The Cambridge Companion to Early Greek Philosophy

by A. A. Long

  4 On the positions of Love and Strife, see Guthrie [16] 179, O’Brien [359] 116–17, Graham [363] 308 n. 39, O’Brien [369] 418–21.

  5 Those rejecting a complete separation of the elements include Bollack [356] vol. 1; Hölscher [360]; Solmsen [361]; Long [362]; Schofield in KRS, 288 n. 1, 299–305; Osborne [364]. Those defending a complete separation include O’Brien [359] and [369]; Barnes [14] 308–11; Wright [358]; Graham [363]; Inwood [357]. If there is a complete separation of elements, there must be two periods of creation of animals and plants, one before and one after the separation, if there is no complete separation, one period of creation suffices. Much of the debate centres on how the various stages of creation identified in the fragments are to be located in the cosmic cycle. For the view that only Love is responsible for zoogony, see Broadie in this volume p. 216.

  6 On Pythagorean teachings on rebirth, see Huffman in this volume, pp. 69–71. For Pythagorean influences on Empedocles, see Kingsley [105].

  7 On Empedocles’ poetry, see Most in this volume, p. 356.

  8 Starting with Tannery [131] and Burnet [6], and followed for example, by Cornford [384] and Vlastos [392].

  9 The list of five derives from Kerferd [390]. A number of earlier studies identified several of the postulates.

  10 Unfortunately, it is not clear whether Anaxagoras is implying that no stretch of a stuff is like any other stretch of it, or only that no phenomenal object is like any other.

  11 For the development of an alternative to homoiomereity, see Graham [387].

  12 Kerferd [390]; Barnes [14] ch. 16; Graham [387].

  13 Barnes [14] ch. 3 has revived the Aristotelian view that some early Ionians, including Anaximenes, were “material monists,” that is, that they posited a basic substance (air, in the case of Anaximenes) which changed its qualities to produce the phenomena of other substances, but which was always present as an underlying principle for them. This view seems to be based on a misreading of the ancient evidence. See Heidel [388]; Cherniss [34] 362ff., esp. 371; Stokes [130] ch. 2; Graham [242] and appendix to this chapter; but to the contrary, see Sedley in this volume p. 123.

  14 For Anaximenes: Cicero De natura deorum 1.10.26, Aetius I.7.13; Heraclitus DK 22 B30, 64, 67.

  15 See Vegetti in this volume p. 273.

  16 The notion that he founded a school is likely to be anachronistic, but for convenience I shall refer to a group of like-minded philosophers by the traditional term of school.

  17 Plato, Tht. 180d-e, Soph. 242c-d; Aristotle, Phys. II.1, Metaph. I.5 986b 18-25.

  18 Plato, Soph. 242d-243a; Aristotle, Phys. I.1 184b18-22.

  19 On this see Mourelatos [309] 130–33; Barnes [14]; and Curd [287], Curd [290]. The term “one” appears with possible implications of monism only in B8.6 and B8.54; in both cases the implication seems tenuous at best. But see Sedley in this volume p. 120.

  20 Thus Mourelatos [309].

  21 The ancient tradition ascribes to most of the early philosophers the view that motion is eternal; however, no fragments express this view except possibly those of Heraclitus, and it seems likely that the principle is derived from Aristotle’s inference that eternal motion is presupposed by their view. Heraclitus, on the other hand, does seem to stress the eternity of process and could provide the dialectical setting for Parmenides’ reaction. Previously it was thought (by Paul Tannery [131] 232–47 followed by Burnet [6] 183ff., 314–15, Cornford [285] ch. 1, and Raven [226]) that Parmenides was reacting to certain Pythagorean views. But no trace of the alleged Pythagorean views has been found (Vlastos [229] 376–77).

  22 On the pluralists as responding to Parmenides’ radical critique, see KRS 351; Barnes [14] 313–17. On the alleged success of the atomist reply, see KRS 433 (almost unchanged from the first edition). On the failure of the whole pluralist project, see Barnes [14] 441–42.

  23 Raven (KRS 358-59) argues that Anaxagoras B1 marks a pointed rejection of Parmenidean monism, timelessness, and indivisibility. But that interpretation depends crucially on how Anaxagoras read Parmenides, a problem we shall discuss below.

  24 Aristotle portrays Leucippus’ atomism as a reaction to the Eleatics, GC I.8 325a2 ff. But as KRS, 409 n. 4, notes, the positions he is said to react to are those of Melissus, not Parmenides. At Aristotle GC I.2 316a13ff., Democritus is shown arriving at his principles by a reasoned rejection of Zenonian positions. These interpretations suggest that atomism is a response to the second generation of Eleatics. And if Zeno and Melissus force a rethinking of neo-Ionian principles, it may be because they are responsible for the view that Parmenides is a strict monist, allowing only one single entity to exist.

  25 Thus Mourelatos [118] 128–30.

  26 On different readings of B8.53-54, see Mourelatos [309] 80–85. The passage resists easy interpretation even with the best tools of philology.

  27 Metaph. 1.5 986b31 ff.

  28 For a reading that seems to me to come closer to Parmenides’ intentions, see Long [304]. For the general position that the pluralists are trying to follow Parmenides’ lead in the Way of Opinion, see now Curd [290] and A. Finkelberg, “Xenophanes’ physics, Parmenides’ doxa and Empedocles’ theory of cosmogonical mixture,” Hermes 125 (1997) 1–16. Curd asserts the strong view that the pluralists are correct in their reading of Parmenides. Finkelberg traces the dualism of Parmenides’ cosmology back to Xenophanes.

  29 As Aristotle takes the passage, Metaph. I.5 986b33-987a2.

  30 Mourelatos [309], esp. 134–35, defends a reading of this type.

  31 See Appendix, p. 176.

  32 Plato, Euthyphro 11a, Aristotle, Cat. chs. 2, 4, 5.

  33 See D. W. Graham, “Aristotle’s discovery of matter,” AGP 66 (1984) 37–51.

  34 On the relation of Zeno’s arguments to Parmenidean doctrine, see McKirahan in this volume, pp. 134, 156; for Melissus, see Sedley, pp. 130–31.

  35 B17.34, 21.13, 26.3.

  36 Cf. Aristotle Phys. VIII.9 265b17 ff.

  37 A crucial question that arises in this context is, just what are Anaxagoras’ seeds? Furley [385] 72–75 argues that they are biological seeds from which living things grow. If that is the case, then even plants and animals are latent in the primeval chaos. But there are many other interpretations of the seeds, and Anaxagoras’ attention to their shapes (?) (ideai), colours, and savours (B4) at least tends to suggest that he is interested in them as loci of phenomenal qualities rather than as sources of biological generation. In any case, at least all stuffs are latent in the original mixture, and if Furley is right, even biological species are latently present.

  38 The term “Eleatic pluralism” has been applied to the early atomists by R.B.B. Wardy, “Eleatic Pluralism,” AGP 70 (1988) 125–46; I wish to apply it especially to Empedocles and Anaxagoras, reserving the possibility that the atomists, for reasons Wardy does not consider, may in some important respect be anti-Eleatic, for example, in Democritus’ claim that “thing is no more than no-thing” (DK 68 B156). In some sense surely the atomists too are Eleatic pluralists, but not necessarily as thoroughly, unabashedly, perhaps ingenuously as their predecessors, because they have faced criticisms from later members of the Eleatic school.

  C.C.W. TAYLOR

  9 The atomists

  Atomism was the creation of two thinkers of the fifth century B.C., Leucippus and Democritus. The former, attested by Aristotle, our primary source, as the founder of the theory, was a shadowy figure even in antiquity, being eclipsed by his more celebrated successor Democritus to such an extent that the theory came to be generally regarded as the work of the latter. Epicurus, who developed and popularised atomism in the late fourth and early third centuries B.C. (following in the tradition of various figures such as Nausiphanes and Anaxarchus, now little more than names), went so far as to deny that Leucippus ever existed. Only a little more is known about Democritus (see p. xix). The precise relation between Leucippus and Democritus is unclear. Plato never mentions either by name. Aristotle and his followers treat Leucippus
as the founder of the theory, but also assign its basic principles to both Leucippus and Democritus; later sources tend to treat the theory as the work of Democritus alone. While it is clear that the theory originated with Leucippus, it is possible that the two collaborated to some extent and almost certain that Democritus developed the theory in a number of areas, for example, extending it to include a materialistic psychology, a sophisticated epistemology, and an account of the development of human society that laid particular stress on the human capacity to learn from chance experience.1

  PHYSICAL PRINCIPLES

  According to Aristotle (GC I.7-8 324a35-325a31), the atomists attempted to reconcile the observable data of plurality, motion, and change with the Eleatic denial of the possibility of coming to be or ceasing to be. Like Anaxagoras and Empedocles, they postulated unchangeable primary things, and explained apparent generation and corruption by the coming together and separation of those things. But their conceptions of the primary things and processes differed radically from those of Anaxagoras and Empedocles. For Anaxagoras the primary things were observable stuffs and properties, and for Empedocles they were the elements, earth, air, fire, and water: for both, the primary processes were mixing and separation of those primary things. By contrast, for the atomists the primary things were not properties or stuffs but physical individuals, and the primary processes were not mixing and separation but the formation and dissolution of aggregates of those individuals. Again, the basic individuals were unobservable, in contrast with the observable stuffs of Anaxagoras and the observable elements of Empedocles. Consequently, their properties could not be observed but had to be assigned to those individuals by theory.

  Since the theory had to account for an assumed infinity of phenomena, it assumed an infinite number of basic individuals, while postulating as few explanatory properties as possible, specifically shape, size, spatial ordering, and orientation within a given ordering.2 All observable bodies are aggregates of basic individuals, which must therefore be too small to be perceived.3 These basic corpuscles are physically indivisible (atomon, literally uncuttable), not merely in fact but in principle; Aristotle reports (GC I.2 316a14-b7) an (unsound) atomistic argument, which has some affinities with one of Zeno’s arguments against plurality (DK 29 B2), that if (as e.g., Anaxagoras maintained) it were theoretically possible to divide a material thing ad infinitum, the division must reduce the thing to nothing. This argument was supported by another for the same conclusion; atoms are theoretically indivisible because they contain no void. On this conception bodies can split only along their interstices; hence, where there are no interstices, as in an atom, no splitting is possible. (The same principle probably accounted for the immunity of the atoms to other kinds of change, such as reshaping, compression, and expansion. All were probably assumed to require displacement of matter within an atom, which is impossible without any gaps to receive the displaced matter.) It is tempting to connect the assumption that bodies can split only along their interstices with the Principle of Sufficient Reason, to which the atomists appealed as a fundamental principle of explanation – arguing, for instance that the number of atomic shapes must be infinite, because there is no more reason for an atom to have one shape than another (Simplicius, In phys. 28.9-10).4 Given the total homogeneity of an atom, they may have thought, there could be no reason why it should split at any point, or in any direction, rather than any other. Hence by the Principle of Sufficient Reason, it could not split at all.

  The programme of reconciling the data of perception with the demands of Eleatic theory led the atomists to posit a void or empty space (a) as that which separates atoms from one another and (b) as that in which they move. Parmenides had argued (DK 28 B.22-25) that there could not be many things if there were no void to separate them, and Melissus had argued (DK 30 B7) that there could be no motion without a void into which the moving object moves; Aristotle attests that the atomists accepted both theses (Phys. IV.5 213a32-34, GC I.8 325a27-28). To the question what it is that separates atoms from one another, and into which they move, their answer was simply “nothing.” “what is not” or “the empty,” which they appear to have treated as interchangeable terms. They did not, then, shrink from the conclusion that what is no more is than what is not (Aristotle, Metaph. I.4 985b8; Plutarch, Adv. Col. 1108f).5 But the assertion that what separates distinct objects is nothing leads straight to incoherence; either there is nothing which separates those objects, in which case they are not separate from one another, or there is something which separates them, in which case “nothing” is the name of something.

  We have no idea whether this challenge was actually put to the atomists, or if it were, how they might have met it. The most we can offer is the following suggestion of an appropriate defence. There is indeed something which separates any two nonadjacent atoms, namely an interval. But an interval is not any kind of thing: it is merely a gap, an absence of anything. So there are indeed gaps between atoms, but gaps are nothings, and when an atom moves, it moves into a gap. But that can hardly be the whole story. For the notion of an interval or gap between objects presupposes a continuous dimension in which the objects and the interval between them are alike situated. That is to say, the atomists’ conception of the void cannot have been merely that of the nonbeing of a physical object; it was at least that of a gap in space, where space is conceived, however inchoately, as a continuous dimension. The atomists also claimed that the void is infinite in extent and used the term “the infinite” as another designation of it; this is most naturally interpreted as the claim that empty space is infinite in extent. They believed, then, that the universe consists of an infinitely large collection of indivisible physical objects (atoms) moving in infinite space, where space is a three-dimensional continuum of which any part may be either occupied or unoccupied.6

  In this empty space the atoms are in a state of eternal motion. This motion is not the product of design, but is determined by an infinite series of prior atomic interactions7 (whence two of Aristotle’s principle criticisms of Democritus, that he eliminated final causation (GA V.8 789b2-3) and made all atomic motion “unnatural” (De caelo III.2 300b8-16)8). The theoretical role of the void in accounting for the separation of atoms from one another has an interesting implication that is recorded by Philoponus (In phys. 494.19-25, In GC 158.26-159.7). Since atoms are separated from one another by the void, they can never strictly speaking come into contact with one another. For if they did, even momentarily, there would be nothing separating them from one another. But then they would be as inseparable from one another as the inseparable parts of a single atom, whose indivisibility is attributed to the lack of void in it (see above); indeed, the two former atoms would now be parts of a single larger atom. But, the atomists held, it is impossible that two things should become one. Holding atomic fusion to be theoretically impossible, and taking it that any case of contact between atoms would be a case of fusion (since only the intervening void prevents fusion), they perhaps drew the conclusion that contact itself is theoretically impossible.9 Hence what appears to be impact is in fact action at an extremely short distance. Rather than actually banging into one another, atoms have to be conceived as repelling one another by some sort of force transmitted through the void. Again, though no source directly attests this, the interlocking of atoms, which is the fundamental principle of the formation of aggregates, is not strictly speaking interlocking, since the principle of no contact between atoms forbids interlocking as much as impact. Just as impact has to be reconstrued as something similar to magnetic repulsion, so interlocking has to be reconstrued as quasi-magnetic attraction. If this suggestion is correct (and it is fair to point out that no ancient source other than Philoponus supports it) it is a striking fact that, whereas the post-Renaissance corpuscular philosophy that developed from Greek atomism tended to take the impossibility of action at a distance as an axiom, the original form of the theory contained the a priori thesis that all action is action at a distance. C
onsequently that impact, so far from giving us our most fundamental conception of physical interaction, is itself a mere appearance that disappears from the world when the description of reality is pursued with full rigour.10

  CHANCE AND NECESSITY

  While the broad outlines of the views of the atomists on these topics can be fairly readily reconstructed, there is much obscurity about the details. The atomists’ universe is purposeless, mechanistic, and deterministic; every event has a cause, and causes necessitate their effects.11 Broadly speaking the process is mechanical; ultimately, everything in the world happens as a result of atomic interaction. The process of atomic interaction has neither beginning nor end, and any particular stage of that process is causally necessitated by a preceding stage. But exactly how the atomists saw the process as operating is obscure. This obscurity is largely attributable to the fragmentary nature of the evidence that we possess, but perhaps the statement of the theory itself was not altogether free from obscurity.

  The fundamental text is the single fragment of Leucippus (DK 67 B1) “Nothing happens at random, but everything from reason and by necessity.” The denial that anything happens “at random” (matên) might well be taken in isolation to amount to an assertion that all natural events are purposive, since the adverb and its cognates frequently have the sense “in vain” (i.e., not in accordance with one’s purpose) or “pointlessly.” If that were the sense of not matên then “from reason” (ek logou) would most naturally be understood as “for a purpose.” These renderings are, however, very unlikely. The majority of the sources follow Aristotle (GA V.8 789b2-3) in asserting that Democritus denied purposiveness in the natural world, explaining everything by mechanistic “necessity.” A reading of Leucippus which has him assert, not merely (contra Democritus) that some, but that all natural events are purposive, posits a dislocation between the fundamental world-views of the two of such magnitude that we should expect it to have left some trace in the tradition. Moreover, the attribution of all events to necessity, a central feature of the mechanistic Democritean world-view, is itself attested in the fragment of Leucippus. We ought, then, to look for an interpretation of the fragment that allows it to be consistent with Democritus’ denial of final causation.