POLARITY AND ANALOGY TWO TYPES OF ARGUMENTATION IN EARLY GREEK THOUGHT BY G.E.R.LLOYD Fellow of King's College, Cambridge CAMBRIDGE AT THE UNIVERSITY PRESS 1966 PUBLISHED BY THE SYNDICS OF THE CAMBRIDGE UNIVERSITY PRESS Bentley House, 200 Euston Road, London N.W. 1 American Branch: 32 East 57th Street, New York, N.Y. 10022 African Office: P.M.B. 5181, Ibadan, Nigeria CAMBRIDGE UNIVERSITY PRESS 1966 Printed in Great Britain at the University Printing House, Cambridge {Brooke Crutchley, University Printer) LIBRARY OF CONGRESS CATALOGUE CARD number: 66-10042 CONTENTS Introduction page i part one: polarity l Theories based on opposites in early Greek thought 15 II The analysis of different modes of opposition 86 part tw o: analogy III The pre-philosophical background 172 IV Metaphor and imagery in Greek cosmological theories 210 V The role of comparisons in particular accounts 304 VI The analysis of argument from analogy 384 part three: conclusion VII The development of logic and methodology in early Greek thought 421 Bibliography 441 Index of Passages Quoted or Refened to 459 General Index 476 V INTRODUCTION AIMS AND PROBLEMS The aims of this study are to describe and analyse two main types of argument and methods of explanation as they are used in early Greek thought from the earliest times down to and including Aristotle, and to consider them, in particular, in relation to the larger problem of the development of logic and scientific method in this period. First I should say why the subject and the period I have chosen seem to me impor­ tant. It is arguable that natural science, cosmology and formal logic all originate (so far as the West is concerned) in Greek philosophy. But while it is convenient and quite justified to see these disciplines as having a definite historical beginning, in each case the influence of previous thought on those who were primarily responsible for initiating the new inquiry raises an intricate issue. The Presocratic philosophers may be credited with the first systematic attempts to give rational accounts of natural phenomena and of the universe as a whole. But both their scientific and their cosmological theories undoubtedly owe a certain debt to the common stock of pre-philosophical Greek beliefs. Aristotle is generally held to be the founder of formal logic as we know it, and this is evidently true in that it was Aristotle who introduced the use of symbols into logic, for example.1 But before Aristotle the dialogues of Plato deal with many problems which may rea­ sonably be considered problems of logic, particularly in con­ nection with the method of Dialectic. And we can trace a certain awareness of some logical and methodological issues earlier still, for it was in the pre-Platonic period that philo­ sophers and scientists first debated the respective merits of ‘reason’ and ‘sensation’ as sources of knowledge, for example, 1 See, for example, Cornford, 5, p. 264. Aristotle himself claimed originality for his work in logic at SE 183 b 34 ff. LPA INTRODUCTION and began to consider the question of the relation between a theory and the grounds or evidence in its support. The period of Greek philosophy which we shall be con­ sidering is marked by notable advances in both the theory and the practice of logic and scientific method. My main problems concern certain aspects of these developments. The first systematic exposition of a set of rules of argument is found in Aristotle, but a number of questions may be raised concerning the use of argument in earlier Greek writers and the circumstances in which various modes of argument came to be recognised and analysed. Even though modes of argument were not at first identified as such, we may never­ theless consider not only what arguments were commonly used in practice in early Greek literature for the purposes of inference or persuasion, but also how they were used; how far, in particular, is it possible to determine what assump­ tions were made concerning the cogency of different types of argument in the period before Plato? How far do pre- Platonic thinkers explicitly formulate the principles or assumptions on which their arguments are based? And how far did Plato carry the analysis of different types of argu­ ment? If we can detect certain developments in the assess­ ment of certain modes of argument in the period before Aristotle, we may also ask how far these developments influenced the actual arguments which were subsequently used, and whether the analysis of techniques of inference led to any major modification in the methods of argument which the philosophers and others employed. One topic that invites study is the use and development of modes of argument in the sixth to the fourth century. An­ other is the related question of the use and development of methods of explanation in the same period. Again it was Aristotle who (in the Posterior Analytics) put forward the first fairly complete theory of what may be called ‘scientific method5, but again certain questions may and should be asked concerning the earlier period. The Presocratic 2 INTRODUCTION philosophers and the medical theorists in the Hippocratic Corpus attempt to elucidate a wide variety of natural phenomena. How far is it possible to decide, either from their actual theories and explanations, or from their explicit pro­ nouncements on method, where these exist, what they ex­ pected of an ‘account’ of a natural phenomenon? As in the study of the development of modes of argument, three types of problem present themselves, first to identify the modes of explanation which were most commonly used in practice, second to trace the development of ideas on methodological problems, and third to analyse the interaction between theory and practice (where one may consider how far the ideas expressed by different Greek philosophers and scientists on the subject of method influenced either the types of theories they proposed or the way in which they attempted to establish them). The immediate purpose of this inquiry is to elucidate the use and development of certain modes of argument and methods of explanation, but I hope thereby to throw some light also on some of the broader issues of the place of informal logic in early Greek thought before the invention or discovery of formal logic. Indeed the broader issu«! raised by the problem of the nature of ‘archaic logic’ can, perhaps, only be discussed in terms of such concrete topics as the use of certain types of argument. In a series of works beginning with Les Fonctions mentales dans les sociétés inférieures (1910)1 Lévy- Bruhl developed the hypothesis of a ‘ pre-logicaT mentality, a mentality which is ignorant of the law of contradiction and which is ‘utterly mystical*, and as this hypothesis had a pro­ found influence on at least one generation of historians of early Greek philosophy, some comments should be made on it here. Gornford, for example, was much influenced by 1 See a\so La Mentalité primitive (1922) and VAme primitive (1927). One typical passage may be cited from La Mentalité primitiae (trans. L. A. Glare, ΐ923>Ρ·43δ): the primitive ‘will always seek the true came [i.e. of ‘natural phenomena’] in the world of unseen powers, above and beyond what we call Nature, in the “metaphysical” realm, using the word in its literal sense’. 3 INTRODUCTION Lévy-Bruhl’s theory in From Religion to Philosophy (1912), and Brunschvicg,1 Reymond,2 Rey,3 and Schuhl,4 in turn, pro­ vide ample evidence of the stimulus which Lévy-Bruhl’s ideas had on classical scholars interested in the problems of the development of logic and science in ancient Greece, even when they challenged certain features of his interpretation. More recently, Snell, in his influential book Die Entdeckung des Geistes,5 still attempted to distinguish between ‘mythical’ and ‘logical’ thought, using these two terms because (as he put it) ‘they effectively describe two stages of human thought’.6 The influence which Lévy-Bruhl’s hypothesis had is all the more remarkable considering the criticisms to which it was subjected from various quarters. Already in 1912 Durkheim contested the postulate of a ‘ pre-logical ’ mentality: ‘Ainsi, entre la logique de la pensée religieuse et la logique de la pensée scientifique il n’y a pas un abîme. L’une et l’autre sont faites des mêmes éléments essentiels, mais inégalement et différemment développés. Ce qui paraît surtout caractériser la première, c’est un goût naturel aussi bien pour les confusions intempérantes que pour les con­ trastes heurtés----- Elle ne connaît pas la mesure et les nuances, elle recherche les extrêmes; elle emploie, par suite, 1 L’Expérience humaine et la causalité physique (ist ed., 1922; 3rd ed. Paris, 1949), especially Book 4, chs. 9 and 10. 2 Histoire des sciences exactes et naturelles dans l’antiquité gréco-romaine (ist ed., 1924; 2nd ed., Paris, 1955), pp. n6ff. 3 La Science dans l’antiquité, vols. 1 and 2 (Paris, 1930, 1933), e.g. vol. i»pp. 434 a 4 Essai sur Information de la pensée grecque (ist ed., Paris, 1934). - In the 2nd edition, 1949, p. xiv and n. 4, Schuhl lists a number of other classical scholars ■who were influenced by Lévy-Bruhl and also notes that Lévy-Bruhl himself modified his thesis in the Carnets (see below, p. 5). 5 2nd ed., Hamburg, 1948, trans. T. G. Rosenmeyer, The Discovery of the Mind (Oxford, 1953), pp. 223 ff. 6 Snell went on to note that these two stages or types of human thought do not exclude one another completely, but he then described the difference be­ tween them, as he conceived it, in the following vague and seemingly confusing terms (3, p. 224): ‘Mythical thought requires receptivity; logic cannot exist without activity. Logic does not materialize until man has become cognizant of the energy within him, and the individuality of his mind. Logical thought is unimpaired wakefulness; mythical thinking borders upon the dream, in which images and ideas float by without being controlled by the will.* 4 INTRODUCTION les mécanismes logiques avec une sorte de gaucherie, mais elle n’en ignore aucun.’1 Moreover Lévy-Bruhl himself radically modified his position in his later writings. In a letter to Professor Evans-Piitchard (written in 1934 and pub­ lished in the British Journal of Sodology in 1952) he granted that the term ‘pre-logicaP was ‘rather unfortunate’, while in the Carnets (no. in, dated 1938, published in 1947 in Revue philosophique) we find the following note (p. 258) : Έη ce qui concerne le caractère “prélogique” de la mentalité primi­ tive, j’avais déjà mis beaucoup d’eau dans mon vin depuis 25 ans; les résultats auxquels je viens de parvenir touchant ces faits [i.e. beliefs reported from certain African societies] rendent cette évolution définitive, en me faisant abandonner une hypothèse mal fondée, en tout cas, dans les cas de ce genre.’ The hypothesis of a ‘ pre-logical ’ mentality has long been discredited,2 and it was eventually abandoned by its author himself. But some, at least, of the difficulties which that hypo­ thesis was invented, to solve remain real difficulties. No doubt it would be unwise to compare the problems facing the student of early Greek thought with those facing the anthro­ pologist too closely3 (and the methods of the two disciplines 1 Les Formes élémentaires de la oie religieuse (Paris, 1912), p. 342. 2 Writing in 1954, G. Lienhardt, for example, began a discussion of primitive modes of thought with the warning that ‘none of us who study savage societies would say, today, that there are modes of thought which are confined to primi­ tive peoples’ (in The Institutions of Primitive Society, ed. E. E. Evans-Pritchard, Oxford, 1954, p· 95)· The most recent full-length study of primitive thought is C. Lévi-Strauss’s La Pensée saunage (Paris, 1962), in which Lévi-Strauss shows with a great wealth of documentation how, for «cample, the members of primitive societies often observe and classify natural species with extraordinary accuracy and minute attention to detail. 3 The cautious judgements οζ among others, Schuhl (pp. 5 ff.) and Guthrie (4, pp. 18 ff.) on the subject of the use of comparative anthropology, provide an important corrective to the tendency to assume that material collected by the anthropologists necessarily affords the key to the understanding of ancient Greek beliefe. Yet in two contrats, in particular, it may prove very useful to consult such material: first, for a negative purpose, to counter a suggestion that some Greek belief is simply the result of a hypothetical ‘ universal tendency of human thought’,* and secondly, with a positive aim, to suggest possible lines of inquiry which we may follow in our attempts to elucidate what at first sight may seem incomprehensible beliefe or practices. 5 INTRODUCTION differ profoundly since the evidence for ancient Greek thought is almost entirely literary). But dissatisfaction with Lévy- BruhPs concept of a ‘pre-logicaT mentality may prompt both the anthropologist and the classical scholar to attempt to give a more adequate account of the informal logic implicit in primitive or archaic thought. Indeed the evidence of early Greek literature, limited though this is, is especially interest­ ing and valuable since it enables us to study not only the nature of the logical principles implicit in archaic beließ, but also the development of logic itself and the gradual recogni­ tion and analysis of those principles. In particular, the evidence for ancient Greek thought in the period down to Aristotle provides us with a unique opportunity to consider how far the invention or discovery of formal logic merely rendered explicit certain rules of argument which were implicitly observed by earlier writers, or to what extent the analysis of various modes of argument involved the modifica­ tion and correction of earlier assumptions. METHOD AND EVIDENCE The broad field of this inquiry is the modes of argument and forms of explanation of early Greek thought down to Aristotle, but the method of approach to this subject which I have adopted should be defined, and some remarks should be made on the nature of the evidence on which this study is based. First on the question of method. In dealing with the types of argument and explanation in early Greek thought two main methods seem to be possible, which might be called for want of better terms the analytic and the synthetic. The first would attempt a full description of the relevant texts, reaching general conclusions only at the end of an exhaustive survey of the evidence : this has the advantage of complete­ ness. The second method would offer preliminary generalisa­ tions as working hypotheses, as it were, which may, and indeed probably will, require modification in the light of the 6