SOURCE: Scott Austin, in an introduction to Parmenides: Being, Bounds, and Logic, Yale University Press, 1986, pp. 1-10.

[In the following excerpt, Austin introduces Parmenides's poem and considers claims that it is self-referentially inconsistent.]

Parmenides was born in Elea, a Greek colony in southern Italy. He was the founder of Western rational theology, as well as of scientific explanation as we now know it. His innovations in logic, in the metaphysical characterization of ultimate reality, and in the construction of standards for explanation have been passed down to us by Plato and Aristotle, who departed from his insights only in order to further the quest for truth that he had begun. He may also have been a civic leader: it is reported that he constructed a code of laws for his city. He is supposed to have sat at the feet of the philosopher Xenophanes when the latter traveled through Italy, and this seems possible in view of his verbal echoes of Xenophanes; perhaps more probable is the story of Parmenides' devotion to his teacher, a noble Pythagorean of humble birth.¹ Plato (Parmenides 127A-B) dramatically represents Parmenides and Zeno as having talked to Socrates in Athens when Socrates was very young; however doubtful this story may be as a way of dating people exactly,² it may nevertheless be useful as a rough index of contemporaneity and would place Parmenides' death, at an advanced age, sometime around the middle of the fifth century B.C.

Parmenides, Zeno, and Melissus are together regarded as comprising the Eleatic school, which was named after Parmenides' home town; of these three Parmenides and Zeno are of the first rank, with Melissus's position somewhat lower (his exact status depends on which scholar you read).³ Parmenides and his two pupils are distinguished by their use of long, tightly woven proofs and by their concern with the methods of proof and with determining which entities there are for proofs to be about. They began a series of reflections on the relation between demonstration and reality that ultimately bears fruit in Socratic and Platonic dialectic. It is difficult to know whether the unity of the school consisted in anything more than this, however. It is now open to question whether Parmenides, Zeno, and Melissus were all monists, as was formerly thought, and the precise relation between Zeno's destructive⁴ paradoxes and Parmenides' affirmative arguments is not as clear today as it was for Plato, who represents Zeno as simply defending Parmenides against his philosophical opponents.

Parmenides' hexameter poem, which may have been titled "On Nature," survives only in fragments quoted and/or translated by later authors, mainly Simplicius,⁵ who transcribed for us because the poem had become hard to find in his time. We do not have other works by Parmenides. The surviving fragments, in the original and in translation, are given in the Appendix, not as a putatively critical edition but for the reader's convenience in referring to the poem. I also offer translation and analysis throughout where textual questions and questions of meaning arise.

The poem is the first-person account of a chariot ride taken by a young man, one who (we are told) knows, far from the beaten track of humanity, in a chariot drawn by cunning mares. Traveling at tremendous speed, escorted by the maiden daughters of the sun, he arrives at Hesiod's⁶ huge, lintel-surmounted gates, where the paths of night and day converge at the edge of the world—the gates are guarded and locked by the figure of avenging Justice. This is a place beyond, and also at the source of, the contrariety and opposition characteristic of ordinary human life. It takes the maidens to persuade Justice to let him through with his train—the gates strain open—but, once admitted, he is given the gracious welcome appropriate to a Homeric hero. The goddess who greets him reassures the charioteer that, though his route has been an unusual one, he is not doomed but, in fact, has been conducted along by Rightness and Justice. And she promises him an account, not only of the unshaking heart of truth, but also of the opinions of mortals, in which there is not true trust. The epic quest for victory and homecoming has now been transformed by Parmenides into the search for truth, and the youth, like Odysseus and Telemachus, is about to receive further positive and negative directions along the road to ultimate reality.⁷

This is a fascinating prologue, and commentators since ancient times have attempted to find meaning in it: it is a shaman's journey, an Orphic allegory, a secularization of the Phaethon myth, a conceptualized version of the journey of the Homeric hero to his home and goal, a journey to a place where normal distinctions between opposites break down.⁸ The last two are certainly part of the poetic apparatus, which works by incorporating themes from other literary works, especially Homer and Hesiod, but the general purpose of the chariot ride is clear and can be apprehended by the imagination alone. A young man journeys to a faraway place with tremendous urgency under the protection and guidance of maidens; the topography of his journey is one in which the normal directions of up and down are muddled; after penetrating through a gate guarded by an initially severe female figure, he receives a warm welcome from another female, his superior, who proceeds to initiate him into the right path and course of inquiry to follow. This (I believe) noncontroversial description of the journey, if seen as the product of someone's imagination, should allow us to sidestep controversies about timing, topography, underlying metaphors, religious overtones, and direction. Initiation, illumination, katabasis (descent), homecoming, the reception of adulthood—all are ways of describing the journey and its meaning; and metaphors from mystery, epic quests, and cosmic archetypes of all sorts will lie beneath the surface simply because of the nature of the subject matter, regardless of whether the author intended to evoke them specifically. We need not try to excavate them all as distinct from each other. And surely there is an element in the initial story which, though the Parmenidean voyager is male, could be generalized into a universal human content for all philosophical imaginations.

The rest of the poem (fragments 2-19) is taken up with the goddess's speech. It is usually divided into the two parts she promises to deliver: the account of truth (fragments 2-8.51) and the account of mortal opinion (fragments 8.51-19). Fragments 2-7 of the "Truth" section prepare the way for the extended deduction of the transcendental attributes of reality that takes place in fragment 8.

The "Truth" section, then, begins in fragment 2 with the enunciation of two "routes" or "ways" of inquiry. Both routes are described rather elliptically: the first, the "path of persuasion," as the route "that it is and that it is not possible for it not to be"; the second as the route "that it is not and that it is right that it should not be"—"And this," says the goddess, "I point out to you as a completely uninformative route: for neither could you get to know what is not (for it cannot be completed) nor could you point it out."⁹ As the poem continues, these brief route-descriptions are expanded; everything incompatible with saying how it is is rejected as inappropriate in a description of being, of what-is, and an attempt is made to separate the two routes clearly from each other, so that the positive one can then be chosen. There is much disagreement among scholars about the meaning of these compressed requirements and prohibitions and the nature and number of the routes they generate….¹⁰

The goddess continues her discussion of the routes in fragment 6: being, the object of the route that it is, must have some connection with language and thought, for it can be, whereas nothing cannot. The goddess bars the youth from the route of nonbeing, but also bars him from the mixed route taken by mortals, those two-headed creatures whose straying minds are steered by Helplessness, who are tossed about like a senseless swarm of the deaf and blind, who consider being and nonbeing to be the same and not the same.

Thus far, then, we have three routes: one of what-is, the route of persuasion and truth; one of what-is-not, which is completely uninformative, and one taken by mortals, who attempt to mix the two routes. After this point in the fragments we have, the route of what-isnot is discussed only as rejected, and the route of mortals is given full exemplification in the "Opinion" section. It remains, then, for the deductions of fragment 8 to present to us the route of what-is, the route of truth.

Along the route of what-is, the goddess says, are many signposts as to how being is ungenerable, unperishing, a whole of a single kind, unmoving, and perfect. These are the transcendental predicates of being, or what-is. What-is is the subject of the route how it is, the subject about which attributions are made. We are supposed to speak and think of what-is as it is, and not as it is not. What-is cannot come from nothing (lines 7-9) because one cannot speak or think of what-is-not; nor can whatis come to be at any particular time (lines 9-11), for there is not sufficient reason why any particular time should be preferred; nothing besides being can come from nothing either (lines 12-13); and so Justice will permit neither coming-to-be nor perishing, and she does not release her bonds (lines 13-15). The goddess goes on to speak of the choice between "it is" and "it is not" as a "decision" or "verdict" (krisis), continuing the metaphors of justice and morality, and says that the latter must remain "unthinkable" and "nameless," while the former must be allowed to be and to be true. (This transcendental connection between being and truth, implicit in the Greek language, is adopted by later ancient and medieval authors, e.g., Thomas Aquinas.¹¹) A short recapitulation of the preceding argument ("If it comes to be, it is not," lines 19-21) concludes the dis cussion of the signpost "ungenerable." The imperishability of what-is is not directly discussed by Parmenides; it may be that he thought a turning of being into nothing to be so inconceivable ("unheard-from perishing," line 21) as not to be in need of discussion. This section of the poem persuaded Empedocles and Anaxagoras that nothing can come from nothing and that absolute generation is therefore impossible.¹²

The proofs continue with the statement that being is indivisible and homogeneous, without any protuberances or areas of greater or lesser density, and that it is not capable of locomotion; it rests immobile and uniform within the bounds of Necessity (lines 22-31). There can be no thinking without being, because there never can be anything else besides being, Doom having sentenced it to be an immobile whole (lines 34-38). And this being is the ultimate referent of all language as well, even of the terms (to come to be and to perish, to change place, to undergo change of color) that mortals use, wrongly persuaded that they apply to reality (lines 38-41). The Truth section of the poem concludes with a recapitulatory metaphor: being is like a well-rounded sphere equal from every side; it is not right for it to be any bigger or any smaller anywhere, since nothingness cannot prevent it from reaching uniformity. And, since there is nothing in its own nature which would cause it to be asymmetrical, it rests evenly within its bounds. "With this," the goddess concludes, "I leave off my trustworthy discourse and thought about Truth. Now learn from me the opinions of mortals, attending to the deceptive ordering of my words." The Parmenidean version of ultimate reality is thus one from which all distinction, difference, change, and plurality have been excluded, yet one which, in accordance with the horror some Greeks had for the infinite, preserves its definiteness so that it can also be the truth, the implicit and single object of all language. Parmenides is thus the first metaphysician (or, if you prefer, theologian) to argue for those eternal attributes also shared by Plato's Forms, by Aristotle's primary movers, and by their descendants in the history of philosophy. This picture of the truth as a single, abiding whole is next contrasted with the picture the mortals subscribe to. The "Opinion" section of the poem is, in fact, a plausible cosmology of the pre-Socratic type, a description of the world we think we live in, done in great detail and with tremendous richness of scope.

The mortals begin by separating two cosmic opposites—fire and night—in familiar pre-Socratic style, and filling the cosmos with them in such a way as to leave no gaps, at least at the beginning of the cosmic process (fragment 8, lines 51-61, fragment 9). Descriptions are then given of ether, sun, and moon (fragment 10), the encircling heaven, the earth, and the stars (fragment 11), a system of cosmic rings of fire and night containing a Daimōn (a divine or semidivine being) who is responsible for the blending of the sexes in copulation (fragment 12), the creation of Love (fragment 13), the statement (apparently for the first time in the West) that the moon receives her light from the sun (fragments 14-15), a discussion of the relative placements of the sexes on different sides of the womb during embryological development (fragment 17), a theory of the generation of gay men (fragment 18), and a concluding statement seeming once again to suggest that the entire scheme is the work of mortal nomenclature and classification (fragment 19). With this the goddess completes the sketch of the misleading opinions of others that she had promised to the youth, "so that no mortal judgment will ever outstrip you." And yet it is a little surprising to find such a wealth of detail in the descriptions of unstable and changing objects from the philosopher of an immutable cosmic Entity.

It seems that, though the propositions of "Opinion" about the sensible world are not ones that the goddess will herself assent to, there is still something that we, as mortals, are supposed to learn from her description of our world. Scholars have disagreed widely about what that something is, all the way from seeing the "Opinion" section as a serious exposition of Parmenidean doctrine to discarding it as Parmenides' rejection of the sensible world as a mere error.¹³ What are we supposed to learn? First, "Opinion" contains many descriptions that recall, and yet are ironically distanced from, the descriptions of being given in "Truth."¹⁴ Fire, one of the two elements separated off by mortals at the beginning of "Opinion," is "in every way the same with itself and not the same with the other" (i.e., with Night). The assertion about sameness is also made about being, as if Fire were trying to imitate the solidity and substantiality of being; but the assertion about difference would not be said of being, and it represents the respect in which the entities of opinion reveal themselves as copies and caricatures. Second, the two sexes, which are distinguished as opposites by the mortals, just as Fire and Night were, immediately try to overcome this distinction and (as Parmenides sees it) revert to some kind of more primal unity in the phenomena of copulation and homosexuality, thus undercutting the sharp boundaries which the mortals in vain try to set up between their opposites. These opposites, which try to mimic the isolation and distinctiveness of being, end up (as he sees it) mixed together in a parody of themselves in different forms of human sexuality. The irony here is supposed to make us see the ontological values of "Truth" as imperfectly mirrored, or only temporarily attained, in the entities of "Opinion."

But "Opinion" is more than a piece of profound irony; it is also (or, perhaps, consequently) a serious attempt to describe things, in considerable detail, as we do in fact perceive them. This aspect of "Opinion" has often raised suspicions that Parmenides had a serious motive in writing it. And yet there is no incongruity in supposing that an author, convinced of the derivativeness or unreality of the sensible world, could nevertheless try to describe that world as exactly as possible. In fact, the better the description, the more effective the irony would be, as if Parmenides had said to his cosmological contemporaries: "I can give as good a descriptive account of the cosmos as you can, starting from a pair of contraries. But, since I have already shown in "Truth" that the logic of contraries only mirrors and distorts the logic of being, it follows that, since the sensible cosmos has been shown to derive from contraries, such a cosmos is only a reflection in a moving pool of waters of an unchanging sun above." In this way a serious sensible cosmology is at the same time an ironic reductio ad absurdum of the sensible world….

With the end of "Opinion," then, the goddess has delivered on the promise she made to the youth at the beginning, to show forth and show through not only the truth but also the untrustworthy opinions of mortals.

…..

Though Parmenides is famous for saying that one ought not say or think what is not (whatever that is), he went on to compose a poem that seems to do almost nothing else. Since Plato's Sophist, perhaps, Parmenides has been almost as famous for apparent inconsistencies¹⁵ as for the rigid dicta that seemed to land him in them. Here I try to extricate him from some if not all his inconsistencies and to suggest a new role for him in the history of mysticism and logic. I suggest that previous interpretations have not dealt adequately with the problem of Parmenides' negative language, note that there is one negative expression (ouk esti, "isn't") which he does not use except innocuously,¹⁶ suggest that his abhorrence of negative language was not an abhorrence of negative facts and extended only so far as a directly and assertorically negated copula, and try to generate a philosophical rationale which will explain both this limited abhorrence and the rest of the poem….

[In what follows, references designated by the letter B are those of the Diels-Kranz edition (see n. 1), so that, for example, B8.42 means "Diels-Kranz fragment 8, line 42." Line numbers without a fragment number almost always refer to lines in fragment 8 unless some other fragment is prominent in the immediately preceding context. Greek is usually translated except in the footnotes or when a translation has already been given. I always use being and what-is to mean the same thing.]

Though we are told that the route of "is not" is completely uninformative and the route of what-is trustworthy, the goddess's language in fragment 8 abounds in denials, negative words and phrases, and contrapositive methods of proof. This fact has long been noted, but it has not, I think, been sufficiently appreciated. Being "is not lacking" (B8.33), "nor is it divisible" (B8.22), is not such as to be here bigger and there smaller than being (B8.47-48), is "ungenerable" (B8.3), "unmoving" (B8.4), "immovable … without beginning, without end" (B8.26-27), and so on. Parmenides says that it is not right for being to be incomplete (B8.32) and then says that it is complete (B8.42)—that is, he has, at least, the right opinion that one would expect from any Greek about the fact that some double-negatives yield a positive.

Moreover, the entire elenchus operates within a framework of negative proof. "Whole" in line 4 is proved in lines 22-26 by denying "divisible," "here somewhat bigger," and "somewhat smaller"; "ungenerable" and "unperishing" in line 3 are proved by denying various sorts of birth and destruction in the lines that follow; and the signpost at the end of line 4, whose meaning must have been "perfect" or "complete," is proved in lines 42-49 through a series of similar denials, involving language which seems doubly negative: "neither any bigger, nor any smaller … nor is there what-is-not … nor is it in any way possible for what-is to be here bigger and there smaller than what-is." The core of each signpost-proof is thus to say what is true about being by denying what is false. And this is not all: not only does the poem contain negative language in a context where negation and its logic are supposedly being ruled out, but the negations seem to cover a great deal of what one might call the "allowable logical ground." I return to this in chapter 2, but I note here that the poem contains not only the assertion of alpha-privative "predicates" like agenēton, "ungenerable," anōlethron, "unperishing," atremes, "unmoving," akinēton, "immovable," anarchon, "without beginning," apauston, "without end," but also the denial of such a predicate (ateleutēton, "incomplete") using the modal expression ou themis estin, "it is not right," in B8.32. Positive (that is, non-alpha-privative) "predicates" are also asserted (oulon mounogenes, "a whole of a single kind," tetelesmenon, "perfect") and denied (diaireton, "divisible," epidees, "lacking"). The triple oude … oude … oude ("neither … nor … nor") of lines 22-25 is echoed in the akinēton … anarchon apauston ("immovable … without beginning, without end") of lines 26-27, and in the tauton … en tautōi … kath'heauto ("the same … in the same place … by itself") of line 29—an interdependence of positives and negatives in proof. There are, finally, negative expressions involving various personified figures: for example, Justice did not allow being to come to be or to perish (B8.13-14; see also lines 7-8, 36-38)….

In the context of this poem, where negation and affirmation are already at issue in the distinction between two routes, surely there is something we are supposed to notice in all this. Did the goddess have to be negative in order to speak at all? If so, why was she so negative? Interestingly, there was one thing she left out; the words ouk esti, "isn't"—which are used as a nickname for the negative route and in certain other ways—are not used in an assertoric negative predication about being, in what Simplicius tells us is the whole of the "Truth" section of the poem. This fact, which is significant in view of the abundance of other negative language, is important for my hypothesis. But, for now, surely the goddess is inviting us to take up the challenge posed by her apparent inconsistency and to respond to it with our nooi, our minds.

It seems to me that one of the following explanations will carry the field, or else nothing will. (1) If all this negative language and style of argument is deliberate, we must either explain the second fragment's enunciation of the two routes so as to allow the negative language to be acceptable on Parmenides' own criteria, or explain the language as a necessary (though perhaps perversely indulged in) self-referential inconsistency. (2) Perhaps Parmenides merely used the Greek language in the only way he could in order to say what he thought was true about being, and did not worry too much about how well his language conformed to his own rules.

The second explanation, if true, would make Parmenides very uninteresting. Luckily, it is improbable. Parmenides knew what he was doing. If he was so extraordinarily sensitive to, and scrupulous about, negative statements as to eliminate the entire sensible world in one fell swoop on the grounds that it cannot be described without making such statements, then he must have noticed that he himself was making what might well be thought to be the same mistake. But, on the contrary, he seems to have thought nothing wrong with the goddess's discourse about being; there is little room in her "trustworthy discourse and thought about truth" (B8.50-51) for self-doubt.

But, if the language is deliberate, it remains to be seen whether and how the goddess's discourse about Truth in fragment 8 is to be brought into accord with her prescriptions and prohibitions in fragments 2 and 6—if, indeed, this is possible and/or desirable at all. Here we return to the first alternative discussed above, and here there is already a well-charted route to follow through the intense scholarly discussion of recent years. The major possibilities thus far considered seem to be these: (1) The language of fragment 8 does violate the canons of fragment 2, if, indeed, the canons do not violate themselves in their very utterance. Parmenides knew this, but this predicament is inevitable given his view and, far from disabling the poem, it is in fact an index of genuine profundity. (This is the Owen line of interpretation, also in Furth and Nussbaum.)¹⁷ (2) The language of fragment 8 does not violate the canons, (a) because it contains only double-negations, that is, (implicitly) positive statements, not singly negative statements;¹⁸ (b) because fragment 8 does not deny that being exists, but only denies that being has certain properties which, if it had them, would involve its being nonexistent in various ways (Tarán, Barnes, and Gallop, among others);¹⁹ (c) because what is said by means of the negations genuinely characterizes what is real, instead of leading one off into a morass of vagueness in which no definite information can be provided, as would be the case if Parmenides had affirmed what he denies in fragment 8 (Mourelatos).²⁰ These three readings of course depend on and provide support for their authors' interpretations of the canons of fragment 2; they all differ from the Owen line of interpretation in that they attempt to explain fragment 8 as containing language that Parmenides would not disallow, language that should be interpreted as being in some sense, implicitly or in effect, positive, and not really negative….

Plato's and Aristotle's criticisms of Parmenides attribute to him an ignorance of, or blindness to, the fact that what-is is and is not in many different ways. In one way, it can be true of a rose that it is not blue, without its being true that the rose does not exist, that it is not. Or, to put it differently, the "is" of predication is to be distinguished from the "is" of existence. Could ignorance of this distinction have led Parmenides to a monistic conclusion? Yes. If one thought that one could not speak or think of what does not exist, and if one were ignorant of the different behavior of "is" used predicatively from "is" used existentially, one might think that all negative predicates were inadmissible (because assimilated to negative existential statements) and so conclude that the only admissible statement is the statement, Being is. This view of Parmenides was taken by Calogero and Raven.²¹ Others have written of a "fused" sense of "is," intending to avoid an accusation of conceptual unclarity while capturing the fact that the argument might seem transferable into both existential and predicative contexts.²² Such views seem at first to have the advantage of actually explaining why Parmenides said what he said, on the basis of an ignorance which might be thought appropriate for a thinker who composed before Plato's Sophist. Parmenides, on this view, exhibited behavior reasonable for one living in an archaic age, and does not seem merely to have argued for or explicated a revised version of a conclusion which he had swallowed beforehand, for example, a (suitably logicized) One Being from Milesian or Pythagorean cosmology, as in Cornford's interpretation.²³

In fact, however, the Calogero-Raven view explains no more than Cornford's. To take for granted that "is" is fused or confused between existence- and predication-contexts is just as much of a presupposition as to take for granted that Being is One. Moreover, given the connection between an existence-predication conflation and monism (the connection that Calogero and Raven wish to establish), "Being is One" could just as well be a conclusion from, or a justification for, a fusion or confusion between predication and existence. Both are equally basic and, at first sight, equally unargued for by Parmenides. Neither is, on this level, a better explanation. One needs a different criterion than explanatory adequacy for choosing between the two interpretations.

But the view that Parmenides fused or confused existence and predication does not hold up well, if only because the goddess says of being that "it is not lacking" (B8.33: esti gar ouk epidees), and she certainly does not mean to say that being does not exist. I do not think that any interpretation which has Parmenidean existence- and predication-statements joined, confused, or moving in parallel can withstand this counterexample without forcing the goddess to break her own laws. But this raises the question of the status of Parmenides' own language, thus entering immediately into dialogue with Owen and Furth, both of whom view this language as a device for transcending itself.

Owen and Furth give somewhat different explanations for the origin and motivations of Parmenides' argument: Owen, at least initially, interprets him as seeking a general, existing subject on the grounds that only such a subject can suffice for language and for thought, while Furth's models for assertions disallowed by Parmenides involve negative predications and/or negative existential statements. Indeed, all the statements refused by Parmenides fall into one or both of Furth's models. For example, if being were to be more somewhere, it would be not less there, and this is, for Furth, inadmissible, as would be divisibility, a hole in being where being is not. Furth's view, then, adequately accounts for the Parmenidean lexicon of transcendental predicates. I will not stress further differences between Owen and Furth here; in what follows I argue only against the general idea that some parts of Parmenidean discourse about Truth in fragment 8 violate the goddess's own laws for speech. Owen associates himself with the idea that fragment 8 is a Wittgensteinian ladder which is jettisoned at the end of the elenchus; but really the ladder is one upon which one cannot even take the first step, given what Parmenides actually says.

Any interpretation of Parmenides which says that the reason for his view was that he did not like negative predications or existence-assertions ought to be confronted first of all with the fact that Parmenidean discourse about Truth is full of precisely those predications and assertions. In view of all this negative language, one is tempted to ask, how could anyone say that the root of Parmenides' theory was an objection to such speech? The poem's entire method is that of negative elenchus, and the goddess tramples through the negative predications as if they were the sweet grapes of the vine. Surely the burden of proof is on a view in which all her words are disqualified right off the bat.

The resort here is to read fragment 8 as an example of a familiar sort of philosophical self-referential inconsistency: the ladder is rendered unnecessary by its very trustworthiness. Owen recommends transcendental argument; Furth suggests that, given a student of intelligence, the same effect could have been achieved by repeated beatings, Zen-fashion, every time the student used negative language. A supple youth like Zeno would, presumably, respond well to this sort of "negative reinforcement."²⁴ But then why bother to compose a poem at all? Why not just say "Being is"—if one can even say that? Practical charity or vanity are plausible motives, but Furth does not give a philosophical reason why Parmenides should have designed his argument to show how being is by denying how it is not. And this is what Parmenides did do, as was indicated above. And if Parmenides had a genuine antipathy towards negative language, would he not at least have reduced his use of it to the bare minimum, even if the constraints of communication forced him into negative statement? This reduction he certainly does not make—if anything, he flaunts his Mephistophelian side. Or, one who thought self-referential inconsistency was unavoidable might embrace it and become promiscuously inconsistent. In that case, why aren't there even more negations, clearly indicating that the goal is silence?

It might be replied that many philosophers have been in the predicament of saying what they know they ought not to say in order to explain why they ought not to say it. Kant's Critique of Pure Reason is an example, if Fichte was right about it; Wittgenstein's Tractatus has already been mentioned; the Platonic likely stories are perhaps another example. Here the philosopher is depicted as both master and victim of a proper obsession, walking about with cast and crutches in order to demonstrate the impossibility of doing so. Or, perhaps, out of a real charity and concern for truth, Parmenides is willing to bend a principle in order to make it accessible. But Wittgenstein and Plato at least say that this is what they are doing and give good reasons for why they need to do it. Why not Parmenides? If he is indeed as scrupulous about negative language as he is represented to be, then the repeated negations in fragment 8 might be thought daring, as if he were flaunting his violation of the principle, inviting the reader to pounce. This might be good pedagogy in certain situations, but it invites misunderstanding if used as a general method.

What is needed is an explanation of how self-referentially inconsistent or paradoxical (Mackenzie) discourse²⁵ could be regarded as "trustworthy" in the sense demanded by the goddess. Her language of trust has to do with the fixity of being within its bounds (B8.13-15, 26-33), with its accessibility to discourse and thought (B2.1-4, B8.16-18, 34-38), and with the guarantees underlying the positive route (B1.26-30). If a transcendent reality is finite as well, then it can be grasped even as we express its transcendence, or so she seems to imply. How, then, could the violation of a prohibition against negative facts in the very statement of the prohibition be trustworthy? First, it might be said to demonstrate the prohibition by showing the (profound) incoherence that results when it is violated; the fact that the incoherence is necessary for the very statement of the prohibition might not be thought to affect this point. Second, both the positive fact (being) and the prohibition which expresses it through violating it might be said to stand in a relation of trustworthy correspondence to each other, if part of the nature of being, so to speak, is that it must always be expressed in this self-inconsistent way. Thus, someone might say, it is not really a failure of language, but rather something natural, when language bites its own tail in attempting to express a truth so fundamental. Language, on this view, would be untrustworthy if it attempted to disguise its inconsistencies, for the truth cannot be communicated except through the inconsistencies. It is the mortals, not the goddess, who attempt to disambiguate the inconsistencies by shoving contrary predicates onto two substrates, Fire and Night. They fail, on this view, because the truth they attempt to grasp is so paradoxical that its very paradoxically is a sign of truth and cannot be dispensed with.

But of all this, in the statement of the goddess and in the theme of the poem, I do not see clear evidence to clue us in, to let us know that this is what we are supposed to expect. And yet the goddess is not shy about her intentions and our duties and warns us, as at the beginning of the "Opinion" section, when the ordering of her words is to be taken as deceptive. I don't find a self-referentially complicated hermeneutic in the assertoric portions of fragment 8. Surely the language of trust is to be taken at face value, at least so far as the directly attributed negations in the fragment are concerned. Whatever the "sound of one hand clapping" is, it is surely not the language of transcendental proof and justice—in effect, the language of an honest oath. This is not to demean a self-undercutting hermeneutic … but only to say that the great burden of proof involved in attributing such a hermeneutic to Parmenides disappears if we begin by baptizing the negations rather than ending by throwing them away. (I discuss the question whether for Parmenides all words turn out in the end to be shadows, and the problem of self-undercutting with goddess, maidens, and chariot, later in this chapter.) Indeed, the question is a profound one: Do certain paradoxes inherent in some meta-languages extend to the object-language as well, or is there a bounded zone within which the object-language obeys the laws that are set for it? My purpose here is to explore ways in which the second alternative might be worth attributing to Parmenides.

Another objection along self-referential lines is based on Plato's reading of Parmenides in the Sophist. This seems a very fruitful and respectable reading of Parmenides as plagued by the inconsistencies that afflict absolute monists, but it becomes clear that Plato regards the inconsistencies as providing a starting-point for his own pluralistic ontology, for a weakening of Eleatic criteria in order to allow nonidentical things to be fully real…. Surely, it might be said, if Plato thinks Parmenides had the inconsistencies, then so may we, and if he thinks they are profound raw material for a new ontology, then we do not have to regard them as lessening Parmenides' philosophical value.

The difficulty is that Plato thinks Parmenides did not go far enough. Plato does not think it possible to secure a coherent discourse about reality on Parmenidean principles even though he respects Parmenides very much; and one reason why he disagrees with Parmenides is that he thinks Parmenides as a monist is self-referentially inconsistent: if there is oneness, there is also being, different from oneness, and so forth. The Parmenides to whom Plato testifies is not an able selfundercutter; he is not a master of the limitations of positive and negative discourse but ultimately becomes immured in a silence which makes discourse impossible—he is not depicted as self-referentially inconsistent in a successfully self-conscious way. If we say that the inconsistencies were designed to break the rules, we are adding to the Platonic picture, going beyond it.

An interesting additional possibility, both for Furth, Nussbaum, and Mackenzie and for my interpretation, is this: the use of both positive and negative language in Parmenides, in Heraclitus, and in the Sophist's exploration of the fabric of interconnections and nonidentities among Forms might conceal a deeper consensus about truth, one suppressed by the disagreements about monism, pluralism, change, and certain sorts of negative language in relation to others. If the articulation of the truth demands both the positive and the negative side (and I suggest in chapter 6 why this might have been so in Greece) then what matters is that, on all these interpretations, Parmenides too thought that this was so, and my disagreement with these other interpreters becomes less important. I claim here, However, that the goddess's direct, assertoric negative language about being is not meant to be inconsistent merely on the grounds that it is negative.

Various criticisms also apply against other readings in contemporary debate. The existential interpretation offered by Tarán, Barnes, and Gallop has the advantage of simplicity and economy.²⁶ It might be concluded from B8.7-9 ("Nor will I allow you to say or think: out of what-is-not. For it is not sayable or thinkable that it is not.") that a mention of what does not exist in an explanation, even of the relatively minimal sort involved in the statement that what-is comes from what-is-not, is an instance of the sort of discourse involved in the negative route, the route of what-is-not. Parmenides' phraseology here in B8.9 (hopōs ouk esti, "how it isn't") is almost the same as that in fragment 2 (hōs ouk esti, "how it isn't"). If one continues along these lines, it is tempting to adopt an existential reading of the assertions and denials in fragment 8 and to say that wrong discourse for Parmenides is simply the attempt to refer directly or implicitly to what does not exist, or the attempt to say anything about it except that it does not exist. There is nothing for negative existential language to hold on to. Thus, on Tarán's reading, there is nothing wrong with using language that is grammatically negative or alpha-privative so long as such discourse in effect serves to attribute to what-is properties or conditions that are in fact incompatible with its having, or containing, or being susceptible to nonexistence—and there is nothing wrong with using the phrase "is not" in a sentence, so long as that sentence serves to say that one ought not to say "is not" (like B8.8-11) or what implies or involves "is not." Moreover, this interpretation, like Furth's, provides us with an easy deduction of the signposts that Parmenides asserts in fragment 8. For, if being comes into being or perishes, then it comes from, or fades into, what does not exist; if it is divided from itself, then there is somewhere it does not exist; and so on.

There is a compelling reason, however, to disagree with an interpretation that has these advantages. There is evidence in the poem to suggest that Parmenides meant to rule out other kinds of "is not" besides the "is not" which says that something does not exist.

First, at the beginning of the "Opinion" section, when Parmenides states the initial error of mortals in constructing their world-scheme, the dichotomy they set up is described as follows: "Fire … in every way the same with itself and not the same with the other." Now surely this is a specimen or a description of mortal discourse, ironically adopted by the goddess in order to demonstrate the insufficiency of that discourse; it is equally certain that the mortals think there is nothing wrong with it. What is wrong with it? It certainly does not, so far as I can see, either say or imply that something does not exist; in fact, the mortals go on to say that no point anywhere is without both light and night; this is certainly not thought, by them or by Parmenides, to be a situation in which there is a hole in the plenum, a place where an ontological distinction between Fire and Night requires there to be a hole of nonbeing. So, clearly, what is wrong with the world of mortals, at least in its initial formulation, cannot be just that it contains something that is thought not to exist.

But there are problems with the exhaustiveness of the existential reading even in the "Truth" section, with Parmenides' discussion of local inhomogeneity in fragment 8. We hear, in two similar but distinguishable passages, that what-is is not bigger or smaller anywhere. There is some discussion in recent literature about just what the inhomogeneity involves, whether it is one of density, or spatial distribution, or even (wrongly, I think) temporal distribution…. But surely a thing can be (though it does not have to be) more dense in one place and less dense in another without thereby containing a zone of nonbeing.²⁷ It is true that if a stuff is less in one place and more in another, then, even if it contains no gaps, it "is not more" in the place in which it is less, and "is not less" in the place in which it is more. But this is not the "is not" of denial of existence. We still have to explain what Parmenides thinks is wrong with certain statements involving negative predications, not just negative existential statements. Moreover, there are negative existential statements within the body of fragment 8 ("For there neither is nor will be an other besides what-is," B8.36-38). For the existential reading, these are allowable because (like fragment 2) they are to be read as ways of telling us how not to talk. Thus they are not taken to be direct discourse but instead part of a self-referentially inconsistent metalanguage. However, I would like to bring these statements, too, directly within the ambit of allowable Parmenidean expression.

I do not rule out the possibility that a combination of the existential criterion with some other criterion against contradiction (of the sort I present later) might be workable at least as a deduction of the predicates of being.²⁸ But if a purely existential reading of the "is" is not wide enough to rule out "Opinion" and preserve "Truth," one might want to look for some other construal of "is." Here the view to deal with is that of Mourelatos. In what follows I argue against his claim that Parmenides' world could have been composed of many monads, each with its own proper or constitutive name. A potentially pluralistic claim is shared by Mourelatos and Barnes.

Basic to a view of the world as a paratactically ordered plurality of monads, without negation-inviting relations among them, is the view that each thing is what it is and not another thing. If there is only one thing in the world, then one doesn't have to worry, since there aren't any other things, though one may need to show that there aren't any others. But if the world has many monads in it, then when I say, "This is Odysseus," I also say or imply, "This is not anything which is not Odysseus." Parmenides seems to say not only how what-is is, but also how it is not possible for it not to be (B2.3). On Mourelatos' reading, in which the statements made in "Truth" are more or less guidelines for what ought to be said, one ought to avoid characterizing Odysseus in terms which (implicitly or explicitly) relate him to some particular other individual, to someone or something that he is not. Now the "is not" in the statement "is not anything which is not Odysseus" is not supposed to be a focused negation directed at some particular thing (say, Telemachus)—if it were, Odysseus would be constituted by the fact that he is not Telemachus, and this is what is inadmissible on Mourelatos' reading. Rather, the "is not" is supposed to be a general negation, undirected, or directed at any other thing in the world. Odysseus, then, is supposedly constituted by the fact that he "is not" any one of those things which are not Odysseus—not that he "is not this one" or "is not that one," but that he is not any of them. This is what must be required by Mourelatos' view (1) if it is to allow more than one thing as a candidate for what-is, and (2) if directed negation, negation applied to a particular "this," is what Parmenides disallows. And a world composed of monads is supposedly unexceptionable, since one is never allowed to think, even on the underlying metaphysical level, that this one (Odysseus) is characterized by not being that one; the "is not" in "This is Odysseus and is not anything which is not Odysseus" is thought to be merely a strengthening of the applicability of the name to the single thing it singles out.

But here one wants to say, in criticism of Mourelatos, that the situation does arise, sooner or later, in which Odysseus meets up with Telemachus, and then the negation will be not only global but also individual; that is, one cannot say, "This is not any of the others" without sooner or later saying. "This is not that one, or that one, or that one"—and this is a feature even of a world of monads, and even if there are no two mem bers of the same family. If the identity of an individual is secured by its not being identical with any other, then it is secured also, sooner or later, by its not being identical with this one and that one. But then constitutive predications, even in a world of monads, sooner or later involve negative assertions on the deeper metaphysical level, in particular the assertion of nonidentity, so long as there is more than one thing, even if such a world is not imagined as containing hidden strands of logical texture in the form of families or classes. And thus the Parmenidean world, as Mourelatos represents it, could not be set up without violating what Mourelatos takes to be Parmenides' own prohibition. Barnes's view is existential, so the statement of nonidentity is not directly thought of as prohibited language; but all pluralistic views seem to run up against the fact that nonidentity statements are characteristic of "Opinion," not of "Truth." It is Fire, not being, which is "in every way the same with itself and not the same with the other" (B8.57-58).

Let me summarize the train of the argument thus far. (1) Either Parmenides had no objection to (at least some kinds of) negative language, or the entire "Truth" section of the poem is a deliberate, merely heuristic device, not only just because it is language, but also because of the kind of language it is. (2) It is worth viewing the negative language seriously as a kind of language to which Parmenides apparently felt himself entitled. Previous interpretations have too often attributed to him a prohibition against the very language that he uses in fragment 8, and this tells me that it is no longer fruitful to try to figure out what is prescribed and prohibited without examining that language more closely. (3) Parmenides knew (at least on the level of right opinion) that one could make a negative predication without making a negative existence-assertion. Thus he could not have fused or confused the two. (4) On the assumption that, for example, "it is not lacking" (B8.33) and "nor is there what-is-not" (B8.46) are acceptable forms of expression for Parmenides, he cannot have objected to sentences which express negative predications or negative existential statements. (5) Nor could the Parmenidean universe have included a plurality of totally real beings.

Notes

¹ For the evidence on Parmenides' life, see Hermann Diels, ed., Die Fragmente der Vorsokratiker, 11th ed. (5th and subsequent editions revised by Walter Kranz) (Berlin: Weidmann, 1964), vol. I, pp. 138-46. (Hereafter cited as Diels-Kranz.)

² For a statement of the case against an uncritical acceptance of the Platonic tradition, see Friedrich Solmsen, "The Tradition about Zeno of Elea Re-examined," in Alexander P. D. Mourelatos, ed., The Pre-Socratics: A Collection of Critical Essays (Garden City: Doubleday, 1974), pp. 368-93.

³ For a contemporary rehabilitation of Melissus, see Jonathan Barnes, The Presocratic Philosophers, 2 vols. (London: Routledge & Kegan Paul, 1979), vol. I, pp. 155-230.

⁴ See ibid.; and Alexander P. D. Mourelatos, The Route of Parmenides (New Haven: Yale University Press, 1970), pp. 130-33.

⁵ See Diels-Kranz, vol. I, pp. 147-65.

⁶ See Theogony 746-57, M. West, ed. (Oxford: Clarendon, 1966), p. 139. Here I think that David Furley, "Notes on Parmenides" (in E. N. Lee, A. P. D. Mourelatos, and R. Rorty, eds., Exegesis and Argument: Studies in Greek Philosophy Presented to Gregory Vlastos [Phronesis Suppl. Vol. I] [Assen: Van Gorcum, 1973], pp. 1-15), has made it clear that the ultimate destination of the journey is one in which both the cosmic opposites have been transcended. The comma is in line 10, after eis phaos, "into the light," not in line 9, so that the maidens who come to conduct the young man themselves come into the light and do not necessarily come to conduct him into the light. But the fact that this opens up the possibility of an infernal destination for all parties does not mean that the house which is thereby reached is the house of darkness alone. Indeed, a katabasis (descent) to the edge of the world could just as well be spoken of as an ascent, given the nature of houses occupied equally by day and night, i.e., transcendent houses.

⁷ The young man's situation vis-à-vis the goddess is, among other things, the epic situation of one obtaining directions or hearkening to advice, as Mourelatos has convincingly shown. In my opinion the comparison with Odysseus and Telemachus does not depend on the actual topography of fragment 1 (which contains manifestly un-Homeric elements) so much as on the speech situation in this and succeeding fragments.

⁸ For the various meanings assigned since antiquity to Parmenides' prologue, see Leonardo Tarán, Parmenides (Princeton: Princeton University Press, 1965), pp. 17-31, and Mourelatos, Route of Parmenides, pp. 1-46. See also Furley, "Notes on Parmenides"; Jaap Mansfeld, Die Offenbarung des Parmenides und die Menschliche Welt (Assen: Van Gorcum, 1964); and Walter Burkert, "Das Proömium des Parmenides und die Katabasis des Pythagoras," Phronesis 14 (1969): 1-30.

⁹ I do not intend to take a stand here on whether "how" or "that" is the correct translation. (See Mourelatos, Route of Parmenides, pp. 47-73; and David Gallop, ed., Parmenides of Elea: Fragments [Toronto: University of Toronto Press, 1984], p. 55.) Any translation that allows the general declarations of fragment 2 to be expanded into the actual predications of fragment 8 is sufficient for my purposes. I especially regret that time constraints did not permit me to work with Gallop's book except in the Appendix.

¹⁰ For the interpretations of the two routes, see Tarán, Parmenides, pp. 32-40; Mourelatos, Route of Parmenides, pp. 47-73, 269-76; Barnes, Presocratic Philosophers, vol. I, pp. 157-65. I am committed in chapter 1 to the existence of three kinds of discourse in the poem, and I offer evidence for the claim that the route of mortals is not simply the negative route but in some sense a mixture of positive and negative. The question of how many routes there are does not depend upon issues in B6.1-3 alone. The partisans of two or three routes should want only that mortal discourse be neither about pure being nor about pure nonbeing, regardless of how one actually numbers the routes or pins them down. Since it is evident that their discourse is not about pure being, but also that they do make positive existence-assertions and assert positive facts in "Opinion," what they say has an intermediate status and as such cannot (to give Parmenides credit) have been disallowed on the grounds that it was totally negative. What they say is not supposed to be about nonbeing. If it is then claimed that their discourse is somehow implicitly on the negative route in such a way that there are still only two routes, the reply will be that this claim is not now essentially different from the one the partisans of three routes are making—they call the third route a mixture of the first two anyway, admitting that it has (at least partly) negative content. But then the question about the number of routes becomes a question more of enumeration than of philosophical message….

¹¹De Veritate, question 21. Truth (De Veritate) (Chicago: Henry Regnery, 1954), pp. 3-32.

¹² See Empedocles, fragment 12, and Anaxagoras, fragment 17, in Diels-Kranz.

¹³ For the doxography, see Tarán, Parmenides, pp. 202-30, and Mourelatos, Route of Parmenides, pp. 194-221.

¹⁴ See Mourelatos, Route of Parmenides, pp. 222-63.

¹⁵ I shall ultimately offer an explanation in which the negative language will be seen as belonging to the route of truth; but (as explained later) most modern interpretations must disqualify at least some of the negations, and so, for them, the opposition between the hōs esti of fragment 2 and the negative language of fragment 8 is a genuine one. The ouk esti mē einai of B2.3 explicitly admits modal negative language, as I shall argue below; but it does not necessarily justify assertoric double-negations, because its ouk esti is modal; thus the problem of fitting the poem's negative language into its own prescriptions is a legitimate and difficult one. (See below for more criticisms of the double-negative view.)

¹⁶ This point about ouk esti may have been brought to my attention by Peter Meadow.

¹⁷ See G. E. L. Owen, "Eleatic Questions," Classical Quarterly 10 (1960): 84-102, reprinted with revisions in R. E. Allen and D. Furley, eds., Studies in Presocratic Philosophy (London: Routledge & Kegan Paul, 1975), vol. II, pp. 48-81; Montgomery Furth, "Elements of Eleatic Ontology," Journal of the History of Philosophy 7 (1968): 111-32, also in Mourelatos, ed., The Pre-Socratics, pp. 241-70; Martha Nussbaum, "Eleatic Conventionalism and Philolaus and the Conditions of Thought," Harvard Studies in Classical Philology 83 (1979): 63-108.

¹⁸ A tempting view—that the negations in fragment 8 are all negations of inadmissible predicates (like oude diaireton estin, in which diaireton is negated) and so admissible as double-negations—has had defenders in the past. See Hermann Fränkel, Dichtung und Philosophie des frühen Griechentums, 2nd ed. (Munich: Ch. Beck, 1962), p. 402 n. 12; and Hans-Georg Gadamer, "Retraktationen zum Lehrgedicht des Parmenides," Varia Variorum: Festschrift Karl Reinhardt (Münster/Cologne: Böhlau, 1952), p. 61. Mourelatos speaks of the predications of fragment 8 as "not negations made de re but rejections de dicto of negations made de re"; Route of Parmenides, p. 53; and Samuel Scolnikov holds a similar view, in which the proofs of the signposts are negations of mortal attributes…. Here one wants to say: (1) If the first negation is rejected because it is negative, so should the second be, unless the second is de dicto. All such accounts thus presuppose a theory of de dicto discourse like Mourelatos'. (2) Even if the double-negation view is adopted, we still need an account of what makes the first (de re) negations wrong, so this account will inevitably have to distinguish true from false predicates in some way. Thus the double-negation view reduces to one of the other views when it comes to deciding what is on the negative route and what is wrong with the negative route. (3) I claim in chapter 3 that almost all the signposts are proved by denying incomplete predicates like "bigger" and "smaller," which (as Scolnikov says) apply in the sensible world. So it is tempting to think that the signposts are pedagogical only, and that they merely direct us away from "Opinion." But this does not come to grips with the fact that some of the signposts are not only asserted (agenēton … estin) but also proved (tetelesmenon esti) in direct assertions involving estin. There is thus a prima facie case for taking the language of fragment 8 to have as many direct predications in it as possible, to have something to say in its own right. (4) On the double-negation view the shifts from de dicto to de re discourse occur at least twice in each sentence, before and after the negation, even with predicate- and subject-negations, whenever there is a negation. It would seem desirable to switch horses in midstream less frequently, and I shall propose to allow negations within an assertoric sentence itself provided that the copula remains unnegated—though I shall also be saying that some of the discourse is de re.

Nobody will deny that Parmenides very often says what is true by denying what is false. The question is how this is to be described. My de dicto discourse can very often be a singly negated rejection of what is unnegatedly false, and my de re discourse can involve negations of subject or predicate. Double-negation, or Mourelatos' view, in which the rejected subjects or predicates commit one to the negative route, does not allow enough amplitude for positive discourse. For the question why certain positives are accepted and others rejected, see below.

¹⁹ See Tarán, Parmenides; Barnes, Presocratic Philosophers, and Gallop, Parmenides of Elea. See also Jonathan Barnes, "Parmenides and the Eleatic One," Archiv für Geschichte der Philosophie 61 (1979): 1-21.

²⁰ Mourelatos, Route of Parmenides. See the other works by Mourelatos listed in the bibliography for his later settings-forth of this and related views.

²¹ Guido Calogero, Studi sull'Eleatismo (Rome: Tipografia del Senato, 1932). G. S. Kirk and J. E. Raven, The Presocratic Philosophers: A Critical History with a Selection of Texts (Cambridge: Cambridge University Press, 1957).

²² Schofield's revision of Kirk and Raven (2nd ed., 1983, p. 246) abandons Raven's original account of Parmenides as confused in favor of an account of the esti as combining existential and predicative considerations. See also Furth, "Elements of Eleatic Ontology."

²³ Francis M. Conford, Plato and Parmenides (London: Routledge & Kegan Paul, 1939).

²⁴ Furth's phrase.

²⁵ M. M. Mackenzie, "Parmenides' Dilemma," Phronesis 27 (1982): 1-13.

²⁶ See Tarán, Parmenides, Barnes, Presocratic Philosophers, and Gallop, Parmenides of Elea.

²⁷ Though Melissus appears to have argued that differences in density involve different amounts of void (fragment 7, sec. 8 in Diels-Kranz), this involves a view of the microstructure of being which is not explicitly stated by Parmenides. I am indebted to David Sedley for referring me to the Melissus passage. My next point about negative existentials in fragment 8 is also made by G. E. L. Owen in his Sather lecture, given at the University of California at Berkeley, "Words as Quarries, Words as Deceivers" (manuscript at the Classics Faculty Library, Cambridge University); Owen proceeds to interpret these negative existentials as essentially positive, but this new distinction between surface and depth has a tendency to make all discourse positive; if even "nor is there an other" is positive, then the underlying standard for positive discourse is no longer merely existential.

²⁸ I am again indebted to David Sedley for pointing this out. If not only nothing or void but also inequities in its distribution are ruled out, then one can get an argument against the incomplete predicates in B8.22-25 and 42-49 using an existential esti, i.e., if it is supplemented by something like "whatever is true in one place must be true in every place to avoid contradiction"—something like the contradiction-criterion I introduce later. But this means that the entire burden of proof is no longer carried by the existentiality of the esti. Moreover, the negative existentials still remain difficult to explain without involving a self-referentially inconsistent metalanguage. And even if this is explained, it will still be difficult to explain why the locution ouk esti is not used in the negative existentials of 36-37 and 46.