necessary for each of the statements to be of a certain quality.

THT. Yes. ES Of what quality then must one say each of these is?

263B

THT. One is, somehow, false, the other true. ES And the true one states the things which are as they are about you. THT. Certainly.

ES The false one instead <states> things other than the things which are <about you>. THT. Yes.

ES It therefore states the things which are not as things which are <about you>. THT. I suppose so.

18 The main MSS read ‘oÃntwj’ (cf. Robinson (1999), 159). In Crivelli (1990), 82, 93, I chose the reading ‘oÃntwj’ (also defended by Frede (1967), 58, Rijlaarsdam (1978), 209, and Szaif (1998), 475−8): the criticisms of O’Brien (1995), 128−30 and (1999), 36−7 led me to change my mind. I now find it more plausible to adopt Cornarius’s emendation ‘oÃntwn’ (cf. Cornarius (1561), 159, 194; Fischer (1771), 32), printed also by all recent eds. Note that at 240B7 one main MS (W) reads ‘oÃntwj’ where two others (B and D, the two representatives of the family β) have ‘oÃntwn’.

19 The second statement introduced as an example is described as being ‘one of the shortest’ (263C3). Hence it must consist of only one name and one verb. Therefore it cannot be the whole ‘Theaetetus with whom I am now speaking is flying’, but ‘Theaetetus is flying’. For this reason in my translation the words ‘with whom I am now speaking’ are enclosed in dashes.

156

ES But <it states> things which are other than things which are about you. For we said that about each thing there are many things which are and many which are not.

THT. By all means.

ES The statement I made about you later, in the first place, according to our definition of what statement is, must itself necessarily be one of the shortest.

THT. So we agreed just now.

263C

ES Moreover it must belong to something. THT. Yes.

ES And if it is not yours, it belongs to nothing else. THT. Certainly.

ES And if it belonged to nothing, it would not be a statement at all, for we pointed out that it was an impossibility that there could not be a statement which was a statement belonging to nothing.

THT. Quite correct.

ES So, things stated about you, but things other stated as the same, and things which are not as things which are— such a combination coming to be from verbs and names by all means seems really and truly to come to be a false statement.

THT. Perfectly true.

263D

Shortly before embarking on defining statement, the ES and Theaetetus had agreed that ‘statement has come to us thanks to the reciprocal interweaving of forms’ (259E5−6). They were thereby probably committing themselves to the claim that every statement involves the reciprocal interweaving of at least two forms. In the case of ‘Man understands’, the reciprocally interwoven forms probably include the kinds man and understanding; in the case of ‘Theaetetus is sitting’, there are perhaps exactly two reciprocally interwoven forms—namely, the kind being (which functions as a vowel-kind, or connector)20 and the kind sitting (being is perhaps also involved in ‘Man understands’).21 Near the beginning of the passage translated above, the ES says that he will make a statement ‘by putting an object together with an action by means of a name and a verb’ (262E13−14). In view of these facts, it may be plausibly inferred that if a speaker produces a primary statement by putting an object, signified by a name, together with an action, signified by a verb, the action in question is always a kind (one of a special type, like the kinds understanding, sitting, and flying),22 whereas the object in question can be anything (e.g. a kind like the kind man, signified by the name ‘man’ within ‘Man understands’, or a perceptible individual like the boy Theaetetus, signified by the name ‘Theaetetus’ within ‘Theaetetus is sitting’ and ‘Theaetetus is flying’). It may also be plausibly assumed that some of the occurrences of ‘to be about’ in the passage translated above are examples of the converse use of ‘to be’: in particular, some of them

20 Cf. above, text to n. 15 of Ch. IV.

21 Cf. Moravcsik (1962), 60; Malcolm (1967), 143−4; Ray (1984), 83−6. The kind being is not expressed by a separate word in the Greek sentences ‘aÃnqrwpoj manqa/nei’ and ‘Qeai¿thtoj ka/qhtai’, but Plato perhaps agrees with Aristotle that every finite form of any verb means the same as the phrase consisting of the corresponding finite form of ‘eiånai’ and the participle of that verb (see Arist. Int. 12, 21b9−10; APr. 1.46, 51b13−15; Ph. 1.2, 185b28−30; Metaph. ∆7, 1017a27−30). Note however that in the Euthyphro (10A5−11A6) Plato seems to distinguish between ‘fe/retai’ and ‘fero/meno/n e)stin’ (I was alerted to this point by Benjamin Morison in conversation).

22 Cf. Szaif (1998), 472.

157

are examples of the converse of the predicative use of ‘to be’ and therefore express the relation which in modern philosophical jargon is expressed by ‘to hold of ’.23

As for true statements, the above passage suggests the following account: a statement composed of a name n and a verb v is true just in case the action signified by v holds of the object signified by n. For example, ‘Theaetetus is sitting’ is true just in case sitting, the action signified by the verb ‘is sitting’, holds of Theaetetus, the object signified by the name ‘Theaetetus’.

The account of false statement is controversial. At least four different exegeses have been suggested.24

(1) According to the Oxford interpretation, a statement composed of a name n and a verb v is false just in case the action signified by v is other than everything that holds of the object signified by n. For instance, ‘Theaetetus is flying’ is false just in case flying, the action signified by the verb ‘is flying’, is other than everything that holds of Theaetetus, the object signified by the name ‘Theaetetus’.25

(2) According to the incompatibility interpretation, a statement composed of a name n and a verb v is false just in case the action signified by v is incompatible with some kind that holds of the object signified by n.26 For instance, ‘Theaetetus is flying’ is false just in case flying is incompatible with some kind that holds of Theaetetus.27

23 Cf. Frede (1967), 52−5, 94−5; Kostman (1973), 194; Frede (1992), 418; Hestir (2003), 8−10. On the converse use of ‘to be’, cf. above, text to n. 98 of Ch. IV.

24 Classifications of the interpretations of Plato’s account of false statement are also in Keyt (1973), 293−5 and Crivelli (1990), 81−91.

25 Cf. Peipers (1883), 173−7; Apelt (1897), 193; Ross (1951), 116; Peck (1962), 61−2; Runciman (1962), 115−8; Crombie (1963), 401, 497; Frede (1967), 58, 95; Owen (1971), 237−8, 260; Wiggins (1971), 294−5; Detel (1972), 104−5; Gosling (1973), 216, 217−18, 220; von Weizsäcker (1973), 237−8; Bondeson (1974), 45; Ferg (1976), 340−1; Johnson (1978), 151; McDowell (1982), 126−7; Galligan (1983), 273−4; Ray (1984), 88−92; Prior (1985), 145; de Rijk (1986), 206; Rudebusch (1990), 602−3; Frede (1992), 419, 420; N. P. White (1993), xxviii−xxix, xxx, xxxii; van Eck (1995), 39−40, 41−2; O’Brien (1999), 35−6, 38, 40; Silverman (2002), 204−5; Hestir (2003), 5.

Some commentators (cf. Heindorf (1810), 435−6; Hackforth (1945a), 57−8; Findlay (1974), 269−70; Guthrie (1978), 156; Mojsisch (1986a), 45; O’Brien (1995), 83, 121−2; Fronterotta (2007), 485−6) do not employ, at those points of their expositions where one would expect them to, an expression like ‘everything’ (or ‘anything’ or ‘all things’), but resort to formulations that mean more or less the same as ‘A statement composed of a name n and a verb v is false just in case the action signified by v is other than the things that hold of the object signified by n’. I take these commentators to subscribe to the Oxford interpretation.

26 On incompatibility cf. above, text to nn. 149−152 of Ch. IV.

27 Cf. Burnet (1914), 288−9; Taylor (1961), 67−8; Lorenz and Mittelstrass (1966), 141−4; Kostman (1973), 195−6; Seligman (1974), 110−2; Graeser (1975), 49−51; Graeser (1983), 166; Movia (1991), 457−8; Dorter (1994), 163.

Sayre (1976), 584−6 and (1983), 236−8 (cf. Matthews (1972), 26−7) proposes a refined version of the incompatibility interpretation: a statement composed of a name n and a verb v is true just in case every form holding of the object signified by n is compatible with the action signified by v; a statement composed of a name n and a verb v is false just in case every form holding of the object signified by n is compatible with the restricted complement of the action signified by v; for every form F, the restricted complement of F is a form K such that for every object x, K holds of x just in case some form incompatible with F holds of x.

158

(3) According to the quasi-incompatibility interpretation, a statement composed of a name n and a verb v is false just in case the action signified by v is other than, but in the same incompatibility range as, some kind that holds of the object signified by n. For instance, ‘Theaetetus is flying’ is false just in case flying is other than, but in the same incompatibility range as, some kind that holds of Theaetetus.28

(4) According to the extensional interpretation, a statement composed of a name n and a verb v is false just in case the object signified by n is other than everything of which the action signified by v holds. For instance, ‘Theaetetus is flying’ is false just in case Theaetetus is other than everything of which flying holds.29

The incompatibility interpretation may be ruled out because it implausibly presupposes that at some points in the Sophist the Greek word ‘eÀteron’ expresses (not otherness, as it does elsewhere in the dialogue, but) incompatibility.30 The extensional interpretation cannot be easily reconciled with the wording of the above passage (although, as I have argued elsewhere, a reconciliation is not impossible).31 The difficulties faced by the quasi-incompatibility interpretation of Plato’s account of negation32 carry over to the corresponding exegesis of his treatment of false statement. This leaves the Oxford interpretation, which moreover has the advantage of fitting well with the earlier account of not being in terms of otherness: since, as principle NIBO requires,33 for something not to be ϕ is for it to be other than everything that is ϕ (where the occurrence of ‘not to be’ and the occurrence of ‘is’ in penultimate position are occurrences of forms of ‘not to be’ and ‘to be’ in their predicative use), it follows that for flying not to be about Theaetetus is for flying to be other than everything that is about Theaetetus ( just substitute ‘about Theaetetus’ for ‘ϕ’ and apply the result to flying). Thus the Oxford interpretation of Plato’s account of falsehood is a logical consequence of an instance of principle NIBO, to which Plato was found to be committed on the basis of the extensional interpretation of his account of negation.34

Other interpretations of Plato on true and false statements. Some commentators interpret Plato in such a way as to commit him to an account of falsehood whereby a statement composed of a name n and a verb v is false just in case the action signified by v is other than the object signified by n.35 Such exegeses are implausible because the position they attribute to Plato is very poor. For,

28 Cf. Ferejohn (1989), 258−62; Szaif (1998), 489−99, 505, 507, 508−9.

29 Cf. Mignucci (1989), 275−7; Crivelli (1990), 91−5; Crivelli (1993), 73−4.

30 Cf. above, text to n. 153 of Ch. IV.

31 Cf. Crivelli (1993), 73−4.

32 Cf. above, subsection to n. 156 of Ch. IV.

33 Cf. above, text to n. 175 of Ch. IV.

34 Cf. Frede (1967), 80; Frede (1992), 420.

35 Cf. Sasso (1991), 204; Casertano (1996), 201; Glasmeyer (2003), 104.

159

since sitting, the action signified by the verb ‘is sitting’, is other than Theaetetus, the object signified by the name ‘Theaetetus’, the account of falsehood in question requires that ‘Theaetetus is sitting’ be false. But Plato explicitly acknowledges that ‘Theaetetus is sitting’ is true.

Other scholars understand Plato in such a way as to commit him to an account of falsehood whereby a statement composed of a name n and a verb v is false just in case the action signified by v is other than the condition which the object signified by n is in.36 Such exegeses run the risk of crediting Plato with the same implausible position as those mentioned in the last paragraph. Is sitting other than the condition which Theaetetus is in? If it is, then the account of falsehood we are now considering requires that ‘Theaetetus is sitting’ be false, the same unhappy consequence faced by the interpretations mentioned in the last paragraph. If instead sitting is not other than the condition Theaetetus is in, then an explanation is called for of what it is for an action to be other than the condition an object is in: after all, sitting is not identical with the condition Theaetetus is in, so one’s first inclination is to say that the former is other than the latter. The most plausible explanation is that for an action to be other than the condition an object is in is for it to be other than every kind that is a component of that condition (on this account, sitting is not other than the condition Theaetetus is in because it is not other than every kind that is a component of that condition—for sitting itself is a kind that is a component of the condition). But then, since for an action to be other than the condition an object is in is for it to be other than every kind that is a component of that condition, i.e. to be other than every kind that holds of the object, it follows that interpretations of the sort we are examining after all coincide with the Oxford interpretation.

According to yet other commentators,37 Plato does not specify necessary and sufficient conditions for falsehood. In their view, Plato is only indicating that a true statement can be made into a false one by replacing its verb with one that signifies an action that is other than the one signified by the verb replaced: e.g. the true statement ‘Theaetetus is sitting’ can be transformed into a false statement by replacing its verb, ‘is sitting’, with another one, like ‘is flying’, which signifies an action other than the one signified by the verb replaced. Plato’s aim is simply to describe a sense in which a false statement ‘states what is not’: a false statement generated by the procedure just described ‘states what is not’ because the action it states about the object it names is other than an action that is about that object. The problem with this exegesis is that the procedure described can transform a true statement not only into a false statement, but also into a true one: if one starts with the true statement ‘Theaetetus is sitting’ and replaces its verb, ‘is sitting’, with another one, ‘is breathing’, which signifies an action other than the one signified by the verb replaced, the result is another true statement, ‘Theaetetus is breathing’. And if one follows the line of thought suggested by the exegesis in question, namely that Plato’s intent is to describe a sense in which a false statement ‘states what is not’, one must accept the very awkward consequence that a true statement generated in accordance with the above procedure also ‘states

36 Cf. Palumbo (1994), 257−8.

37 Cf. Bluck (1957), 185; Rijlaarsdam (1978), 209−10; Jordan (1984), 128.

160

what is not’ because the action it states about to the object it names is other than an action that is about that object.

Five difficult remarks by the ES. Five remarks made by the ES on the statements chosen as examples (at 263B4−5, 263B7, 263B9, 263B11, and 263D1−4) are particularly important, but fraught with exegetical difficulties. One interpretative problem arises because some of these remarks contain plural noun-phrases which are constructed around participles and begin with the definite article ‘the’: these phrases are most plausibly regarded as universal quantifiers,38 but it is not immediately clear what role universal quantifiers might play at the points in question. Further exegetical difficulties are due to the opaque formulae ‘to state things as they are’ and ‘to state things as thus and such’. In the following subsections I attack these difficulties and I offer paraphrases of these remarks.

The ES’s first remark is:

[ρ1] And the true one states the things which are as they are about you. (263B4−5)

Sentence [ρ1] is my English rendering of the Greek ‘le/gei de\ au)tw½n o( me\n a)lhqh\j ta\ oÃnta w¨j eÃstin periì sou= ’ (263B4−5). I take the ‘w(j’ (263B4) to be comparative, as most translators do.39 A few translators40 regard ‘w(j’ as declarative, thereby offering something like: ‘And the true one states the things which are to be about you’. There is no substantial difference between the two translations because (as we shall see below) ‘… states the things which are as they are about you’ is logically equivalent to ‘… states about you the things which are about you to be’, which in turn is logically equivalent to ‘… states the things which are to be about you’. Several commentators prefer to take ‘w(j’ as declarative. Some support this interpretation by pointing out that shortly later ‘w(j’ has undoubtedly a declarative effect: at 263B9 we find ‘… it states the things which are not as [w¨j] things which are <about you>’ and at 263D1−4 we read ‘Things stated about you, but things other stated as [w¨j] the same, and things which are not as [w¨j] things which are—such a combination […] seems really and truly to come to be a false statement’.41 This reason is unconvincing. In fact, it might be turned on its head. For, in the

38 For the use of the article with participles in Greek cf. Smyth (1920/56), 288.

39 Cf. Ficino (1484), 364; Cornarius (1561), 159; Fowler (1921), 439; Diès (1925), 382; Martini (1931), 274; Fraccaroli (1934), 204; Cornford (1935), 310; Arangio-Ruiz (1951), 255; Taylor (1961), 175; Warrington (1961), 217; Zadro (1971), 251; Cambiano (1981), 475; Cavini (1982), 142; Roggerone (1990), 173; Mazzarelli (1991), 305; Vitali (1992), 145; Cordero (1993), 195; N. P. White (1993), 58−9; Brann, Kalkavage, and Salem (1996), 76; Bianchini (1997), 138; Giardini (1997), 613; Ambuel (2007), 240; Fronterotta (2007), 483.

40 Cf. Wagner (1856), 159; Apelt (1914), 118; Rijlaarsdam (1978), 207−10; Benardete (1984), 61; Meinhardt (1990), 175; Duerlinger (2005), 135.

41 Cf. Frede (1967), 57; Detel (1972), 100−1; Keyt (1973), 288−91; Jordan (1984), 128; Frede (1992), 418; Szaif (1998), 467.

161

sentence of 263B4−5, ‘w(j’ is construed with the indicative ‘eÃstin’ whereas in the sentences of 263B9 and 263D1−4 it is construed with the participle ‘oÃnta’ and the adjective ‘ta\ au)ta/’.42 Plato could have used the construction with the participle also at 263B4−5 in a sentence which would have (not merely allowed but) required the ‘w(j’ to have a declarative effect (‘le/gei de\ au)tw½n o( me\n a)lhqh\j ta\ oÃnta w¨j oÃnta periì sou=’).43 So, had Plato intended the ‘w(j’ in the sentence of 263B4−5 to have a declarative effect, he would have followed it up otherwise than he actually does. Therefore this ‘w(j’ is not declarative. Some commentators offer another consideration in favor of treating the ‘w(j’ in the sentence of 263B4−5 as declarative: they argue that such a treatment chimes with the account of falsehood at 240D6−241A2.44 But the parallel with 240D6−241A2 is too feeble to prove anything: for this earlier passage speaks only about falsehood (not truth) and does not contain ‘w(j’ at all. In order to adjudicate between the two ways of understanding the ‘w(j’ in the sentence of 263B4−5 and the two corresponding translations, one must look at the other passages in Plato’s dialogues where ‘w(j’ is used in formulae about truth. Some such passages (Cra. 385B7−8 and R. V 477B11−12) are ambiguous in the same way as Sph. 263B4−5; but others (Euthd. 284C7−284D7, R. V 478A7, and Prm. 161E4−5) require ‘w(j’ to be comparative.45 So, the external evidence points towards taking the ‘w(j’ in the sentence of Sph. 263B4−5 as comparative.

In [ρ1] the phrase ‘about you’ (‘periì sou=’) is probably to be construed in common with ‘states’ (‘le/gei’, cf. 263A5, 263A10−11, 263C1, and 263D1) and ‘things which are’ (‘oÃnta’, cf. 263B11−12).46 Since ‘eiånai’ prefers ‘peri¿’ and the accusative,47 ‘oÃnta periì se/’ would be standard. But ‘le/gein’ prefers ‘peri¿’ and the genitive, so that ‘le/gei periì sou=’ is standard. Since the phrase beginning with ‘peri¿’ is construed in common with ‘le/gei’ and ‘oÃnta’, one of the two constructions had to give way: the one involving the accusative did.48 Granted that in

42 Cf. Ray (1984), 128−9.

43 Cf. X. An. IV 4, 15.

44 Cf. Szaif (1998), 467; Hestir (2003), 3−4.

45 Cf. Ap. 38A7; Chrm. 161A10; Smp. 201C7; Euthd. 281E1−2; 285E9−286A3; Hp.Ma. 282A4; 284E8−9; 300B3; Hp.Mi. 367A5; Men. 78C1−2; 79A2; 96C2; 99B10; Phd. 68C4; 114D1−2; Cra. 384C8−9; 423A7; 440A5; Prm. 128E5−6; 161E4; Lg. X 893E5; Anon. Dissoi Logoi 4.2. I think that in the two ambiguous passages also ‘w(j’ is comparative: only if ‘w(j’ is comparative is the account at Cra. 385B7−8 sufficiently general (cf. Derbolav (1972), 175; Lafrance (1984), 73; Notomi (1999), 188) and the comparative reading of ‘w(j’ in R. V 477B11−12 is dictated by the parallel at 478A7.

46 Some commentators construe ‘about you’ only with ‘states’ (cf. Rijlaarsdam (1978), 209), others with ‘things which are’ and possibly also with ‘they are’ (cf. Frede (1992), 418), others with both ‘states’ and ‘they are’ (cf. Szaif (1998), 467−8), yet others with all of ‘states’, ‘things which are’, and ‘they are’ (cf. Robinson (1999), 159).

47 Cf. above, text to n. 98 of Ch. IV.

48 Cf. Robinson (1999), 159.

162

[ρ1] ‘about you’ is construed in common with ‘states’ and ‘things which are’, a first step towards a plausible paraphrase of [ρ1] is:

[∗] The true statement states about you the things which are about you as they are.

Looking at [∗] inspires the following second step towards a plausible paraphrase of [ρ1]:

[∗∗] ∀x (x is about you & the true statement mentions x → the true statement states about you x as it is).

In [∗∗] the universal quantifier ‘∀x ’ corresponds to the definite article ‘the’ within the phrase ‘the things which are about you’ occurring in [∗]. In [∗∗] the formulae ‘x is about you’ and ‘the true statement mentions x ’ express two conditions which jointly restrict the range of the universal quantifier ‘∀x ’. Although the second of these restrictive conditions is not expressed in [∗] (nor in [ρ1]), it may be plausibly supplied on the basis of the context. By way of analogy, consider the sentence ‘Ivan treats the enemies of the Revolution as they deserve’. It cannot be paraphrased by ‘∀x (x is an enemy of the Revolution → Ivan treats x as an enemy of the Revolution deserves)’. Rather, it is to be paraphrased by ‘∀x (x is an enemy of the Revolution & Ivan has to do with x → Ivan treats x as an enemy of the Revolution deserves)’, where the range of the universal quantifier ‘∀x ’ is restricted by two conditions, expressed by the formulae ‘x is an enemy of the Revolution’ and ‘Ivan has to do with x ’, and the second condition is not expressed in the paraphrased sentence.

Granted that [∗∗] is a faithful rendition of [∗], and therefore of [ρ1], its component formula ‘the true statement states about you x as it is’ calls for an explanation. Let us go back to [∗]. If a statement states about you things which are about you, it can do this in two ways: either affirmatively, in which case the statement states about you things which are about you to be, or negatively, in which case the statement states about you things which are about you not to be. But a statement states about you things which are about you as they are just in case it states about you things which are about you to be, and a statement states about you things which are about you otherwise than how they are just in case it states about you things which are about you not to be. On the basis of the first of these two biconditionals, [∗∗] may be plausibly regarded as logically equivalent to:

[π1] ∀x (x is about you & the true statement mentions x → the true statement states about you x to be).

Let [π1] be the paraphrase of [ρ1].49

49 Frede (1992), 419−20 suggests a different paraphrase of [ρ1], which in effect amounts to: ‘∀x ∀y (x is a true statement about you & x states y to be about you → y is one of the things which are about you)’. My only reservation with this paraphrase is that it speaks of all true statements about Theaetetus while [ρ1] seems to speak of one particular such statement (i.e. ‘Theaetetus is sitting’).

163

The sentences ‘The statement states about you things which are about you to be’ and ‘The statement states about you things which are about you not to be’ are instances of the schema ‘ϕ states about ψ χ to µ’, a schema where ‘ϕ’ is supposed to be replaced by an expression denoting statements (e.g. ‘the statement’), ‘ψ’ by one denoting objects (e.g. ‘you’), ‘χ’ by one denoting actions (e.g. ‘things which are about you’), and ‘µ’ by an appropriate form of a verb (e.g. ‘be’ or ‘not be’). This schema is awkward and its instances are of dubious grammaticality. But it has a virtue: it highlights that stating is stating something about something and is therefore exercised (not on a single proposition-like item, but) on two items, i.e. an action (denoted by what replaces ‘χ’) and an object (denoted by what replaces ‘ψ’). The form of a verb which goes to substitute ‘µ’ expresses the way in which the stating of an action about an object is carried out: if what is described is an affirmative predicative statement then what goes to substitute ‘µ’ is ‘be’ (because an affirmative predicative statement states about an object, e.g. you, an action, e.g. sitting, to be), while if what is described is a negative predicative statement then what goes to substitute ‘µ’ is ‘not be’ (because a negative predicative statement states about an object, e.g. you, an action, e.g. flying, not to be). The schema ‘ϕ states about ψ χ to µ’ enables us to steer clear of attributing to Plato the view that the speech-act of stating is exercised on a single proposition-like item, a view which, as we know from the aporetic arguments about not being, Plato probably rejects. An instance of this schema occurs in [π1] and others will occur later in the paraphrases of some of the ES’s later remarks about the statements chosen as examples.

The ES’s second remark is:

[ρ2] The false one instead <states> things other than the things which are <about you>. (263B7)

The expressions ‘states’ and ‘about you’ are supplied from [ρ1]. As in the case of [ρ1], ‘about you’ is to be construed in common with ‘states’ and ‘things which are’. We therefore obtain, as a first step towards a plausible paraphrase of [ρ2]:

[#] The false statement states about you things other than the things which are about you.

Looking at [#] inspires the following paraphrase of [ρ2]:

[π2] ∃x (the false statement states x about you & ∀y ( y is about you → x is other than y)).

In [π2] the universal quantifier ‘∀y ’ corresponds to the definite article ‘the’ within the phrase ‘the things which are about you’ in [#]. The particular quantifier ‘∃x ’ corresponds to the absence of the definite article ‘the’ in front of the expression ‘things other’ in [#].

The ES’s third remark is:

[ρ3] It therefore states the things which are not as things which are <about you>. (263B9)

164

As in the case of [ρ2], the phrase ‘about you’ is supplied from [ρ1] and is to be construed in common with ‘states’ and ‘things which are not’. So, if we ignore the occurrence of ‘therefore’ in second position, a first step towards a plausible paraphrase of [ρ3] is:

[+] The false statement states about you the things which are not about you as things which are.

The omitted ‘therefore’ signals that ‘things which are not about you’ is inferred from ‘things other than the things which are about you’ in [ρ2]: the inference relies on principle NIBO (the version involving the predicative uses of ‘not to be’ and ‘to be’, with ‘about you’ substituted for ‘ϕ’).50

We still need to do some work with [+]. In particular, we must still unpack the ‘as’. If we take ‘as’ to introduce the way in which the false statement states about you the things which are not about you,51 we obtain the following second step towards a plausible paraphrase of [ρ1]:

[++] The false statement states about you the things which are not about you to be.

On the basis of [++] we finally obtain the following plausible paraphrase of [ρ3]:

[π3] ∀x (x is not about you & the false statement mentions x → the false statement states about you x to be).

In [π3] the universal quantifier ‘∀x ’ corresponds to the definite article ‘the’ within the phrase ‘the things which are not about you’ occurring in [++]. In [π3] the formulae ‘x is not about you’ and ‘the false statement mentions x ’ express two conditions which jointly restrict the range of the universal quantifier ‘∀x ’. Although the second of these restrictive conditions is not expressed in [++] (nor in [ρ3]), it may be plausibly supplied on the basis of the context (as in the case of [∗∗] and [∗]).

The ES’s fourth remark is:

[ρ4] But <it states> things which are other than things which are about you. (263B11)

‘It states’ is supplied on the basis of [ρ3]. As in earlier cases, ‘about you’ is construed in common with ‘states’ and ‘things which are’. [ρ4] does not add a new thesis. Its main contribution is to clarify an obscure point of the earlier remarks by specifying that the ‘things which are’ mentioned in [ρ2] are ‘things which are about you’ (the association of ‘oÃntwn’ with ‘peri¿’ and the genitive is somewhat unusual, but is probably due to the influence of the understood ‘le/gei’).

The ES’s fifth remark is:

50 Cf. above, text to n. 33 and text to n. 175 of Ch. IV.

51 Cf. LSJ s.v. ‘w(j’ C I 2; s.v. ‘le/gw’ (B) III 2; A. A. 672; Pl. R. V 463B13−14; Sph. 263D1−2; Plt. 293E4−5; Ti. 50B3. Analogous constructions of ‘w(j’ with verbs different from ‘le/gein’ occur in the Sophist : cf. 219A5−6; 225C2−3; 238B1; 239A8−9; 249B2−3; 255B8−9; 255C1.

165

[ρ5] Things stated about you, but things other stated as the same, and things which are not as things which are—such a combination coming to be from verbs and names by all means seems really and truly to come to be a false statement. (263D1−4)

Remark [ρ5] makes two points concerning false statements about you. The first is that a false statement states about you ‘things other […] as the same’ (263D1−2), the second that a false statement states about you ‘things which are not as things which are’ (263D2). Consider the first point: a false statement states about you ‘things other […] as the same’ (263D1−2). Things other than what? They could be things other than things which are about you (cf. [ρ4]), but they could also be things other than you.

The first answer, according to which a false statement states about you things other than things which are about you as the same, faces a difficulty. For it leads one to expect that a false statement states about you things other than things which are about you as the same as things which are (‘about you’ may be omitted at the end because it would duplicate its occurrence after ‘states’). This can be made to work as an acceptable description of a false statement in only one way, namely by understanding it as a claim to the effect that a false statement states about you things other than all things which are about you as the same as some things which are. But the integration of two different quantifiers, ‘all’ and ‘some’, is unnatural.52

This difficulty faced by the first answer induces me to prefer the second one, according to which a false statement states about you things other than you as the same. If this is correct, the first point of [ρ5] describes false affirmative identity statements about you and the description offered is similar to that of the mistakes which dialectic enables one to avoid: for we were told that what is peculiar to dialectical science is ‘to divide in accordance with kinds and not to regard as other an identical form nor as identical one which is other’ (253D1−2).

The second point made by [ρ5] is that a false statement states about you ‘things which are not as things which are’ (263D2). On the basis of remarks [ρ1]−[ρ4] it may be plausibly assumed that ‘about you’ is to be supplied after ‘things which are not’, so that the statement is described as stating about you things which are not about you as being. The analysis of what it is for a kind not to be about you developed in remarks [ρ2]−[ρ4] is presupposed: in other words, it is presupposed that for a kind not to be about you is for it to be other than everything that is about you. Thus, the second point made in [ρ5] covers false affirmative predicative statements about you.

If the account of the last two paragraphs is correct, [ρ5] is not simply offering an alternative description of what goes on in the example of false statement which has been the focus of the discussion so far, i.e. ‘Theaetetus is flying’, but is a generalization covering all false affirmative statements about Theaetetus, i.e. false affirmative statements of identity about Theaetetus as well as false affirmative predicative statements about Theaetetus. The result that [ρ5] is a general description of all false affirmative statements about Theaetetus agrees with the overall tone of the remark, and especially with its second part: the words ‘such a combination coming to be from

52 Cf. Szaif (1998), 491−4.

166

verbs and names by all means seems really and truly to come to be a false statement’ suggest that what was offered in the immediately preceding lines was a description covering more than ‘Theaetetus is flying’.

Observations on the ES’s remarks. Propositions [π1] and [π3] are not definitions of truth and falsehood. Nor is it the case that [π1] is a truth condition specification for ‘Theaetetus is sitting’ and [π3] a falsehood condition specification for ‘Theaetetus is flying’. Rather, [π1] is an accurate description of what goes on from a semantic point of view in ‘Theaetetus is sitting’, one which suffices to establish that this statement is true; analogously, [π3] is an accurate description of the semantic situation of ‘Theaetetus is flying’, one that suffices to establish that this statement is false.

There is something awkward about remarks [ρ1] and [ρ3], an awkwardness that is brought to light by their being paraphrased by universally quantified sentences in which the ranges of the universal quantifiers are restricted (in part) by conditions which are not expressed in them. One wonders why Plato puts such awkward remarks in the mouth of the ES. Could he not have made his points more clearly by using slightly different formulations? Specifically, could he not have avoided [ρ1] by using instead ‘The true one states a thing which is as it is about you’, which we would then have paraphrased by ‘∃x (x is about you & the true statement states about you x to be)’? Could he not have avoided [ρ3] by using instead ‘It states a thing which is not as a thing which is about you’, which we would then have paraphrased by ‘∃x (x is not about you & the false statement states about you x to be)’?

The reason why Plato adopts [ρ1] and [ρ3], despite their awkwardness, has probably to do with the origin of the expressions that play key roles in them. The expression ‘to state the things which are as they are about you’, which plays a key role in [ρ1], is an expansion of ‘to state the things which are’, which Plato standardly uses to describe true speech; parallelwise, the expression ‘to state the things which are not as things which are about you’, which plays a key role in [ρ3], is an expansion of ‘to state the things which are not’, which Plato standardly uses to describe false speech.53 The reason why Plato uses ‘to state the things which are as they are about you’ and ‘to state the things which are not as things which are about you’ in [ρ1] and [ρ3] is probably that he wants to preserve the connection with his standard ways of describing true and false speech.

One might be inclined to go even further.54 One might claim that since the expressions which play key roles in [ρ1] and [ρ3] are there because Plato wants to preserve the connection with his standard ways of describing true and false speech, we should not analyze them as we did in earlier subsections. More precisely, one might claim that since the occurrences of the definite article ‘the’ in [ρ1] and [ρ3] which we rendered by awkward occurrences of universal quantifiers in the paraphrases [π1] and [π3] are legacies of Plato’s standard ways of describing true and false speech, we are allowed to refrain from rendering them by universal quantifiers, a rendering that sounds quite awkward.

53 For Plato’s standard ways of describing true speech and false speech cf. above, subsection to n. 65 of Ch. II.

54 Cf. Szaif (1998), 468.

167

This further step is tempting and I do not think that it can be ruled it out with complete confidence. Nevertheless, on balance, I resist it. My main reason for this resistance is that the expansion of the expressions standardly used by Plato to describe true and false speech (i.e. the expressions ‘to state the things which are’ and ‘to state the things which are not’) into the expressions which play key roles in [ρ1] and [ρ3] (i.e. the expressions ‘to state the things which are as they are about you’ and ‘to state the things which are not as things which are about you’) involves a change in syntactic structure. The new syntactic structure cannot be brushed under the carpet, but must be adequately accounted for. This is what is done by [π1] and [π3]. What the explanation of [ρ1] and [ρ3] as derived from Plato’s standard ways of describing true and false speech does provide is, as I said, a reason for Plato’s adopting [ρ1] and [ρ3] despite their awkwardness.

Falsehood and not being. The idea that to state a falsehood is to state what is not resurfaces—but, thanks to the analysis of statement and the account of not being in terms of otherness, it is now innocuous. The venom is taken out in two steps. In the first, we realize that a false statement states what is not in that it states about an object what is not about it. This first step, whose main achievement is to replace the intractable notion of ‘complete’ not being with the manageable one of ‘specific’ not being (e.g. not being about Theaetetus), is based on the analysis of statement. At last we understand why there was a need to define statement in order to realize that not being combines with it: the sophist’s last defense was not so silly as it appeared at first. What the sophist wanted was an explanation of whether and how the manageable notion of ‘specific’ not being is relevant to false statement, his hope being that the first or the second aporetic argument about not being (237B7−237E7, 238A1−238C12), which purport to show that it is impossible to say what is not, could be given another run.55 In the second step, we capitalize on the manageability of the notion of ‘specific’ not being: we realize that to state about Theaetetus what is not about him is to state about Theaetetus what is other than everything that is about him. This second step relies on the account of not being in terms of otherness.56

It also becomes clear why Plato couches his analysis of false statement in the somewhat artificial jargon of the converse of the predicative use of ‘to be’ and ‘not to be’. The reason is that he wants to retain a link between his innocuous talk of ‘stating what is not about …’ (which brings in the converse of the predicative use of ‘not to be’) and the troublesome ‘stating what is not’ (which turns on the complete use of ‘not to be’). The link is given by the fact that just as the troublesome ‘stating what is not’ may be felt as meaning ‘stating what is not the case’ or ‘stating what is not true’, so also the innocuous ‘stating what is not about …’ may be felt as meaning ‘stating what is not the case about …’ or ‘stating what is not true of …’. In other words, both in the troublesome ‘stating what is not’ and in the innocuous ‘stating what is not about …’, ‘what is not’ may be given a veridical reading, whereby it means ‘what is not the case’ or ‘what is not

55 Cf. Szaif (1998), 411.

56 Cf. N. P. White (1993), xxviii−xxix.

168

true’.57 It would not have been possible to retain such a link if, for instance, the analysis of false statement had been formulated by means of the predicative direct use of ‘not to be’: had Plato said that a false statement states what is not thus and so (e.g. flying) to be thus and so (e.g. flying), his analysis could not have been felt to rely on a veridical use of ‘not to be’ (there is no reasonable way of understanding ‘stating what is not thus and so’ as meaning ‘stating what is not the case’ or ‘stating what is not true’).

The cross-reference at 263B11−12. At 263B11−12 the ES remarks: ‘We said that about each thing [periì eÀkaston] there are many things which are and many which are not’. The cross-reference is to 256E6−7, where the ES and Theaetetus agreed that ‘about each of the forms [periì eÀkaston … tw½n ei¹dw½n] what is is a lot whereas what is not is of indefinite multitude’. The argument leading to this agreement (256D11−256E5) shows that the ground for claiming that ‘about each of the forms what is is a lot’ (256E6) is that for every form F there are many forms such that for every K among them, K is about F in that F is K because F partakes of K; and that the ground for claiming that ‘about each of the forms […] what is not is of indefinite multitude’ (256E6−7) is that for every form F there are indefinitely many forms such that for every K among them, K is not about F in that F is not K because F is other than everything that is K by being identical with K.58 Thus, the earlier passage (256E6−7) displays the converse of the identity use of ‘not to be’: ϕ is not about σ in that σ is not ϕ because σ is other than everything that is ϕ by being identical with ϕ. The account of false statement in which the later passage (263B11−12) is embedded requires the converse of the predicative use of ‘not to be’: ϕ is not about σ in that σ is not ϕ because σ is other than everything that is ϕ by partaking of ϕ. The earlier passage provides the ground for the claim that is needed by the later one. The ground is an analogical one: just as for every form F there are indefinitely many forms such that for every K among them, K is not about F in that F is not K because F is other than everything that is K by being identical with K, so also (analogical passage) for every object x there are many forms such that for every K among them, K is not about x in that x is not K because x is other than everything that is K by partaking of K. Plato expects his readers to carry out this analogical grounding on their own: in other words, they must carry out the minor modification of the earlier analysis of what it is for ϕ not to be about σ (the one involving the converse of the identity use of ‘not to be’), the modification that yields the analysis of what it is for ϕ not to be about σ which is required by the account of false statement (the one involving the converse of the predicative use of ‘not to be’). Note that we get two (logically equivalent)59 analyses of what it is for ϕ not to be about σ that fit the account of false statement: first, for ϕ not to be about σ is for ϕ to be other than everything that is about σ; second, for ϕ not to be about σ is for σ to be other than everything that partakes of ϕ. It is precisely because the second analysis is different from (albeit logically equivalent to) the first that

57 Cf. Szaif (1998), 470−1, 475.

58 Cf. above, text to n. 110 of Ch. IV.

59 For this equivalence, cf. propositions [18]−[20] of the Appendix.

169

independent evidence is provided by the claim that ‘about each thing […] there are many things which are not’ (263B11−12).

One might still wonder why at 263B11−12 Plato provides this independent evidence by appealing to a passage, 256E6−7, that displays the converse of the identity use of ‘not to be’, thereby skipping over the whole section, 257B1−258C6, where the predicative use of ‘not to be’ is discussed.60 The reason is that the section where the predicative use of ‘not to be’ is discussed, namely the section 257B1−258C6, contains no claim involving the converse of the predicative use of ‘not to be’ applied to forms: the first part of this section, 257B1−257C4, discusses the direct predicative use of ‘not to be’, while the second part, 257C5−258C6, is concerned with negative attributes and their instantiation. The closest earlier occurrence of a claim involving the converse of a use of ‘not to be’ applied to forms is at 256E6−7: here the converse of a use of ‘not to be’ applied to forms was the converse of the identity use of ‘not to be’ and, as I said, Plato trusts that his readers will be able to switch to the converse of the use of ‘not to be’ as is required by the analogical grounding of the claim that is relevant to 263B11−12.

Why a discussion of the direct predicative use of ‘not to be’? At this point the question will be probably raised of what the point is of having a discussion about the direct predicative use of ‘not to be’ and of showing that the direct predicative use of ‘not to be’ does not imply non-existence (which is what principle NIBNE states). A plausible answer to this question may be found by considering what could have happened if no discussion about the direct predicative use of ‘not to be’ had taken place and if the point that the direct predicative use of ‘not to be’ does not imply non-existence had not been secured.

First of all, suppose there had been no discussion of the direct predicative use of ‘not to be’. Then we would have remained without an explanation of the converse of the predicative use of ‘not to be’ and we would therefore have been without an account of what it is for flying not to be about Theaetetus. In other words, we would not have been in a position to say that for flying not to be about Theaetetus is for it to be other than everything that is about Theaetetus. The key here is to treat the converse of the predicative use of ‘not to be’ as a case of the direct predicative use of ‘not to be’: ‘not to be about Theaetetus’ is after all an instance of the schema ‘not to be ϕ’, just substitute ‘about Theaetetus’ for ‘ϕ’.

Secondly, suppose that the point that the direct predicative use of ‘not to be’ does not imply non-existence had not been secured. Then the proposed account of falsehood would have remained exposed to a revamped sophistic objection: namely, that since flying is not about Theaetetus, it follows that flying does not exist. The effects of this revamped objection would have been as devastating as those of the sophist’s original paradox (it would have followed, for instance, that if ‘Theaetetus is flying’ is false then it is meaningless because the action its verb should signify does not exist so that its verb fails to signify an action).61

60 Cf. McDowell (1982), 116.

61 Cf. above, text to n. 181 of Ch. IV.

170

Paradoxes that arise from confusing naming with stating. As I noted earlier, a remark made by the ES presupposes that stating and naming are different. Moreover, the ES and Theaetetus stress that the two statements offered as examples, ‘Theaetetus is sitting’ and ‘Theaetetus is flying’, are about the same object: Theaetetus. These facts are probably due to one version of the falsehood paradox being based on confusing stating with naming. Were one to confuse stating with naming, one would be likely to identify the statement ‘Theaetetus is flying’ with the complex name ‘the flying Theaetetus’ and to conclude that ‘Theaetetus is flying’ is not false, but meaningless (because the absence of a flying Theaetetus allegedly makes the complex name ‘the flying Theaetetus’ meaningless).62 Hence the point of indicating that stating and naming are different. Moreover, were statements identical with complex names, then not only would ‘Theaetetus is flying’ be identical with ‘the flying Theaetetus’, but also ‘Theaetetus is sitting’ with ‘the sitting Theaetetus’. The two statements would then be about different items (the flying Theaetetus and the sitting Theaetetus) and would therefore not contradict one another. Hence the point of emphasizing that the two statements are about the same object.

From false statement to false belief and appearance. The ES and Theaetetus define thought (dia/noia) as ‘the inner conversation of the soul with itself that occurs without voice’ (263E4−5), belief (do/ca) as the soul’s inner silent affirmation or denial that comes as a conclusion of the inner silent conversation (263E10−264A3, cf. Tht. 189E6−190A6; 196A4−7; Ti. 37A2−37C5; Phlb. 38C5−38E8),63 appearance (fantasi¿a) as the ‘blending of perception and belief ’ (264B2), and statement (lo/goj) as the ‘efflux that passes through the mouth with sound’ (263E7−8).64 Given that an account of false statement has been attained, so that statements can be both true and false, it immediately follows that beliefs and appearances also can be both true and false.

The expression ‘appearance’ (‘fantasi¿a’) is a Platonic neologism: there is no occurrence of it in Greek literature before Plato.65 Its earliest occurrence in the Sophist is at 260C9. The context of this occurrence might suggest that ‘appearance’ (‘fantasi¿a’) is merely a stylistic variant of ‘apparition’ (‘fa/ntasma’). For what Plato says is that ‘if error exists, all things are necessarily full of images [ei¹dw/lwn] and likenesses [ei¹ko/nwn] and appearance [fantasi¿aj]’ (260C8−9). Since apparitions are one type of images alongside likenesses (cf. 235B8−236C8), the association of appearance with images and likenesses suggests that ‘appearance’ is a stylistic variant of ‘apparition’. This must however be wrong: for apparitions are intrinsically false (because they are unfaithful images that present themselves as faithful) whereas appearances can be both true and false. Appearances are more like perceptual beliefs, of which some are false but others are true.

62 According to some commentators (e.g. Owen (1971), 245), Plato himself earlier succumbed to such a version of the falsehood paradox.

63 Cf. Crivelli, (1998), 20−1; Szaif (1998), 495−6.

64 A reminiscence of a definition in Gorgias’ style: cf. Men. 76C4−76E2.

65 Cf. Notomi (1999), 250.

171

What is the language of the soul’s inner silent conversation? Is it a divine language? A language of images? Or the language the speaker feels most comfortable with at the moment? No answer is forthcoming from Plato’s works.66 Another important unasked question is: can thinkers always tell, by introspecting their consciousness, what the contents of their beliefs are?67

The impact of Plato’s philosophy of language. Many of Plato’s views in philosophy of language have had a remarkable impact on later thinkers. Aristotle picked up and developed some of these views, such as the idea that truth and falsehood are qualities of statements (SE 22, 178b27−8), the distinction between primary and non-primary statements (Int. 5, 17a8−9, 17a20−4), and the analysis of primary statements into names and verbs (chapters 2−5 of de Interpretatione). The falsehood paradox never surfaced again as a serious threat: the Sophist laid it to rest. Alongside these elements of success in the area of philosophy of language, a conspicuous absence in a neighboring field should be mentioned: the study of argument and of its validity does not appear in Plato. Its development is perhaps the most notable of the many achievements of his most distinguished pupil.

66 Cf. Crivelli (1998), 21−3.

67 Cf. N. P. White (1992), 241.

APPENDIX PLATO’S THEORY OF TRUTH AND FALSEHOOD FOR PRIMARY

STATEMENTS: FORMAL PRESENTATION

Alphabet. Let me first give the alphabet of the formal language in which the formal presentation of Plato’s theory of truth and falsehood is formulated.

[1] The logical symbols of the alphabet are:

two parentheses: ‘( ’ and ‘)’; one one-argument propositional connective: ‘¬’; four two-argument propositional connectives: ‘&’, ‘∨’, ‘→’, and ‘↔’; two quantifiers: ‘∀’ and ‘∃’; one two-place relation constant: ‘=’.

These logical symbols are understood in the usual way.

[2] The descriptive symbols of the alphabet include:

ten one-place relation constants: ‘S ’, ‘Sy’, ‘Sn’, ‘Sa’, ‘Se’, ‘Si’, ‘So’, ‘K ’, ‘T ’, and ‘F ’; four two-place relation constants: ‘O ’, ‘A ’, ‘D ’, and ‘H ’; infinitely many individual variables: ‘x ’, ‘y ’, ‘z ’, ‘u ’, ‘v ’, ‘x0’, ‘x1’, ‘x2’, …; infinitely many individual parameters: ‘a ’, ‘b ’, ‘c ’, ‘d ’, ‘e ’, ‘f ’, ‘g ’, ‘a0’, ‘a1’, ‘a2’, …; infinitely many one-place relation parameters: ‘R 0

1 ’, ‘R 11 ’, ‘R 2

1 ’, …; infinitely many two-place relation parameters: ‘R 0

2 ’, ‘R 12 ’, ‘R 2

2’, …; ⋮

The meanings of the relation constants are best explained by considering them in context:

[3] Every instance of every schema in the following left-hand side column has the same meaning as the corresponding instance of the schema on the same line in the following right-hand side column:

‘Sα’ ‘α is a primary statement’ ‘Syα’ ‘α is a singular affirmative primary statement’ ‘Snα’ ‘α is a singular negative primary statement’1 ‘Saα’ ‘α is a universal affirmative primary statement’ ‘Seα’ ‘α is a universal negative primary statement’

1 The subscripts ‘y ’ and ‘n ’ abbreviate ‘yes’ and ‘no’.

173

‘Siα’ ‘α is a particular affirmative primary statement’ ‘Soα’ ‘α is a particular negative primary statement’2 ‘Kα’ ‘α is a kind’ ‘Tα’ ‘α is true’ ‘Fα’ ‘α is false’ ‘Oαβ’ ‘α is an object signified by the name in β’ ‘Aαβ’ ‘α is an action signified by the verb in β’3 ‘Dαβ’ ‘α is other than β’ 4 ‘Hαβ’ ‘α holds of β’

Formulae. Formulae are formed in the standard way. The usual conventions for economizing on parentheses are adopted (the conjunctive and the disjunctive connective, ‘&’ and ‘∨’, are taken to have more binding power than the conditional and the biconditional connective, ‘→’ and ‘↔’).

The definitions of truth and falsehood rely on seven assumptions:

[4] ∀x (Sx ↔ Sy x ∨ Sn x ∨ Sa x ∨ Se x ∨ Si x ∨ So x).

[5] ∀x ((Sy x → ¬(Sn x ∨ Sa x ∨ Se x ∨ Si x ∨ So x)) & (Sn x → ¬(Sa x ∨ Se x ∨ Si x ∨ So x)) & (Sa x → ¬(Se x ∨ Si x ∨ So x)) & (Se x → ¬(Si x ∨ So x)) & (Si x → ¬So x)).

[6] ∀x (Sx → ∃y (Ayx & ∀z (Azx → z = y) & Ky)).

[7] ∀x (Sy x ∨ Sn x → ∃y (Oyx & ∀z (Ozx → z = y))).

[8] ∀x (Sa x ∨ Se x ∨ Si x ∨ So x → ∃y (Oyx & ∀z (Ozx → z = y) & Ky)).

[9] ∀x∀y (Dxy ↔ ¬x = y).

[10] ∀x∀y (Hxy → Kx).

Here is the definition of truth, in the format of truth condition specifications for primary statements:

[11] ∀x (Sy x → (Tx ↔ ∃y ∃z (Oyx & Azx & Hzy))).

[12] ∀x (Sn x → (Tx ↔ ∃y ∃z (Oyx & Azx & ∀u (Hzu → Dyu)))).

[13] ∀x (Sa x → (Tx ↔ ∃y ∃z (Oyx & Azx & ∀u (Hyu → Hzu)))).

2 The subscripts ‘a ’, ‘e ’, ‘i ’, and ‘o ’ are the traditional symbols for (respectively) universal affirmative, universal negative, particular affirmative, and particular negative statements.

3 ‘O ’ and ‘A ’ abbreviate ‘object’ and ‘action’.

4 ‘D ’ abbreviates ‘distinct’.

174

[14] ∀x (Se x → (Tx ↔ ∃y ∃z (Oyx & Azx & ∀u (Hyu → ∀v (Hzv → Duv))))).

[15] ∀x (Si x → (Tx ↔ ∃y ∃z (Oyx & Azx & ∃u (Hyu & Hzu)))).

[16] ∀x (So x → (Tx ↔ ∃y ∃z (Oyx & Azx & ∃u (Hyu & ∀v (Hzv → Duv))))).

The definition of falsehood is straightforward:

[17] ∀x (Sx → (Fx ↔ ¬Tx)).

Some results. Here is the fundamental theorem of the present account of the truth conditions of primary statements:

[18] ∀x∀y (¬Hxy ↔ ∀u (Hxu → Dyu)).

Proof. Suppose that ¬Hab. Suppose that Hac. Suppose that b = c: then Hab, which contradicts the initial hypothesis. Therefore ¬b = c. Hence (by [9]) Dbc. But c was arbitrary, so ∀u (Hau → Dbu). Therefore ¬Hab → ∀u (Hau → Dbu). Vice versa, suppose that ∀u (Hau → Dbu). Suppose that Hab: then Dbb, so that (by [9]) ¬b = b, which is absurd. Hence ¬Hab. Therefore ∀u (Hau → Dbu) → ¬Hab. Hence ¬Hab ↔ ∀u (Hau → Dbu). But a and b were arbitrary, so ∀x∀y (¬Hxy ↔ ∀u (Hxu → Dyu)).

There are two other interesting results:

[19] ∀x∀y (¬Hxy ↔ ∀u (Huy → Dxu)).

Proof. Suppose that ¬Hab. Suppose that Hcb. Suppose that a = c: then Hab, which contradicts the initial hypothesis. Therefore ¬a = c. Hence (by [9]) Dac. But c was arbitrary, so ∀u (Hub → Dau). Therefore ¬Hab → ∀u (Hub → Dau). Vice versa, suppose that ∀u (Hub → Dau). Suppose that Hab: then Daa, so that (by [9]) ¬a = a, which is absurd. Hence ¬Hab. Therefore ∀u (Hub → Dau) → ¬Hab. Hence ¬Hab ↔ ∀u (Hub → Dau). But a and b were arbitrary, so ∀x∀y (¬Hxy ↔ ∀u (Huy → Dxu)).

[20] ∀x∀y (∀u (Hxu → Dyu) ↔ ∀u (Huy → Dxu)).

Proof. A trivial consequence of [18] and [19].

The next proposition is a formulation of the falsehood conditions for singular affirmative primary statements. It matches Plato’s account of ‘Theaetetus is flying’:

[21] ∀x (Sy x → (Fx ↔ ∃y ∃z (Oyx & Azx & ∀u (Huy → Dzu)))).

Proof. Suppose that Sy a. Then (by [4]) Sa, so that (by [6]) ∃y (Aya & ∀z (Aza → z = y) & Ky). Let then b be such that Aba & ∀z (Aza → z = b) & Kb. Since Sy a, it follows (by [7]) that ∃y (Oya & ∀z (Oza → z = y)). Let then c be such that Oca & ∀z (Oza → z = c). Since Sy a, it

175

follows (by [11]) that Ta ↔ ∃y ∃z (Oya & Aza & Hzy). Since Sa, it follows (by [17]) that Fa ↔ ¬Ta.

Suppose that Fa. Then ¬Ta, so that ¬∃y ∃z (Oya & Aza & Hzy). Suppose that Hbc: then Oca & Aba & Hbc, so that ∃y ∃z (Oya & Aza & Hzy), contrary to an earlier result. Therefore ¬Hbc. Hence (by [19]) ∀u (Huc → Dbu). Therefore Oca & Aba & ∀u (Huc → Dbu), so that ∃y ∃z (Oya & Aza & ∀u (Huy → Dzu)). So Fa → ∃y ∃z (Oya & Aza & ∀u (Huy → Dzu)).

Vice versa, suppose that ∃y ∃z (Oya & Aza & ∀u (Huy → Dzu)). Suppose that ∃y ∃z (Oya & Aza & Hzy). Since ∃y ∃z (Oya & Aza & ∀u (Huy → Dzu)), let d and e be such that Oda & Aea & ∀u (Hud → Deu). Since ∃y ∃z (Oya & Aza & Hzy), let f and g be such that Ofa & Aga & Hgf. Since ∀u (Hud → Deu), it follows (by [19]) that ¬Hed. Since Aga and ∀z (Aza → z = b), g = b. Since Aea and ∀z (Aza → z = b), e = b. Since g = b and e = b, g = e. Since Ofa and ∀z (Oza → z = c), f = c. Since Oda and ∀z (Oza → z = c), d = c. Since f = c and d = c, f = d. Since Hgf and g = e and f = d, it follows that Hed, contrary to an earlier result. Therefore ¬∃y ∃z (Oya & Aza & Hzy), so that ¬Ta, whence Fa. So ∃y ∃z (Oya & Aza & ∀u (Huy → Dzu)) → Fa.

Therefore Fa ↔ ∃y ∃z (Oya & Aza & ∀u (Huy → Dzu)). But a was arbitrary, so ∀x (Sy x → (Fx ↔ ∃y ∃z (Oyx & Azx & ∀u (Huy → Dzu)))).

176

ReferencesReferencesReferencesReferences

AA. VV. 1953, Actes du XIème Congrès International de Philosophie (Bruxelles, 20−26 Août 1953), XII, Amsterdam and Louvain.

AA. VV. (trs.) 1971, Platone: Opere complete, Rome and Bari. Ackrill, J. L. 1955, ‘SUMPLOKH EIDWN’ = Ackrill (1997), 72−9. —— 1957, ‘Plato and the Copula: Sophist 251−259’ = Ackrill (1997), 80−92. —— 1964, ‘Demos on Plato’, Journal of Philosophy 61: 610−13. —— 1970, ‘In Defence of Platonic Division’ = Ackrill (1997), 93−109. —— 1997, Essays on Plato and Aristotle, Oxford. Ambuel, D. 2007, Image and Paradigm in Plato’s Sophist, Las Vegas, Zürich, and Athens. Anton, J. P. and Preus, A. (eds.) 1983, Essays in Ancient Greek Philosophy, II, Albany. Apelt, O. 1891a, ‘Die Ideenlehre in Platos Sophistes’ = Apelt (1891b), 67−99. —— 1891b, Beiträge zur Geschichte der griechischen Philosophie, Leipzig. Apelt, O. (ed. and comm.) 1897, Platonis Sophista, Leipzig. —— (trans. and comm.) 1914, Platons Dialog Sophistes, Leipzig. Arangio-Ruiz, V. (trans. and comm.) 1951, Platone: Sofista, Bari. Ast, F. 1835−38, Lexicon Platonicum sive Vocum Platonicarum Index, Leipzig. Aubenque, P. 1991, ‘Une occasion manquée: la genèse avortée de la distinction entre l’ “étant” et le “quelque chose” ’

= Aubenque and Narcy (1991), 365−85. Aubenque, P. and Narcy, M. (eds.) 1991, Études sur le Sophiste de Platon, Rome. Bambrough, R. (ed.) 1965, New Essays on Plato and Aristotle, London. Bekker, I. (ed.) 1814−21, Anecdota Graeca, Berlin. —— (ed.) 1826, Platonis Scripta Graece Omnia, London. Benardete, S. (trans. and comm.) 1984, Plato’s Sophist, Chicago and London. Berger, F. 1965, ‘Rest and Motion in the Sophist ’, Phronesis 10: 70−7. Bianchini, B. (trans.) 1997, Platone: Sofista, Rome. Blackburn, S. and Simmons, K. (eds.) 1999, Truth, Oxford. Bluck, R. S. 1957, ‘False Statement in the Sophist ’, Journal of Hellenic Studies 77: 181−6. —— 1963, Plato’s Sophist: A Commentary, ed. by G. C. Neal, Manchester 1975. Bondeson, W. 1972, ‘Plato’s Sophist : Falsehood and Images’, Apeiron 6: 1−6. —— 1974, ‘Plato’s Sophist and the Significance and Truth-Value of Statements’, Apeiron 8: 41−7. —— 1976, ‘Some Problems about Being and Predication in Plato’s Sophist 242−249’, Journal of the History of

Philosophy 14: 1−10. Bordt, S. J., M. 1991, ‘Der Seinsbegriff in Platons “Sophistes”: Eine Untersuchung zu 242b6−249d5’, Theologie und

Philosophie 66: 493−529. Bosanquet, B. 1888/1911, Logic or the Morphology of Knowledge, 2nd edn, Oxford. Bostock, D. 1984, ‘Plato on “Is Not” (Sophist 254−9)’, Oxford Studies in Ancient Philosophy 2: 89−119. Brann, E., Kalkavage, P., and Salem, E. (trans.) 1996, Plato’s Sophist or the Professor of Wisdom, Newburyport, MA. Brown, L. 1986/99, ‘Being in the Sophist: A Syntactical Enquiry’ = Fine (1999), 455−78. —— 1994, ‘The Verb “To Be” in Greek Philosophy: Some Remarks’ = Everson (1994), 212−36. —— 1998, ‘Innovation and Continuity: The Battle of Gods and Giants, Sophist 245−249’ = Gentzler (1998),

181−207. Brunschwig, J. 1988, ‘The Stoic Theory of the Supreme Genus and Platonic Ontology’ = Brunschwig (1994),

92−157. —— 1994, Papers in Hellenistic Philosophy, trans. by J. Lloyd, Cambridge. Burnet, J. 1914, Greek Philosophy: Thales to Plato, London. Burnet, J. (ed.) 1900−7, Platonis Opera, Oxford. Burnyeat, M. F. 2002, ‘Plato on How Not to Speak of What Is Not: Euthydemus 283a−288a’ = Canto-Sperber and

Pellegrin (2002), 40−66. —— 2003, ‘Apology 30B2−4: Socrates, Money, and the Grammar of GIGNESQAI’, Journal of Hellenic Studies 123:

1−25. Büttgen, P., Diebler, S., and Rashed, M. (eds.) 1999, Théories de la phrase et de la proposition de Platon à Averroès,

Paris. Cambiano, G. (trans.) 1981, Platone: Dialoghi filosofici, II, Turin. Campbell, L. (ed. and comm.) 1867, The Sophistes and Politicus of Plato, Oxford.

177

Canto-Sperber, M. and Pellegrin, P. (eds.) 2002, Le style de la pensée: Recueil de textes en hommage à Jacques Brunschwig, Paris.

Carchia, G. 1997, La favola dell’essere: Commento al Sofista, Macerata. Casertano, G. 1996, Il nome della cosa: Linguaggio e realtà negli ultimi dialoghi di Platone, Naples. Cavini, W. (trans. and comm.) 1982, La dialettica, Florence. Cavini, W. 1995, ‘Naming and Argument: Diaeretic Logic in Plato’s Statesman’ = Rowe (1995), 123−38. Charlton, W. 1995, ‘Plato’s Later Platonism’, Oxford Studies in Ancient Philosophy 13: 113−33. Cherniss, H. 1957, ‘The Relation of the Timaeus to Plato’s Later Dialogues’, American Journal of Philology 78:

225−66. Cordero, N.-L. 1986−87, ‘Le non-être absolu dans le Sophiste de Platon’, Annuaire de l’École Pratique des Hautes

Études, Ve Section 95: 282−5. Cordero, N.-L. (trans. and comm.) 1993, Platon: Le Sophiste, Paris. Cornarius, I. (trans.) 1561, Platonis … opera … <et> Eclogae decem, Basel. Cornford, F. M. 1935, Plato’s Theory of Knowledge: The Theaetetus and the Sophist of Plato Translated with a

Running Commentary, London and New York. Crivelli, P. 1990, ‘Il Sofista di Platone: non essere, negazione e falsità’, Atti e Memorie dell’Accademia Toscana di

Scienze e Lettere ‘La Colombaria’ 55: 9−104. —— 1993, ‘Plato’s Sophist and Semantic Fragmentation’, Archiv für Geschichte der Philosophie 75: 71−4. —— 1998, ‘ALLODOCIA’, Archiv für Geschichte der Philosophie 80: 1−29. Crombie, I. M. 1963, An Examination of Plato’s Doctrines, II, 2nd edn, London 1967. Dancy, R. M. 1999, ‘The Categories of Being in Plato’s Sophist 255c-e’, Ancient Philosophy 19: 45−72. Dascal, M., Gerhardus, D., Lorenz, K., and Meggle, G. (eds.) 1992, Sprachphilosophie–Philosophy of Language–La

philosophie du langage, I, Berlin and New York. Daues, S. J., V., Holloway, S. J., M. R., and Sweeney, S. J., L. (eds.) 1966, Wisdom in Depth. Essays in Honor of Henri

Renard, S. J., Milwaukee. Davies, M. 1981, Meaning, Quantification, Necessity: Themes in Philosophical Logic, London, Boston, and Henley. Denyer, N. 1991, Language, Thought and Falsehood in Ancient Greek Philosophy, London and New York. Derbolav, J. 1972, Platons Sprachphilosophie im Kratylos und in den späteren Schriften, Darmstadt. de Rijk, L. M. 1986, Plato’s Sophist: A Philosophical Commentary, Amsterdam, Oxford, and New York. des Places, É. 1970, Lexique de la langue philosophique et religieuse de Platon, Paris. Detel, W. 1972, Platons Beschreibung des falschen Satzes im Theätet und Sophistes, Göttingen. De Vogel, C. J. 1953, ‘Platon a-t-il ou n’a-t-il pas introduit le mouvement dans son monde intelligible? (Critique des

interprétations modernes de Soph. 249a et de Tim. 31b)’ = AA. VV. (1953), 61−7. de Vries, W. 1988, ‘On “Sophist” 255B−E’, History of Philosophy Quarterly 5: 385−94. Diès, A. (ed. and trans.) 1925, Platon: Le Sophiste, Paris. Dixsaut, M. 1991, ‘La négation, le non-être et l’autre dans le Sophiste’ = Aubenque and Narcy (1991), 165−213. Dorter, K. 1994, Form and Good in Plato’s Eleatic Dialogues: The Parmenides, Theaetetus, Sophist, and Statesman,

Berkeley, Los Angeles, and London. Duerlinger, J. (trans.) 2005, Plato’s Sophist: A Translation with a Detailed Account of Its Theses and Arguments, New

York. Duke, E. A., Hicken, W. F., Nicoll, W. S. M., Robinson, D. B., and Strachan, J. C. G. (eds.) 1995, Platonis Opera, I,

Oxford. Dürr, K. 1945, ‘Moderne Darstellung der platonischen Logik: Ein Beitrag zur Erklärung des Dialoges “Sophistes” ’,

Museum Helveticum 2: 166−94. Edwards, P. (ed.) 1967, The Encyclopedia of Philosophy, VI, New York and London. Everson, S. (ed.) 1994, Language, Cambridge. Ferejohn, M. T. 1989, ‘Plato and Aristotle on Negative Predication and Semantic Fragmentation’, Archiv für

Geschichte der Philosophie 71: 257−82. Ferg, S. 1976, ‘Plato on False Statement: Relative Being, A Part of Being, and Not-Being in the Sophist’, Journal of

the History of Philosophy 14: 336−42. Ficino, M. 1484, Platonis Sophista = Bekker (1826), X 340−67. Findlay, J. N. 1974, Plato: The Written and Unwritten Doctrines, London. Fine, G. 1993, On Ideas: Aristotle’s Criticism of Plato’s Theory of Forms, Oxford. Fine, G. (ed.) 1999, Plato 1: Metaphysics and Epistemology, Oxford. Fischer, J. F. (ed.) 1771, Iani Cornarii Eclogae in Dialogos Platonis Omnes, Leipzig. Fowler, H. N. (ed. and trans.) 1921, Plato: Theaetetus, Sophist, London and New York. Fraccaroli, G. (trans. and comm.) 1934, Platone: Il Sofista e l’Uomo politico, Florence.

178

Frank, D. H. 1985, ‘On What There Is: Plato’s Later Thoughts’, Elenchos 6: 5−18. Frede, M. 1967, Prädikation und Existenzaussage: Platons Gebrauch von ‘… ist …’ und ‘… ist nicht …’ im Sophistes,

Göttingen. —— 1992, ‘Plato’s Sophist on False Statements’ = Kraut (1992), 397−424. —— 1996, ‘Die Frage nach dem Seienden: Sophistes’ = Kobusch and Mojsisch (1996), 181−99. Frege, G. 1892, ‘Über Begriff und Gegenstand’, Vierteljahrsschrift für Wissenschaftliche Philosophie 16: 192-205. —— 1918−19, ‘Die Verneinung: Eine logische Untersuchung’, Beiträge zur Philosophie des deutschen Idealismus 1:

143−57. Fronterotta, F. (trans. and comm.) 2007, Platone: Sofista, Milan. Galligan, E. M. 1983, ‘Logos in the Theaetetus and the Sophist’ = Anton and Preus (1983), 264−78. Gentile, M. (trans.) and Plebe, A. (comm.) 1965, Platone: Il Sofista, Florence. Gentzler, J. (ed.) 1998, Method in Ancient Philosophy, Oxford. Gerson, L. P. 2006, ‘The “Holy Solemnity” of Forms and the Platonic Interpretation of Sophist ’, Ancient Philosophy

26: 291−304. Giardini, G. (trans.) 1997, Platone: Tutte le opere, I, Rome. Glasmeyer, C. 2003, Platons Sophistes: Zur Überwindung der Sophistik, Heidelberg. Gómez-Lobo, A. 1977, ‘Plato’s Description of Dialectic in the Sophist 253d1−e2’, Phronesis 22: 29−47. —— 1981, ‘Dialectic in the Sophist: A Reply to Waletzki’, Phronesis 26: 80−3. Gosling, J. C. B. 1973, Plato, London. Graeser, A. 1975, Platons Ideenlehre – Sprache, Logik und Metaphysik: Eine Einführung, Bern and Stuttgart. —— 1983, Sophistik und Sokratik, Plato und Aristoteles, München. Grote, G. 1875, Plato, and the Other Companions of Sokrates, 3rd edn, London. Gulley, N. 1962, Plato’s Theory of Knowledge, London. Guthrie, W. K. C. 1978, A History of Greek Philosophy, V, Cambridge. Hackforth, R. 1945a, ‘False Statement in Plato’s Sophist’, Classical Quarterly 39: 56−8. —— 1945b, Plato’s Examination of Pleasure, Cambridge. Hamlyn, D. W. 1955, ‘The Communion of Forms and the Development of Plato’s Logic’, Philosophical Quarterly 5:

289−302. Harte, V. 2002, Plato on Parts and Wholes: The Metaphysics of Structure, Oxford. Heinaman, R. 1981, ‘Self-Predication in the Sophist’, Phronesis 26: 55−66. —— 1982−83, ‘Communion of Forms’, Proceedings of the Aristotelian Society n.s. 83: 175−90. —— 1983, ‘Being in the Sophist’, Archiv für Geschichte der Philosophie 65: 1−17. Heindorf, L. F. (ed. and comm.) 1810, Platonis Dialogi Selecti, IV, Berlin. Hermann, K. F. (ed.) 1851, Platonis dialogi secundum Thrasylli tetralogias dispositi, I, Leipzig. Herrmann, F.-G. (ed.) 2006, New Essays on Plato: Language and Thought in Fourth-Century Greek Philosophy,

Swansea. Hestir, B. E. 2003, ‘A “Conception” of Truth in Plato’s Sophist’, Journal of the History of Philosophy 41: 1−24. Hicken, W. F. 1957, ‘Knowledge and Forms in Plato’s Theaetetus’, Journal of Hellenic Studies 77: 48−53. Johnson, P. A. 1978, ‘Keyt on eÀteron in the Sophist’, Phronesis 23: 151−7. Jordan, R. W. 1984, ‘Plato’s Task in the Sophist’, Classical Quarterly n.s. 34: 113−29. Jowett, B. (trans.) 1892, The Dialogues of Plato, 3rd edn, Oxford. Kahn, C. H. 2004, ‘A Return to the Theory of the Verb be and the Concept of Being’, Ancient Philosophy 24:

381−405. Ketchum, R. J. 1978, ‘Participation and Predication in Sophist 251−260’, Phronesis 23: 42−62. Keyt, D. 1969, ‘Plato’s Paradox that the Immutable is Unknowable’, Philosophical Quarterly 19: 1−14. —— 1973, ‘Plato on Falsity: Sophist 263B ’ = Lee, Mourelatos, and Rorty (1973), 285−305. —— 1994, rev. of Pelletier (1990), Noûs 28: 117−19. Kobusch, T. and Mojsisch, B. (eds.) 1996, Platon: Seine Dialoge in der Sicht neuer Forschungen, Darmstadt. Kostman, J. P. 1973, ‘False Logos and Not-Being in Plato’s Sophist’ = Moravcsik (1973b), 192−212. —— 1989, ‘The Ambiguity of “Partaking” in Plato’s Sophist’, Journal of the History of Philosophy 27: 343−63. Kraut, R. (ed.) 1992, The Cambridge Companion to Plato, Cambridge. Kühner, R. and Gerth, B. 1892−1904, Ausführliche Grammatik der Griechischen Sprache, 3rd edn, Hannover and

Leipzig. Künne, W. 2004, ‘Die “Gigantomachie” in Platons Sophistes. Versuch einer analytischen Rekonstrution’, Archiv für

Geschichte der Philosophie 86: 307−21. Lacey, A. R. 1959, ‘Plato’s Sophist and the Forms’, Classical Quarterly n.s. 9: 43−52.

179

Lafrance, Y. 1984, ‘Sur une lecture analytique des arguments concernant le non-être (Sophiste, 237b10−239a12)’, Revue de Philosophie Ancienne 2: 41−76.

Lauer, S. J., Q. 1966, ‘The Being of Nonbeing in Plato’s Sophist’ = Daues, Holloway, and Sweeney (1966), 141−56. Lee, E. N. 1972, ‘Plato on Negation and Not-Being in the Sophist ’, Philosophical Review 81: 267−304. Lee, E. N., Mourelatos, A. P. D., and Rorty, R. M. (eds.) 1973, Exegesis and Argument: Studies in Greek Philosophy

Presented to Gregory Vlastos, Assen. Lentz, W. 1997, ‘The Problem of Motion in the Sophist ’, Apeiron 30: 89−108. Lewis, F. A. 1975, ‘Did Plato Discover the Estin of Identity?’, California Studies in Classical Antiquity 8: 113−43. —— 1976, ‘Plato on “Not” ’, California Studies in Classical Antiquity 9: 89−115. Lorenz, K. and Mittelstrass, J. 1966, ‘Theaitetos fliegt: Zur Theorie wahrer und falscher Sätze bei Platon’, Archiv für

Geschichte der Philosophie 48: 113−52. MacKay, R. W. (trans.) 1868, The Sophistes of Plato: A Dialogue on True and False Teaching, London and Edinburgh. Madvig, J. N. 1871, Adversaria critica ad sriptores graecos, Copenhagen. Malcolm, J. 1967, ‘Plato’s Analysis of to\ oÃn and to\ mh\ oÃn in the Sophist’, Phronesis 12: 130−46. —— 1983, ‘Does Plato Revise His Ontology in Sophist 246c−249d?’, Archiv für Geschichte der Philosophie 65:

115−27. —— 1985a, ‘Vlastos on Pauline Predication’, Phronesis 30: 79−91. —— 1985b, ‘On What Is Not in any Way in the Sophist ’, Classical Quarterly n.s. 35: 521−3. —— 2006a, ‘Some Cautionary Remarks on the “is”/“teaches” Analogy’, Oxford Studies in Ancient Philosophy 31:

281−96. —— 2006b, ‘A Way Back for Sophist 255c12−13’, Ancient Philosophy 26: 275−89. Marcos de Pinotti, G. E. 1994, ‘Négation, fausseté et non-être dans le Sophiste’, Revue de Philosophie Ancienne 12:

153−70. Martini, E. (trans.) 1931, Platone: Il sofista = Martini (1974), 225−78. —— (1974), Platone: Tutte le opere, ed. by G. Pugliese Carratelli, Florence. Mates, B. 1979, ‘Identity and Predication in Plato’, Phronesis 24: 211−29. Matthews, G. 1972, Plato’s Epistemology and Related Logical Problems, London. Mazzarelli, C. (trans.) 1991, Platone: Sofista = Reale (1991), 261−314. McDowell, J. 1982, ‘Falsehood and Not-Being in Plato’s Sophist’ = Schofield and Craven Nussbaum (1982),

115−34. McPherran, M. L. 1986, ‘Plato’s Reply to the “Worst Difficulty” Argument of the Parmenides: Sophist 248a−249d’,

Archiv für Geschichte der Philosophie 68: 233−52. Meinhardt, H. 1968, Teilhabe bei Platon: Ein Beitrag zum Verständnis platonischen Prinzipiendenkens unter besonderer

Berücksichtigung des ‘Sophistes’ , Freiburg and München. Meinhardt, H. (trans. and comm.) 1990, Platon: Der Sophist, Stuttgart. Mignucci, M. 1989, ‘Esistenza e verità nel Sofista di Platone’, Atti dell’Accademia di Scienze Morali e Politiche –

Società Nazionale di Scienze, Lettere e Arti in Napoli 100: 267−81. Mohr, R. D. 1982, ‘The Relation of Reason to Soul in the Platonic Cosmology: Sophist 248e−249d’ = Mohr (2005),

189−95. —— 1985, ‘Plato on Time and Eternity’ = Mohr (2005), 51−80. —— 2005, God & Forms in Plato, USA. Mojsisch, B. 1986a, ‘Platons Sprachphilosophie im “Sophistes” ’ = Mojsisch (1986b), 35−62. Mojsisch, B. (ed.) 1986b, Sprachphilosophie in Antike und Mittelalter (Bochumer Kolloquium, 2.−4. Juni 1982),

Amsterdam. Moravcsik, J. M. E. 1962, ‘Being and Meaning in the Sophist’, Acta Philosophica Fennica 14: 23−78. —— 1973a, ‘Plato’s Method of Division’ = Moravcsik (1973b), 158−80. —— 1992, Plato and Platonism: Plato’s Conception of Appearance and Reality in Ontology, Epistemology, and Ethics,

and Its Modern Echoes, Oxford and Cambridge, MA. Moravcsik, J. M. E. (ed.) 1973b, Patterns in Plato’s Thought, Dordrecht. Movia, G. 1991, Apparenze, essere e verità: Commentario storico-filosofico al “Sofista” di Platone, Milan. Natali, C. 1995, ‘Commenti su Cavini’ = Rowe (1995), 139−40. Notomi, N. 1999, The Unity of Plato’s Sophist: Between the Sophist and the Philosopher, Cambridge. Nuchelmans, G. 1973, Theories of the Proposition: Ancient and Medieval Conceptions of the Bearers of Truth and

Falsity, Amsterdam and London. O’Brien, D. 1995, Le non-être: Deux études sur le Sophiste de Platon, Sankt Augustin. —— 1999, ‘Théories de la proposition dans le Sophiste de Platon’ = Büttgen, Diebler, and Rashed (1999), 21−41. Owen, G. E. L. 1953, ‘The Place of the Timaeus in Plato’s Dialogues’, Classical Quarterly n.s. 3: 79−95.

180

—— 1957, ‘A Proof in the periì i¹dew½n’, Journal of Hellenic Studies 77: 103−11. —— 1965, ‘Aristotle on the Snares of Ontology’ = Bambrough (1965), 69−95. —— 1966, ‘Plato and Parmenides on the Timeless Present’, The Monist 50: 317−40. —— 1968a, ‘Dialectic and Eristic in the Treatment of the Forms’ = Owen (1968b), 103−25. —— 1971, ‘Plato on Not-Being’ = Vlastos (1971), 223−67. Owen, G. E. L. (ed.) 1968b, Aristotle on Dialectic: The Topics—Papers of the Third Symposium Aristotelicum, Oxford. Palumbo, L. 1994, Il non essere e l’apparenza: Sul Sofista di Platone, Naples. Peck, A. L. 1952, ‘Plato and MEGISTA GENH of the Sophist: A Reinterpretation’, Classical Quarterly n.s. 2: 32−56. —— 1962, ‘Plato’s Sophist: the sumplokh\ tw½n ei¹dw½n’, Phronesis 7: 46−66. Peipers, D. 1883, Ontologia Platonica: Ad Notionum Terminorumque Historiam Symbola, Leipzig. Pelletier, F. J. 1975, ‘“Incompatibility” in Plato’s Sophist’, Dialogue 14: 143−6. —— 1983, ‘Plato on Not-Being: Some Interpretations of the SUMPLOKH EIDWN (259E) and Their Relation to

Parmenides’ Problem’, Midwest Studies in Philosophy 8: 35−65. —— 1990, Parmenides, Plato, and the Semantics of Not-Being, Chicago and London. Philip, J. A. 1966, ‘Platonic Diairesis’, Transactions and Proceedings of the American Philological Association 97:

335−58. —— 1968, ‘False Statement in the Sophistes’, Transactions and Proceedings of the American Philological Association 99:

315−27. Politis, V. 2006, ‘The Argument for the Reality of Change and Changelessness in Plato’s Sophist (248e7−249d5)’ =

Herrmann (2006), 149−75. Prior, W. J. 1980, ‘Plato’s Analysis of Being and Not-Being in the Sophist’, Southern Journal of Philosophy 18:

199−211. —— 1985, Unity and Development in Plato’s Metaphysics, La Salle, IL. Przełęcki, M. 1981, ‘On What There Is Not’, Dialectics and Humanism 8: 123−9. Quine, W. V. O. 1950, Methods of Logic, 3rd edn, London 1974. Ray, A. C. 1984, For Images: An Interpretation of Plato’s Sophist, Lanham and London. Reale, G. (ed.) 1991, Platone: Tutti gli scritti, Milan. Reeve, C. D. C. 1985, ‘Motion, Rest, and Dialectic in the Sophist’, Archiv für Geschichte der Philosophie 67: 47−64. Rijlaarsdam, J. C. 1978, Platon über die Sprache: Ein Kommentar zum Kratylos, Utrecht. Ritter, C. 1910, Neue Untersuchungen über Platon, München. Roberts, J. 1986, ‘The Problem about Being in the Sophist’, History of Philosophy Quarterly 3: 229−43. Robinson, D. B. 1999, ‘Textual Notes on Plato’s Sophist’, Classical Quarterly n.s. 49: 139−59. —— 2001, ‘The Phantom of the Sophist: to\ ou)k oÃntwj ou)k oÃn (240a−c)’, Classical Quarterly n.s. 51: 435−57. Roggerone, G. A. (trans. and comm.) 1990, I dialoghi platonici del Forestiero di Elea: Sofista e Politico, Settimo

Milanese. Rosen, S. 1983, Plato’s Sophist: The Drama of Original and Image, New Haven and London. Ross, D. 1951, Plato’s Theory of Ideas, Oxford. Rowe, C. J. (ed.) 1995, Reading the Statesman: Proceedings of the III Symposium Platonicum, Sankt Augustin. Rudebusch, G. 1990, ‘Does Plato Think False Speech Is Speech?’, Noûs 24: 599−609. Runciman, W. G. 1962, Plato’s Later Epistemology, Cambridge. Russell, B. 1903, The Principles of Mathematics, 2nd edn, London 1937. —— 1910a, Philosophical Essays, 2nd edn, London 1966. —— 1910b, ‘On the Nature of Truth and Falsehood’ = Russell (1910a), 147−59. —— 1912, The Problems of Philosophy, 2nd edn, Oxford 1967. Ryle, G. 1960, ‘Letters and Syllables in Plato’, Philosophical Review 69: 431−51. —— 1967, ‘Plato’ = Edwards (1967), 314−33. Sasso, G. 1991, L’essere e le differenze: Sul ‘Sofista’ di Platone, Bologna. Sayre, K. M. 1976, ‘Sophist 263B Revisited’, Mind n.s. 85: 581−6. —— 1983, Plato’s Late Ontology: A Riddle Resolved, Princeton. Schanz, M. (ed.) 1887, Platonis … Sophista, Leipzig. Schleiermacher, F. (trans.) 1807, Platons Werke, II 2, Berlin. —— (trans.) 1857, Platons Werke, II 2, 3rd edn, Berlin. Schofield, M. and Craven Nussbaum, M. (eds.) 1982, Language and Logos: Studies in Ancient Greek Philosophy

Presented to G. E. L. Owen, Cambridge. Seligman, P. 1974, Being and Not-Being: An Introduction to Plato’s Sophist, The Hague. Silverman, A. 2002, The Dialectic of Essence: A Study of Plato’s Metaphysics, Princeton. Slings, S. R. (ed.) 2003, Platonis Rempublicam, Oxford.

181

Smyth, H. W. 1920/56, Greek Grammar, rev. by G. M. Messing, Cambridge, MA. Stallbaum, G. (ed. and comm.) 1840, Platonis Sophista, Gotha. Stough, C. 1990, ‘Two Kinds of Naming in the Sophist’, Canadian Journal of Philosophy 20: 355−81. Swiggers, P. 1984, ‘Théorie grammaticale et définition du discours dans le Sophiste de Platon’, Les études classiques 52:

15−7. Szaif, J. 1998, Platons Begriff der Wahrheit, 2nd edn, München. Tarski, A. 1944, ‘The Semantic Conception of Truth and the Foundations of Semantics’ = Blackburn and Simmons

(1999), 115–143. Taylor, A. E. 1926, Plato: The Man and His Work, London. Taylor, A. E. (trans.) 1961, Plato: The Sophist and the Statesman, ed. by R. Klibansky and E. Anscombe, London. Teloh, H. 1981, The Development of Plato’s Metaphysics, University Park and London. Thorp, J. 1984, ‘Forms, Concepts and TO MH ON’, Revue de Philosophie Ancienne 2: 77−92. Toms, E. 1962, Being, Negation and Logic, Oxford. Trevaskis, J. R. 1966, ‘The me/gista ge/nh and the Vowel Analogy of Plato’, Phronesis 11: 99−116. —— 1967, ‘Division and Its Relation to Dialectic and Ontology in Plato’, Phronesis 12: 118−29. van Eck, J. 1995, ‘Falsity without Negative Predication: On Sophistes 255e−263d’, Phronesis 40: 20−47. —— 1997, ‘A Note on Sophist 257b9−c3’, Mnemosyne 50: 75−7. —— 2000, ‘Plato’s Logical Insights: On Sophist 254d−257a’, Ancient Philosophy 20: 53−79. —— 2002, ‘Not-Being and Difference: On Plato’s Sophist, 256D5−258E3’, Oxford Studies in Ancient Philosophy 23:

63−84. Vitali, M. (tr. and comm.) 1992, Platone: Il Sofista, Milan. Vlastos, G. 1970, ‘An Ambiguity in the Sophist’ = Vlastos (1981), 270−322. —— 1981, Platonic Studies, 2nd edn, Princeton. Vlastos, G. (ed.) 1971, Plato: A Collection of Critical Essays, I, Garden City. von Weizsäcker, C.-F. 1973, ‘Die Aktualität der Tradition: Platons Logik’, Philosophisches Jahrbuch der Görres-

Gesellschaft 80: 221−41. Wagner, F. W. (ed. and trans.) 1856, Platons Sophist, Leipzig. Waletzki, W. 1979, ‘Platons Ideenlehre und Dialektik im Sophistes 253d’, Phronesis 24: 241−52. Warrington, J. (trans.) 1961, Plato: Parmenides, Theaitetos, The Sophist, The Statesman, London and New York. White, A. R. 1970, Truth, London and Basingstoke. White, N. P. 1992, ‘Plato (427−347)’ = Dascal, Gerhardus, Lorenz and Meggle (1992), 234−44. White, N. P. (trans.) 1993, Plato: Sophist, Indianapolis and Cambridge, MA. Wiggins, D. 1971, ‘Sentence Meaning, Negation, and Plato’s Problem of Non-Being’ = Vlastos (1971), 268−303. Wittgenstein, L. 1922, Tractatus Logico-Philosophicus, London. Xenakis, J. 1957, ‘Plato on Statement and Truth-Value’, Mind n.s. 66: 165−72. Zadro, A. 1961, Ricerche sul linguaggio e sulla logica del Sofista, Padua. Zadro, A. (trans.) 1971, Platone: Sofista = AA. VV. (1971), II 181−258.