When Socrates falls into doubt at the end of the first part of the Parmenides after failing to salvage his theory of forms, Parmenides recommends training, and in the second part of the dialogue he demonstrates one version of the exercise. In four deductions he adopts a positive hypothesis—if [the] one is—and considers the consequences for the one and other things on that hypothesis; then in four more deductions he starts from the opposing negative hypothesis—if [the] one is not—and considers the consequences for them on that hypothesis.Footnote 1 Were it not for the repeated claims that training of this sort is needed to discover the truth (Parm. 135c8-d6, 136a4-c5, e1-3), the exercise might seem like sheer gymnastics, an impression Parmenides encourages in a summary at the end of the whole exercise:
Let us then say that [viz. “if [the] one is not, nothing is”]—and also that, as it seems, whether [the] one is or is not, it and the others both are and are not, appear and do not appear, all things in all ways, both in relation to themselves and in relation to each other.—Very true.Footnote 2 (Parm. 166c2-5)
This ending balances the conflicting conclusions of the preceding deductions and calls attention to their contradictions.Footnote 3
In my view the final summary is not the true ending of the exercise, but the one that readers will take away if they miss the argument across the deductions, an argument culminating in the final line of Deduction 8 just before the summary begins and recalled in its opening line: “Let us then say that” (Parm. 166c2-3):
Then if we were to say, to sum up, “if [the] one is not, nothing is (ouden estin)”, wouldn’t we speak correctly?—Absolutely. (Parm. 166b7-c2)
If the one is not, there is nothing at all—no world—to explain.Footnote 4 The conclusion is obviously false, since we inhabit some sort of world. The false consequent of the conditional casts doubt on the antecedent, the negative hypothesis itself, and presses the interlocutor to abandon it in favour of the positive hypothesis that the one is. But he cannot simply go back, because—as we shall see—the first part of the exercise compels him again and again to reject the positive hypothesis in favour of its negative counterpart. And yet the negative hypothesis yields a plainly false conclusion.
Since one or the other of the two hypotheses must be true, we should ask whether there is some way to save the positive hypothesis, and there is. I shall argue that to save the positive hypothesis, Socrates needs the give up his thesis in his long speech in Part I that forms cannot partake of other forms (Parm. 129d6-130a2). In some cases a form must even partake of its own opposite, an outcome Socrates declared he would find astonishing, even monstrous (Parm. 129b1-3, b6-c3). According to Socrates’ theory, ordinary things can have opposite features—for instance, Socrates himself is both one (man) and many (parts)—and this is supposed to be unproblematic because he partakes of opposite forms, the form of oneness and the form of multitude (Parm. 129c4-d2); but the forms themselves cannot have opposite features without calling for further explanation, and for him that would entail further (indeed an infinite number of) forms subject to the same problem. So he insists that the form of oneness is simply one and not many.Footnote 5 Socrates challenges Parmenides to prove him wrong, and Part II is that refutation. The exercise teaches two main lessons: first, there must be forms, since there is a world to explain; and second, the one must partake of natures other than its own, including its own opposite, contrary to Socrates’ assertion in Part I. Socrates needs to work out how the one can play the explanatory role required by his theory even though it is both one and many.
1. Plan of the Exercise in Parmenides Part II
Parmenides makes several prescriptions for the exercise before he undertakes it. First, Socrates should adopt the manner of the exercise he heard from Zeno (Parm. 135d8), who would start with a hypothesis that things are many, and show that they have incompatible properties (e.g., they are both like and unlike, both limited and unlimited, both one and many). By demonstrating that the same things cannot have incompatible properties at the same time, Zeno attacked the thesis of Parmenides’ opponents and came to the aid of Parmenides’ own thesis that the all is one. Second, Parmenides approves of something Socrates said to Zeno. While Zeno displayed the compresence of opposites in visible things, Socrates challenged him to display the same difficulty in the case of forms (Parm. 129d6-130a2), and Parmenides now recommends that Socrates follow Zeno’s method but apply it to intelligible things—forms—not concrete particulars (Parm. 135d8-e4).Footnote 6 Third, whereas Zeno in his book started from a single hypothesis and derived various consequences, Parmenides advises Socrates not only to hypothesize that something is, but also to hypothesize that that same thing is not (Parm. 135e8-136a2).
Parmenides also makes a fourth recommendation: once Socrates has completed the upcoming exercise, he should repeat it, taking some other entity as the subject of a positive or negative hypothesis (Parm. 136b1-c5). For example, Socrates should hypothesize that likeness is or that it is not, and examine the consequences on each hypothesis for likeness itself and other things. He should do it again for unlikeness, change, rest, generation, destruction, being, and not-being. In short, says Parmenides, the same method applies to whatever Socrates might hypothesize as being or not being or as having any other property.Footnote 7 He does not demonstrate this extended program in Part II, but restricts himself to examining one hypothesis and its negation. In my view we get a second version of the exercise—the exercise about being—across parts of the Theaetetus and Sophist. Footnote 8 In this paper, I shall argue that the exercise about oneness in the Parmenides shows why Socrates needs to undertake the exercise about being.
Puzzled by Parmenides’ third point about the structure of the exercise, Socrates asks for clarification, and Parmenides gives an illustration, using Zeno’s hypothesis:
If [the] many are, what must the consequences be both for the many themselves in relation to themselves and in relation to the one, and for the one in relation to itself and in relation to the many? And, in turn, on the hypothesis—if [the] many are not—you must again examine what the consequences will be both for the one and for the many in relation to themselves and in relation to each other. (Parm. 136a5-b1)
Parmenides’ illustration suggests that, starting from each hypothesis (positive or negative), Socrates should consider the consequences both for the entity (or entities) hypothesized (in this case the many) and for other things (in this case the one), and whichever of the two he derives consequences for, he should consider the consequences for it in relation to itself and in relation to the other things. The illustration indicates that there will be paired deductions, and the number of pairs depends on how Parmenides carves up the parts.
When we look at the actual exercise in Part II and its positive and negative hypotheses about the one, we find not two, four, or eight deductions, but nine. On closer inspection this irregularity can be readily explained, because one deduction stands out from the overall scheme. Whereas four deductions start from the positive hypothesis and four from the negative hypothesis, an additional deduction following the second opens—“Let us speak yet a third time”—and begins not from the original positive hypothesis about the one but from a hypothesis combining conclusions of the second and first deductions: “If the one is as we have described it—being both one and many (Deduction 2) and neither one nor many (Deduction 1), and partaking of time (Deduction 2)” (Parm. 155e4-6). Given its reliance on both Deductions 1 and 2, we should regard this deduction as an appendix to both, and not as part of the overall plan described at the outset.Footnote 9
I begin by describing what happens in Part II without interpretation. Each of the two main parts of the exercise, beginning from the positive or negative hypothesis, has four deductions as parts, for a total of eight deductions. Two deductions in each foursome examine consequences for the entity hypothesized, and two examine consequences for other things. One member of each pair derives positive consequences (X is both F and not-F, for various values of “F”), the other negative (X is neither F nor not-F for those same values). The scheme in Part II contains a number of anomalies besides the Appendix, and in particular Parmenides reverses the order of the first and second deductions relative to the other six (the first deduction derives negative consequences, the second positive; thereafter, the order is consistently positive-negative). The table below gives the basic structure (+ for positive, – negative):

2. Competing Interpretations of the Exercise
Some interpreters regard Part II as a parody of other philosophers’ methods, but most scholars take the exercise seriously and find in it a constructive message.Footnote 10 Among those who take Part II seriously, a substantial number contend that all or most of the conclusions in Part II are acceptable, some scholars arguing that apparently conflicting deductions treat different subjects—ones essentially characterized by different properties—while others urge that opposing deductions treat the same subject from different perspectives, examining what is true of it in virtue of its own nature (by definition) or what is true of it through displaying features (not by definition).Footnote 11 According to another broad tradition, paired deductions treat the same subject but present real antinomies designed to prod Socrates to recognize that the theory of forms presented in the first part of the dialogue involves some mistaken assumptions.Footnote 12 On the antinomist view most of the conclusions are validly deduced, given the premises, but are unacceptable—unsound.
My view shares the spirit of the antinomist approach but does not fit squarely in that tradition, because I take Part II to contain a single argument composed of nine deductions and designed to provoke the student to find a way to save the positive hypothesis that the one is. All the deductions treat the same subject, the Platonic form of oneness. Deductions 1 and 2 reach opposite conclusions helpfully stated in a summary of the positive hypothesis at the end of Deduction 4—the one is everything (panta) (Deduction 2) and not even one (oude hen) (Deduction 1) (Parm. 160b2-3)—but the second deduction has the same implication as the first, since the one cannot be everything without having manifold incompatible properties. Thus both Deductions 1 and 2 lead independently to the negative hypothesis that the one is not. The Appendix tries to combine the results of Deductions 1 and 2, but its proposal cannot succeed and again points to the negative hypothesis. Deductions 3 and 4, which examine consequences for other things on the hypothesis that the one is, also reach opposite conclusions—that the others have various properties (Deduction 3) and that they have none (Deduction 4)—but Deduction 3 spells out many fruitful results, which are then discarded in Deduction 4. Deductions 3 and 4 stand outside the main argument by ignoring the overwhelming reasons already indicated to reject the positive hypothesis. At the same time, Deduction 3 holds the key to saving it, but to save it, the student must locate the mistaken assumption that allows Deduction 3 to be undermined in Deduction 4.
It may be helpful to contrast my view with the second constructive approach, since my interpretation shares some close affinities with it but differs markedly in its conception of the overall design of the exercise in Part II and of the new rounds recommended by Parmenides.
In her acclaimed book on the Parmenides, Constance Meinwald takes the illustration about Zeno (quoted above) to structure the exercise about oneness and future rounds featuring other entities, and she argues that paired deductions in the second part of the Parmenides examine the same subject either pros heauto (in relation to itself) or pros ta alla (in relation to other things). Deductions considering the subject pros heauto investigate the nature of the subject (what is involved in—i.e., inside—its nature), whereas deductions considering the subject pros ta alla examine features it displays (features it has by standing in a relation to entities typically, though not always, outside its nature).Footnote 13 According to Meinwald, the Parmenides captures two kinds of predication Michael Frede finds in the Sophist but uses the phrases, pros heauto (“in relation to itself”) and pros ta alla (“in relation to the others”) in place of the terminology in the Sophist, kath’ hauto (“by/in virtue of itself”) and pros alla (“in relation to others”).Footnote 14
On two fundamental points I agree with Meinwald. First, I agree that all the deductions treat the same subject, the Platonic form of oneness. Second, I also agree that the first two deductions, which make up the bulk of Part II (nearly two-thirds of that long stretch of argument), consider the one from two distinct perspectives. Deduction 1 examines what the one is by itself (in virtue of its own nature), and Deduction 2 examines the one in relation to other things. Beyond that our interpretations radically diverge. Meinwald thinks that the apparently conflicting conclusions of paired deductions should be equally acceptable once they are read in light of the pros heauto or pros ta alla qualifiers, and that Parmenides repeats the strategy four times in the exercise about oneness and recommends the same design in future exercises treating different subjects. By contrast, I think that Parmenides demonstrates in the first two deductions that each approach, taken on its own, yields extreme and unacceptable results—results furthermore in conflict with each other. If I am right that each strategy is excessive, it would be unproductive to repeat those strategies over and over. In my view both the Appendix and Deduction 3 try in different ways to find a middle path, the Appendix by arguing (unsuccessfully) that the conflicting conclusions of Deductions 1 and 2 can be true at different times, Deduction 3 by integrating the two approaches. The argument of Deduction 3 is itself undermined in Deduction 4, and I argue that to find a way back to the constructive results in Deduction 3, the inquirer must identify and reject the mistaken assumption that ruled out those results.
As a first step in promoting my alternative, I consider the merits and some disadvantages of Meinwald’s proposal. Her proposal is attractive because the two phrases, pros heauto and pros ta alla, are emphasized at three key moments in the discussion: in Parmenides’ illustration of the exercise with reference to Zeno (Parm. 136a4-b1, quoted above), and in two summaries of the exercise—a review of the positive hypothesis at the end of Deduction 4 (Parm. 160b2-4), and a summary of the whole exercise at the end of Deduction 8 (Parm. 166c2-5, quoted at the start of this paper).
Consider again the illustration about Zeno: Socrates is to start with the positive hypothesis, that the many are, and examine the consequences for the many in relation to themselves (pros hauta) and in relation to the one (pros to hen) and for the one in relation to itself (pros hauto) and in relation to the many (pros ta polla); then, starting from the negative hypothesis that the many are not, he is again to examine the consequences for the one and for the many in relation to themselves (pros hauta) and in relation to each other (pros allēla) (Parm. 136a5-b1).Footnote 15 In Meinwald’s view, the illustration forecasts eight sections in the upcoming exercise, and they will come in pairs, with one member examining consequences for the subject pros heauto—in virtue of its own nature—the other examining properties it displays, properties it has by standing in a relation to natures typically other than its own (pros ta alla). If we ignore the Appendix, Part II does indeed consist of eight deductions, as the illustration so interpreted suggests.
Meinwald’s view is especially tempting, because the first two deductions do indeed examine what the one is by itself (solely in virtue of its oneness) and in relation to other things. If she is right that pros heauto and pros ta alla capture the two perspectives, the striking repetition of these phrases in the illustration about Zeno, description of new rounds, and summaries is powerful evidence for the design of the exercise as Meinwald conceives it.
If we examine the actual exercise in the second part of the Parmenides, however, we find that the phrases pros heauto and pros ta alla, though used throughout the exercise, do not have the force that Meinwald attributes to them, but typically express a subject’s relation to itself or other things.Footnote 16 When Parmenides wants to say that an entity X stands in a certain relation to itself or other things, his usual terminology is: “X has the relation (sameness, unlikeness, etc.) in relation to (pros) itself and/or in relation to (pros) other things,” as in “the one has unlikeness in relation to the others (pros ta alla)” (Parm. 161a6).Footnote 17 An example using the pros-phrases from Deduction 6 shows that he considers relations to self and others in the same deduction: “Furthermore, it [the one] would have neither likeness nor difference in kind in relation to itself (pros heauto) or in relation to the others (pros talla)” (Parm. 164a2-4). In fact most of the deductions consider the subject(s) both in relation to itself/themselves and in relation to other things.Footnote 18 Given Parmenides’ frequent use of these expressions in the exercise, we can see why they are repeated four times in the illustration about Zeno. It is highly relevant in an exercise about the many to consider what they are in relation to themselves (same as themselves, different from themselves, like, unlike, equal, unequal, etc.) and in relation to the one (collective wholes, unified, discrete individuals). Similarly in an exercise about the one it is useful to ask: What is the one in relation to itself (same as itself, like, etc.)? And what is it in relation to the many (different from them, unlike, etc.)? The emphatic repetition of these phrases in the illustration is striking, but that emphasis does not indicate that the phrases structure steps in the exercise.Footnote 19
As I understand it, the illustration captures only four steps: (1) If X is, what follows for it? (2) If X is, what follows for other things? (3) If X is not, what follows for it? and (4) If X is not, what follows for other things? In the actual exercise each step has two “claws”, one arriving at positive consequences, the other negative for the same subject (one or others). Pairs of “claws” demonstrate antithetical conclusions.Footnote 20
While I deny that pros heauto and pros ta alla play a structuring role in the design of the exercise, I agree with Meinwald that Deduction 1 considers what the one is by itself (in virtue of its own nature: kath’ hauto) and that Deduction 2 considers the one in relation to other things (pros alla), and I am largely sympathetic to her analysis of those two deductions. But I disagree that Deductions 1 and 2 reach acceptable conclusions. Showing that the conclusions are too extreme will prepare the ground for me to argue that Parmenides seeks a middle way in later deductions.
Significant from my perspective but downplayed by Meinwald, at the end of Deduction 1 Parmenides invites his interlocutor to reject his final conclusions and thus suggests that at least some conclusions in the exercise are unacceptable. Starting from the positive hypothesis about the one, he considers consequences for it and concludes that it is nothing at all—not even one—but then adds a qualification: “if we are obliged to trust this argument” (Parm. 141e12-142a1). He goes on to say that since the one is not, it has no name or account, and there is no knowledge, perception, or opinion of it, and then asks: “Can these things be so concerning the one?” to which his interlocutor, young Aristotle, replies: “I certainly don’t think so” (Parm. 142a6-8).Footnote 21 By calling into question the conclusions of the first deduction, Parmenides alerts the audience that they should not accept all conclusions in Part II.Footnote 22 Deduction 1 is the only time in Part II that the interlocutor expresses misgivings, but this one occasion is telling. Not only does Deduction 1 reach an unacceptable conclusion, but all the supposedly pros heauto deductions (Deductions 1, 4, 6, and 8) conclude that the subject (one or others) is nothing at all. This recurrent result scarcely seems productive, if Parmenides is investigating the nature of things.
I now present the design of the exercise as I understand it, and we start at the very beginning with the positive hypothesis itself.
3. The Positive Hypothesis
Many scholars agree that the first and second deductions present opposing conclusions about the same subject, but we must first establish that these two deductions start from the same hypothesis—often unquestioned but far from evident. I shall argue that they start from variants of the same hypothesis. Showing this will pay a valuable dividend, because the hypotheses in Deductions 1 and 2 can be distinct versions of the same hypothesis only if one and the same form—being—operates in two ways. Getting clear on the positive hypothesis will enable us to see how Deductions 1 and 2 fit together as the first two steps of a larger dialectical strategy.
Parmenides sometimes expresses the positive hypothesis as εἰ ἕν ἐστιν ((ei hen estin) Deduction 1) and sometimes as ἓν εἰ ἔστιν ((hen ei estin) Deductions 2, 3, and 4).Footnote 23 While word order in English often affects the meaning of a sentence, Greek is an inflected language, and so the order of the words need not affect the meaning. That said, the Greek in both versions is underdetermined, since we can construe the adjective hen either as grammatical subject, “if [the] one is,” or as predicate (with grammatical subject understood), “if it—the one—is one”. I shall argue that we should translate the hypothesis in the first deduction as “if it—the one—is one,” and in the second, third, and fourth as “if [the] one is”.Footnote 24 The syntactically distinct versions suggest that the “is” has different meanings in the two contexts, but I shall contend that it has the same meaning in both. This second part of my argument depends on showing that a single form—being—can be both a monadic property and a link to further properties.
To see that the hypothesis is to be construed in different ways in different deductions, we should start by considering evidence from the first part of the dialogue and the transition to Part II, because different sections support different alternatives. I will not go through all the evidence here, but will simply give some illustrations to support my conclusions. At the beginning of Part I, Socrates mentions Zeno’s and Parmenides’ hypotheses and the hypothesis Zeno denies, and he uses “one” and “many” as predicates, as in the hypothesis Zeno denies—“if things are many” (ei polla esti ta onta) (Parm. 127e1-2)—here the grammatical subject ta onta is explicitly mentioned, and “many” is the predicate. He states Parmenides’ hypothesis—“the all is one” (hen ... einai to pan) (Parm. 128a8-b1)—and again states a grammatical subject, using “one” as the predicate. The opening part of the dialogue thus suggests that “one” and “many” will be predicates in the hypotheses of Part II.
But in his prescriptions for the upcoming exercise in the transitional section introducing Part II, Parmenides uses “one” and “many” as grammatical subjects. Recall that he says that the exercise will be just what Socrates heard from Zeno, except that Socrates should focus on forms rather than visible things; and that, whereas Zeno started from a single hypothesis and derived various consequences, Socrates must hypothesize not only “if each thing is” (ei estin hekaston) and consider the consequences, but also “if that same thing is not” (ei mē esti to auto touto) (Parm. 135e8-136a2). Here “each thing” and “the same thing” serve as grammatical subjects in the positive and negative hypotheses, while “is” and “is not” are predicates. I interpreted Parmenides’ illustration about Zeno accordingly (Parm. 136a4-b1). Contrary to our observations about the early part of the dialogue, the bulk of the evidence from the transitional section promotes the expectation that “one” and “many” will be subject-expressions in the hypotheses in the upcoming exercise.
Before we assume that Parmenides has revised the hypothesis in the course of Part I, we should notice one last piece of evidence from the transitional section. When Parmenides finally yields to the entreaties of his companions and agrees to give a demonstration, he proposes to start with himself and his own hypothesis, saying that he will hypothesize about the one itself and consider what the consequences must be “if it is one or if it is not one” (eite hen estin eite mē hen) (Parm. 137b4). This translation, taking “one” as predicate, is the only idiomatic rendering of the Greek without emending the text.Footnote 25 We need feel no surprise that Parmenides states his own hypothesis this way, with “one” as predicate rather than as subject, given Socrates’ earlier characterization of Parmenides’ claim: “the all is one” (128a8-b1)—“if it is one” simply follows from that formulation. I take this statement to set the stage for the hypothesis in Deduction 1 of Part II.
Although Parmenides does not explicitly use a subject-expression when he first states the hypothesis in Deduction 1, “the one” is to be understood as the subject, because he mentions the one in the main clause of the sentence (the first consequence): “If it is one, the one would not be many” (Parm. 137c4-5). This translation is confirmed a few lines later: “Therefore, if the one is to be one (ei hen estai to hen), it will neither be a whole nor have parts” (Parm. 137d2-3). Here the adjective hen is the predicate, and the substantive-expression to hen is the grammatical subject. This result makes sense, because Parmenides said at the outset that he would hypothesize about the one itself, and stated the positive and negative versions of his own hypothesis as “if it is one or if it is not one” (Parm. 137b4). So we should understand the hypothesis in Deduction 1 as: “if it—the one—is one.”
An additional piece of evidence comes from the beginning of Deduction 2 (Parm. 142b5-c7), where Parmenides discusses the meaning of the positive hypothesis and distinguishes two ways to understand it—as “if [the] one (is) one” (ei hen hen) and “if [the] one is” (ei hen estin). He says that now (in Deduction 2) we should construe the hypothesis as “if [the] one is” and that the hypothesis means that the one partakes of being.Footnote 26 Someone might object that ei hen hen—whose literal translation is “if one one”—is not the version of the hypothesis in Deduction 1, since we must supply the “is,” but Parmenides has a reason to omit the verb here.Footnote 27 The first deduction considers the one by itself independent of everything else. The original version of the hypothesis, “if [the] one is one,” assumes that the one partakes of oneness, and it becomes evident as the deduction proceeds that it partakes only of oneness. Parmenides then reveals a shocking result at the end of the first deduction, that the one cannot partake of its own character—oneness—without also partaking of being, a character distinct from oneness, to link it to its oneness (Parm. 141e9-142a1). Because that result follows from the hypothesis and is not presupposed in its original formulation, Parmenides omits the troublesome “is” in stating the hypothesis at the start of the second deduction to capture the intent of the hypothesis in Deduction 1, that the one has only one feature, its oneness.
To understand why Parmenides takes the hypothesis in different ways in Deductions 1 and 2, and why they are nonetheless variants of the same hypothesis, we need to make sense of “is” in the two versions. In the hypothesis “if [the] one is one,” the “is” is an incomplete predicate linking the grammatical subject to the grammatical complement. In Greek both the subject and complement are adjectives—not two substantive-expressions or one substantive and one adjective—and so we must decide whether the “is” serves as an identity-sign, as in the sentence, “Zeus is Jupiter,” in which case the expressions on the left and right side of “is” refer to the same entity; or whether it serves as a copula, as in the sentence “Simmias is large,” in which case only the subject-expression refers to something, while the predicate ascribes a property to it. I quote again the last sentence of the first paragraph of the first deduction, since it indicates that the “is” functions as a copula: “Therefore, if the one is to be one, it will neither be a whole nor have parts” (Parm. 137d2-3). Because the hypothesis here includes a substantive-expression as grammatical subject and an adjective as complement, the hypothesis is a predication, and a predication of a special Platonic sort: a self-predication. According to the hypothesis in Deduction 1, the one has its own proper character—the one is one.
In the version of the hypothesis in the second deduction, “if [the] one is,” the “is” functions syntactically as a complete predicate. If the verb also functions semantically as a complete predicate, it requires nothing further to interpret its meaning. Whereas English has a separate verb “exists” to serve as a complete predicate, classical Greek had no separate verb but attributes existence to something by means of the verb “to be” without a complement. For this reason many interpreters of the Parmenides have translated the positive hypothesis as “if [the] one exists.” If that were the correct construal of the hypothesis in the second deduction, the hypotheses in the two versions would have different meanings, since the incomplete “is” (copula) and the complete “is” (“exists”) would play different semantic roles. The difference is evident in a sentence such as, “Pegasus is a winged horse, but he does not exist.” We ascribe properties to Pegasus even though he is an imaginary object, not an actual living creature.
Plato’s use of the verb “to be” has been much debated, especially in connection with the Sophist, and scholars dispute whether the syntactically distinct uses of the verb in “X is F” and “X is” reflect a semantic difference. We need not enter that debate here, but merely take our cue from what Parmenides says at the end of the first deduction and start of the second.Footnote 28 At the end of Deduction 1 he argues that the one cannot even be one because, to be one, the one would have to partake of being, which links it to its oneness, and so would not only be one but also be (Parm. 141e9-142a1), contrary to the first consequence of the first deduction that the one is simply one and not many (Parm. 137c4-5). The beginning of Deduction 2 emphasizes that the one partakes of being and that being is distinct from oneness (Parm. 142b5-c7). These passages indicate that being is a feature distinct from oneness and that it can function in two ways: it plays a relational role in linking something to a property it has (stated in a sentence such as “the one is one” or “the one is large”) and it is itself a monadic property the subject has (stated in the sentence “the one is”). In either case, the subject is by partaking of being. The fifth deduction makes trouble for the idea that being both links something to a property it has and is itself a property the entity has, since the subject will need an infinite number of links to have any character at all (Parm. 161e3-162b3). The conception of being is thus problematic, but the shift from one version of the positive hypothesis to the other ceases to be perplexing, because the same entity—being—functions in two ways. Since this is so, it is no longer remarkable that, without warning or noticeable misgivings, the speakers switch back and forth between two versions of Parmenides’ and Zeno’s hypotheses earlier in the dialogue. The “is” is univocal in the two versions.
To sum up: Deductions 1 and 2 adopt different versions of the same hypothesis. The version in the first deduction characterizes the one in a particular way, namely as one, whereas the version in the second deduction characterizes the one as being, treating being as a property that can stand alone or link the one to its own or other natures. This observation is of considerable importance in connection with being in the Sophist, where the Eleatic Stranger claims that being operates in two ways—auto kath’ hauto (itself by itself) and pros alla (in relation to other things) (Soph. 255c13-d7). To judge from the Parmenides, the same entity functions in two fundamental ways, as a monadic property, and as a relational property linking the subject to its own or other properties.
4. The First Antinomy
We need not go through the details of the opening pair of deductions to grasp their strategy, some elements of which have already emerged from our previous discussion. The most important point, however, can be shown only now that we have distinguished the two versions of the positive hypothesis in Deductions 1 and 2, according to which the hypothesis in Deduction 1 is: “If it—the one—is one.”
Parmenides opens the first deduction (Parm. 137c4-142a8) and derives as the first consequence Socrates’ thesis about oneness from his long speech in Part I:Footnote 29
If it [the one] is one, the one would not be many, would it?—No, how could it? (Parm. 137c4-5)
In his long speech in Part I, Socrates, having just declared that there is nothing surprising about sensible things being both one and many, offers a challenge to Zeno and Parmenides:
But if he [Zeno or Parmenides] should demonstrate this thing itself, what one is (ho estin hen), to be many, or, conversely, the many to be one—at that I’ll be astonished. (Parm. 129b6-c1)
Parmenides takes up the challenge in the exercise in Part II. Having asserted Socrates’ thesis as the first consequence in Deduction 1, he uses it as a premise in the rest of the deduction and derives many further consequences: the one is neither F nor not-F for many values of “F”—neither changing nor resting, neither the same as nor different from itself or other things, neither like nor unlike itself or others, and so on—and he crowns the argument with the disturbing conclusion that the one is not even one but nothing at all (Parm. 141e9-142a1). Parmenides arrives at that stunning result by arguing that the one cannot be one without partaking of being, a feature different from its oneness, to link it to its oneness. To be one, the one must have two features, oneness and being.Footnote 30 While the consequences of the first deduction are validly deduced, given the initial premise and first consequence, the conclusion that the one itself is nothing at all is unacceptable, and as I have already stressed, Parmenides invites young Aristotle to reject it and his further consequences (Parm. 142a1-8).
To avoid the conclusion of the first deduction and save the positive hypothesis, Parmenides opens the second deduction by weakening Socrates’ thesis to allow the one to partake of being to link it to its oneness: the one is now allowed to partake of two natures, oneness and being (Parm. 142b5-c7). But that seemingly small amendment has far-reaching consequences, because the one turns out to be both one and unlimited in multitude (two arguments: Parm. 142c7-143a3 and 143a4-144e7), and once that has been established, Parmenides argues that the one is both F and not-F for the same values denied of it in Deduction 1—both changing and resting, both the same as itself and different from itself, both the same as the others and different from them, and so on—and ultimately everything.Footnote 31 That result will not do either—since to be everything, the one will have many mutually exclusive properties. Both Deductions 1 and 2 independently show that the positive hypothesis must be rejected in favour of the negative hypothesis.
The order of Deductions 1 and 2 is reversed (negative-positive) relative to the rest of the paired deductions in Part II (which are all ordered positive-negative), because Deduction 1 sets the stage for the whole of Part II, starting from Socrates’ thesis in Part I, that the one itself is one and not many. Socrates challenged Parmenides to prove him wrong (Parm. 129d6-130a2), and the exercise in Part II is that refutation—an indirect argument using Socrates’ thesis as the first step in Deduction 1. Socrates is wrong, because his thesis entails unacceptable results: if the one is one (positive hypothesis in its first version), the one is nothing (conclusion of Deduction 1); if the one is not (negative hypothesis), there is nothing at all (conclusion of Deduction 8). To avoid these consequences Socrates must give up his thesis and admit that the one is both one and many. And yet that admission seems unpromising too, because as Parmenides shows in Deduction 2, if the one partakes of being to link it to its oneness, that concession—allowing the one to have two features, oneness and being—has massive unwanted consequences, that the one is everything.
The major problem in the First Antinomy (and indeed in the whole of Part II) is a conception of being shared by both deductions, that being is a nature outside the nature of beings, including oneness. Parmenides alerts Socrates and the reader to the issue when he comments on Socrates’ failure at the end of Part I:
Only a very gifted man can come to know that for each thing there is some kind, a being itself by itself (ousia autē kath’ hautēn); but only a prodigy more remarkable still will discover that and be able to teach someone else who has sifted all these difficulties thoroughly and critically for himself. (Parm. 135a7-b2)
What is the being auto kath’ hauto for each thing? In particular, what is the being of oneness? By assuming that being stands outside the one, Deduction 1 shows that the one cannot even be one without partaking of a nature other than its own and so, to have its own nature, it must be more than one. Deduction 2, starting from the hypothesis that the one is, derives as its first consequence that the one partakes of being, and as its second that being differs from oneness:
If [the] one is, can it be, but not partake of being?—No it cannot.—So there will be the being of the one, and that is not the same as the one. (Parm. 142b5-8)
From there Parmenides argues that the one is everything. How should the inquirer resolve the First Antinomy—that the one is both everything and nothing?32
5. Appendix and Second Antinomy
Parmenides makes two attempts to reconcile Deductions 1 and 2, in the Appendix and Deduction 3. In the Appendix he says he will speak of the one a third time and now takes as his hypothesis conclusions of both Deductions 1 and 2: “If the one is as we have described it—being both one and many (Deduction 2) and neither one nor many (Deduction 1), and partaking of time (Deduction 2)—must it not, because it is one, sometimes partake of being, and in turn because it is not [one], sometimes not partake of being?”Footnote 33 In the Appendix Parmenides proposes to accept the (independently impossible) conclusions of Deductions 2 and 1, and taking them as contradictory, claims that the one is in those opposite states (both one and many/neither one nor many) at different times.Footnote 34 But even supposing that the opposing states were possible, this solution will not work, because the contradiction Parmenides aims to avoid by appeal to different times reappears at the instant of switching, a moment he locates outside of time.Footnote 35 The Appendix cannot save the positive hypothesis by allowing the conclusions of Deductions 1 and 2 to be true at different times, and moreover the contradiction at the instant reinforces the implication of Deductions 1 and 2 individually: the positive hypothesis must be abandoned in favour of the negative hypothesis.
Deduction 3 takes a different tack in its attempt to reconcile the first two deductions. Focusing on consequences for the others, on the hypothesis that the one is, it considers what the others are both in relation to the one (pros to hen) and in virtue of themselves (hē d’ heautōn phusis kath’ heauta) (Parm. 158d5-6; cf. 158b5-7, c5-7, e2-3). This deduction aims to show what oneness contributes to other things (including unity, individuality, and internal structure), and what the others are by themselves without oneness. Without oneness they are by nature indefinite multitudes lacking distinctive natures and boundaries. The deduction reveals that the tasks of Deductions 1 and 2 cannot be completed separately but must be integrated: for the others to be the definite things that they are, they must partake of oneness, a form inside their natures, which unifies their parts and marks them off as individuals. This insight about the others suggests a way to overcome the problem at the end of Deduction 1: whereas it appeared that the one cannot even be one by itself, it can after all be one by itself if being is inside its nature, somehow structuring oneness as what it is. Deduction 3 also indicates a way to avoid the rampant proliferation in Deduction 2, because the nature of oneness should give it stability and thus put constraints on its relations to things outside its nature. In my view Deduction 3 is the place where Parmenides indicates how to save the positive hypothesis.
Deduction 4 rules out the conclusions of Deduction 3 by arguing that the others cannot partake of the one without fragmenting it into many. Thus Deduction 4 again takes for granted Socrates’ thesis from Part I, that the one is simply one and not many, and once more reaches unacceptable results, this time for the others, which end up having no definite features since they do not partake of the one. To get past the fourth deduction and return to the third, Socrates must give up his thesis in his long speech in Part I, and find a way to make sense of the idea that the one is both one and many.
6. Conclusion
On the reading of Part II that I propose, the nine deductions contribute to a single coherent argument with a powerful message. The exercise presents antinomies, but it also displays a striking progression, with later deductions criticizing or building on conclusions of previous ones.Footnote 36 Deduction 2 opens by reassessing the meaning of the original hypothesis that led to the dubious conclusions at the end of Deduction 1 (namely that the one neither is nor is one); the Appendix tries and fails to keep all the conclusions of Deductions 1 and 2 by arguing that both can be true at different times; and Deduction 3 in discussing the others indicates (it does not show, since it draws consequences for the others) that the one can be both one (in virtue of itself) and many (in relation to other things), a conclusion then undermined in Deduction 4. Socrates must correct the error in the fourth deduction to find his way back to the constructive third deduction.
The four deductions that make up the negative hypothesis follow a similar strategy but not the same pattern. Deductions 5 and 6 develop opposing views about not-being (that it is a feature of things like any other, and that it is the absence of features), followed in Deduction 7 by a remarkable picture of things appearing to be what they are not, a picture destroyed by Deduction 8. But unlike the fourth deduction of the positive hypothesis which challenges the audience to find a way back to the constructive third deduction, the eighth deduction of the negative hypothesis does not lead back to the seventh, but instead presses the interlocutor and audience to reject the negative hypothesis itself in favour of the positive hypothesis earlier abandoned. On this reading Part II develops one vast argument with all nine deductions intricately linked to one another.
In the transitional section to Part II, Parmenides says that there must be forms, if Socrates is to have anywhere to turn his thought, and if he is to preserve the capacity for dialectic (135b5-c3), a means of getting at the truth.Footnote 37 By going through the exercise in Part II we realize that there must be stable entities, and they must explain our world, and we see this truth because it is evident that the conclusion of Deduction 8 is not true: There is a world to be explained.
To avoid the outcome of the eighth deduction, Socrates must abandon the negative hypothesis giving rise to it and return to the positive hypothesis. To save the positive hypothesis he must abandon his thesis in Part I and instead agree that forms partake of other forms, and in the case of oneness that the one is both one and many. At the same time, he should not accept the thesis in the unbridled manner of Deduction 2 but in the judicious manner of Deduction 3 and recognize that an entity has its own character in virtue of itself and other characters through its participation in natures other than its own. What a thing is in virtue of itself gives it stability (thus avoiding the protean situation in Deduction 2) and puts constraints on what other features the entity has or can have. A form has many features, in addition to its own nature, and in some cases its own opposite, and those features are explained by natures external to it. Compresent opposites need not undermine the one’s explanatory function.
The distinction between what something is in virtue of itself (kath’ hauto) and what it is in relation to other things (pros alla) helps to preserve the explanatory role of forms, but not completely, because there remains a vexing problem about being. Can the one be one simply in virtue of itself? Deduction 1 argues that the one must partake of being even to have its own character, and Deduction 5 exploits the same idea by showing that an entity must partake of infinite shares of being to have any character at all, including its own. This is the problem of participation—the problem of being as a relational link between an entity and a property it has—and though a solution is hinted in Deduction 3, the problem is left unresolved in the Parmenides. To solve it, we must make sense of that most perplexing form, the form of being. The next round of the exercise needs to be about being, and in my view Plato presents the exercise across parts of the Theaetetus and Sophist, modeled on the treatment of the positive hypothesis about oneness in fours steps in the second part of the Parmenides. But that is a story for another occasion.Footnote 38
Acknowledgements: I would like to thank Sylvana Chrysakopoulou for organizing the stimulating and rewarding conference on the Parmenides in Chania, Crete, in 2011, as well as the participants for valuable discussion. Thanks to Benoît Castelnérac and Dialogue’s assistants for editing the volume. I particularly thank Constance Meinwald for constructive disagreement at the conference, and before and since: I have learned an enormous amount from her, and our differences have been invariably helpful to me. Thanks also to Sandra Peterson for penetrating comments on an earlier version of this paper.
