120 100 21MB
English Pages 254 [272] Year 1997
DECAYED GODS
STUDIES IN GREEK AND ROMAN RELIGION EDITED BY H.S . VERSNEL IN CO-OPERATION WITH F.T. VANSTRATEN
VOLUME 7
DECAYED GODS ORIGIN AND DEVELOPMENT OF GEORGES DUMEZIL'S "IDEOLOGIE TRIPARTIE"
BY
WOUTER W. BELlER
E.J. BRILL LEIDEN • NEW YORK • K0BENHAVN • KOLN 1991
The paper in this book meets the guidelines for permanence and durability of the Committee on Production Guidelines for Book Longevity of the Council on Library Resources.
Library of Congress Cataloging-in-Publication Data Belier, Wouter W. [Feiten, fouten en fabels. English] Decayed gods : origin and development of Georges Dumezil's "ideologie tripartie" I by Wouter W. Belier. p. cm.-(Studies in Greek and Roman religion, ISSN 0169-9512 ; V. 7) Translation of: Feiten, fouten en fabels, originally presented as the author's thesis (Ph. D.)-Leiden. Includes bibliographical references and index. ISBN 90 04 09487 3 (alk. paper) 1. Indo- Europeans-Religion. 2. Indo- Europeans-Civilization. 3. Dumezil, Georges, 1898I. Title. II. Series. BL660.B4513 1991 291' .089'034-dc20 91-18815 CIP
ISSN ISBN
0169-9512 90 04 09487 3
© Copyright 1991 by E.]. Brill, Leiden, The Netherlands All rights reserved. No part of this book may be reproduced or translated in any form, by print, photoprint, microfilm, microfiche or any other means without written permission from the publisher Authorization to photocopy items for internal or personal use is granted by E.]. Brill provided that the appropriate fees are paid directly to Copyright Clearance Center, 27 Congress Street, SALEM MA 01970, USA. Fees are subject to change. PRINTED IN THE NETHERLANDS
CONTENTS Preface ......................................................................................................................... xi Abbreviations ............................................................................................................ xiv 1 INTRODUCTION ................................................................................................. 1 Indo-Europeans ....................................................................................................... 1 The term "Indo-European" ............................................................................... 1 Indo-European "Urheimat" and culture .......................................................... 3 Dumezil's view of the "Urheimat" and Proto-Indo-European culture ....... .4 Tripartition: a general introduction ...................................................................... 7 The pre-1930 period............................................................................................ 7 "La prehistoire indo-iranienne des castes" ...................................................... 8 "La prehistoire des flamines majeurs" ............................................................. 9 The system: a synopsis ...................................................................................... 10 The "sociale" tripartition ............................................................................. 10 General principle of classification .............................................................. 12 Tripartition in the pantheon ........................................................................ 13 The discussion ........................................................................................................ 15 Indifference ........................................................................................................ 15 Acclaim ............................................................................................................... 16 The pioneers .................................................................................................. 16 Academic respectability ............................................................................... 16 Elaboration .................................................................................................... 17 Criticism ............................................................................................................. 17 Criticism and rejection of (secondary) interpretations ............................ 17 Criticism of the conclusion ........................................................................... 19 Criticism by providing better models .......................................................... 19 Criticism of the explicit or implicit political implications ........................ 19 Conclusion .............................................................................................................. 20
VI
CON1ENTS
2 METHODOLOGY .............................................................................................. 21 The methodological concepts .............................................................................. 21 Typological versus genetic................................................................................ 21 "Ensembles"....................................................................................................... 21 Myth as an entity ........................................................................................... 22 The world of the gods as an entity............................................................... 24 Religion as an entity ...................................................................................... 26 "Centre propre" ................................................................................................ 27 "Ideologie, conception, fonction" ........................................................... ,....... 28
Summary: The methodological concepts .............................................................. 31
The auxiliary hypotheses ....................................................................................... 31 Auxiliary hypotheses preceding the origin of the tripartite model............. .31 Auxiliary hypotheses in connection with the tripartite model ..................... 33 Historicisation ............................................................................................... 34 Summary: Historicisation .............................................................................. 43 . shift: t he "gl"tssement guerrter . " ..........................................44 The G ermaruc Summary: The Germanic shift: the "glissement guerrier" ........................... 46 The reforms of Zarathustra ......................................................................... 46
Summary: The reforms of Zarathustra ........................................................ .49 The Celtic shift .............................................................................................. 50 The Indian shift ............................................................................................. 50 Disappearance of the functional structure ................................................. 51
Summary: The Celtic shift; The Indian shift; Disappearance of the functional structure .................................................... 52 Summary: The auxiliary hypotheses ...................................................................... 52 The paradigmatic framework ............................................................................... 52 Culture/the supernatural .................................................................................. 53 First response ................................................................................................ 53 Second response ............................................................................................ 54 Third response ............................................................................................... 55
Summary: Culture/the supematural .................................................................. 55
Conscious/unconscious ..................................................................................... 56
Summary: Conscious/unconscious ................................................................... 57
CONTENTS
vii
3 THE TRIPARTITE SYSTEM ........................................................................... 58 Tripartition in India............................................................................................... 58 Social tripartition ............................................................................................... 58 Tripartition in the pantheon............................................................................. 59 Summary: Tripartition in India ............................................................................. 67 Tripartition in Iran ................................................................................................ 67 Social tripartition in ancient Iran .................................................................... 67 Scythian tripartition .......................................................................................... 68 Summary: Scythian tripartition ......................................................................... 71 Tripartition among the families of the Nartes ............................................... 72 Summary: Tripartition among the families of the Nartes ................................ 75 Tripartition in the "pantheon" ......................................................................... 75 The post-Zoroastrian period ....................................................................... 75 The Amesha Spentas .................................................................................... Summary: Tripartition in the ''pantheon" ........................................................ 79
n
Tripartition in Rome ............................................................................................. 79 Social tripartition in ancient Rome ................................................................. 79
Summary: Social tripartition in ancient Rome ................................................. 86
Jupiter, Mars and Quirinus .............................................................................. 86
Summary: Jupiter, Mars and Quirinus .............................................................. 95
Romulus and his companions and the Roman kingship cycle ..................... 95
Summary: Romulus and his companions and the Roman kingship cycle .................................................................................................... 100
Tripartite German material ................................................................................ 101 Odhinn, Thorr and Freyr ................................................................................ 101 The problem ................................................................................................ 101 The "glissement militaire" .......................................................................... 102 The gods ofUppsala ................................................................................... 103 The curse of Egill ........................................................................................ 105 The sons of Mannus .................................................................................... 106 Other evidence and "bilan" ....................................................................... 107 Summary: Odhinn, Thorr and Freyr ............................................................... 109 Social tripartition ............................................................................................. 109 Celtic tripartition ................................................................................................. 110 Social tripartition ............................................................................................. 110 The Tuatha De Danann .................................................................................. 112 Greek tripartition ................................................................................................ 113
Summary: Celtic tripartition; Greek tripartition ................................................. 113
viii
CONTENTS
4 THEMITRA-VARUNACONCEPT ............................................................ 114 Introduction ......................................................................................................... 114 Mitra- V aruna .................................................................................................... 116 Summary: Mitra - Varuna ................................................................................... 121 Mithra- Ahura Mazda ....................................................................................... 123 Summary: Mithra - Ahura Mazda ...................................................................... 125 Vohu Manah- Asha ........................................................................................... 126 Summary: Vohu Manah -Asha .......................................................................... 129 Romulus- Numa ................................................................................................. 130 Summary: Romulus - Numa ............................................................................... 135 Jupiter- Dius Fidius/Fides ................................................................................ 135 Summary: Jupiter- Dius Fidius/Fides ................................................................ 138 Odhinn- Ullr/Mithotyn!I'yr .............................................................................. 139 Odhinn/Othinus- Ullr/Ollerus, Mithotyn ................................................... 139 Odhinn- Tyr .................................................................................................... 141 Summary: Odhinn - Ullr/Mithotyn/Tyr .............................................................. 143 Greek bipartition ................................................................................................. 144 SUMMARY: THE MITRA-VARUNA CONCEPT ......................................... 144
5 BIPARTITAMINORA ..................................................................................... 146
"Le Borgne"- "Le Manchot" ............................................................................ 146 Odhinn- Tyr .................................................................................................... 146 Horatius Codes- Mucius Scaevola .............................................................. 148 The Celtic "Le Borgne" - "Le Manchot" .................................................... 151 Bhaga and Savitr .............................................................................................. 153 Cyclops - 100-handed giants .......................................................................... 154 Taxrnoruw- Ahriman ..................................................................................... 155 Summary: "Le Borgne"- "Le Manchot" ........................................................... 156 Concepts related to the bipartition ................................................................... 156 The "polytechnicien" versus the "grand magicien" .................................... 156 The "mauvais roi temporaire" ............................................ 1•••••••••••••••••••••••••• 157 The "grimace herolque" ................................................................................. 158 Summary: Concepts related to the biparlition .................................................... 159
CONTENTS
ix
Theoretical precursors of the bipartition ......................................................... 159 Le probleme des Centaurs ............................................................................. 159 Iran ................................................................................................................ 160 India .............................................................................................................. 161 Greece .......................................................................................................... 162 Rome ............................................................................................................ 162 Conclusion ................................................................................................... 163 Ouranos - V aruna ........................................................................................... 163 Flamen - Brahman .......................................................................................... 165 Jeunesse, eternite, aube .................................................................................. 166 Relation between the theoretical precursors and the MN concept ......... 167 Regnum- interregnum = Mitra - V aruna ............................................. 167 Brahman- raj = Mitra- Varuna ............................................................. 169 Summary: Theoretical precursors of the bipartition ........................................... 172 The exclusiveness of the bipartition .................................................................. 173
Summary: The exclusiveness of the bipartition .................................................. 176
6 TRIPARTITAMINORA .................................................................................. 1n "Aesir"- "Vanir" ................................................................................................ 177 Introduction ..................................................................................................... 177 The Germanic and Roman conflicts ............................................................. 177 The Tuatha De Danann and the Fomoire .................................................... 189 Indra and the Nasatyas ................................................................................... 191 The conflict among the Nartes ....................................................................... 194 Summary: "Aesir"- "Vanir" ............................................................................... 196 Indo-European eschatology ............................................................................... 198 Introduction ..................................................................................................... 198 The interpretation of the Mahabharata ........................................................ 198 The "suite romaine" ........................................................................................ 210 Summary: Indo-European eschatology .............................................................. 217 Goddesses............................................................................................................. 217
Summary: Goddesses ........................................................................................... 226 7 EVALUATION .................................................................................................. 228 Verifiability........................................................................................................... 229 Verification ...................................................................................................... 229 Falsification ...................................................................................................... 232
X
CONTENTS
Precision and inconsistencies ............................................................................. 233 Precision ........................................................................................................... 233 Inconsistencies ................................................................................................. 237 Conclusion ............................................................................................................ 239 SELECfED BIBLIOGRAPHY ........................................................................... 241 INDEX ...................................................................................................................... 247 List of figures: Figure 1: Two models for the resemblances between the IndoEuropean languages ................................................................................ 2 Figure 2: The "sociale" tripartition ...................................................................... 12 Figure 3: Tripartition in the pantheon ................................................................. 14 Figure 4: The Roman and the Indian "champs ideologiques" ......................... 35 Figure 5: The "pantheon" in the post-Zoroastrian period ................................ 76 Figure 6: The Amesha Spentas ............................................................................. 78 Figure 7: The socio-ethnic structure of Rome (according to Varro) .............. 80 Figure 8: Career of Romulus and the kingship cycle ....................................... 100 Figure 9: The Germanic shift and the Rigsthula .............................................. 104 Figure 10: Tripartition in the Rigsthula ............................................................... 110 Figure 11: Celtic social tripartition ...................................................................... 111 Figure 12: Mitra- Varuna {1952) ........................................................................ 119 Figure 13: Mitra- Varuna {1958 and 1968) ........................................................ 120 Figure 14: Mitra- Varuna {1977) ........................................................................ 122 Figure 15: The Germanic and the Celtic king..................................................... 172 Figure 16: "Aesir"- "Vanir" conflict .................................................................. 197 Figure 17: The generation of Vicitravirya and the ascending generation ....... 198 Figure 18: Descendants of Vicitravirya ............................................................... 199 Figure 19: Descendants of Pandu ........................................................................ 199 Figure 20: The Pandavas as a functional group .................................................. 200 Figure 21: "Dieux souverains mineurs" ............................................................... 204 Figure 22: Indo-Iranian "dieux souverains mineurs" ......................................... 205 Figure 23: Roman "dieux souverains mineurs" .................................................. 206 Figure 24: Tripartition in the Roman republic ................................................... 212 Figure 25: Mahabharata and the "suite romaine" .............................................. 216 Figure 26: The "auxiliaires exterieurs" ................................................................ 216
PREFACE If the science of culture is to be scientific, it has to be based on theory. Theories
explain the data and allow for predictions which are in theory verifiable. The field of application of these theories may be (components of) a single culture or more than one culture, distributed in space and time. It is even possible to come up with explanations for the cultural behaviour of "humanity''. The larger the field of application becomes, the more difficult it becomes to test the theory. It is easier to critically assess and verify theories which offer an explanation of a specific problem within confmed spatial and temporal parameters than theories which have "humanity'' as their object. The theory of the "ideologie/conception tripartie des Indo-Europeens" developed by Dumezil between 1938 and 1986 is situated somewhere between these extreme poles of generality and particularity. On the one hand, Dumezil's theory is intended to provide an explanation for data drawn from an extremely wide range of material: the hypothetical Proto-Indo-European mother culture and its many descendants (the various Indo-European cultures) over a period of some five millennia; on the other hand, Dumezil regularly stresses the scientific and empirical character of his theory. His oeuvre consists of some 17,000 pages. Besides publications on such themes as the Caucasian languages, he has developed a theory based on this "conception tripartite" in which he qualifies Proto-Indo-European culture in terms of a division of society and the pantheon into three groups with different functions: (religious and juridical) sovereignty, military force and fertility or prosperity. Dumezil used this theory to indicate the traces of a tripartite conception in a number of cultures belonging to the Indo-European linguistic family. For example, he connected the three Indian varnas (classes): the brahmans (the priestly class), the kshatriyas (the military aristocracy) and the vaishyas (the classes of farmers and traders) with a triad of gods from the Roman pantheon associated with the archaic flamines maiores: Jupiter, Mars and Quirinus. Dumezil also claimed to have found the traces of such a tripartite conception in the Celtic, Germanic, Iranian and Greek traditions. None of the supporters or opponents in the history of the reception of Dumezil's theory has attempted a (critical) reconstruction of the genesis and development of this theory to date. A reconstruction of this kind is not necessarily a prerequisite for an assessment of the value of the theory, since the "context of development" and the "context of justification" can be considered independently of one another. In the present case, however, we shaH see that this reconstruction is of great importance for the assessment of the value of the
xii
PREFACE
theory. Hence the primary objective of the present investigation is a critical reconstruction of the theory in question. The second objective of the investigation is to arrive at a critical assessment and evaluation of this theory. We cannot claim to be able to determine the correctness or incorrectness of the theory as a whole, because of the complexity and disparity of the ensemble of hypotheses, research objects, concepts, models and material covered (by the latter I mean the "factual material" which can be understood with the assistance of the theory) and the number of cultures covered by the theory. It is the extent to which the theory is usable (operational) that will be assessed. Once the parameters for making this assessment have been determined, a comparison can be made with the model. Although Dumezil has not founded a school, various researchers have used his theory as the starting point for their own scientific activity. While their work is influenced by Dumezil, they have had little or no effect on the development of his theory. They will therefore not be taken into account in the present monograph. In the introductory chapter (chapter 1) I shall discuss a number of concepts which are presupposed in a consideration of the theory. The chapter also contains a brief summary of the theory to serve as a guide during the critical reconstruction, and an outline of the discussion of the theory which has taken place and its scientific importance. This is followed by the critical reconstruction of the genesis and development of the theory, combining two different approaches. Firstly, in chapter 2 I present a collection and analysis of statements by Dumezil on methodology. Which concepts does he use; what are his explicit theoretical presuppositions; which auxiliary hypotheses does he posit to account for a possible gap between theory and fact; and what does he have to say about the paradigmatic framework he uses? Chapters 3 to 6 deal with the interpretation of the data. What is the material covered? It can be described in two ways. There is the possibility of a strictly chronological account. Changes in the interpretation of the material which Dumezil takes from one culture at a given moment result in simultaneous changes in the interpretation of similar material which Dumezil takes from other cultures. An account of this kind would require the simultaneous analysis of the model from one moment to another over its entire scope and in every ramification, resulting in a clear view of the development of the theory as a whole but a limited view of the components. The alternative is a thematic account, in which the various themes are dealt with on an individual basis and are traced from beginning to end within their own subcontext. This method provides a good view of the components themselves, but it cannot demonstrate the connections between the changes. I have opted for a combination of the two methods. The presentation is divided thematically into a number of chapters. Thus chapter 3 deals with the tripartite series. Chapters 4 and 5 deal with the opposition within the first function; chapter 4 with the Mitra-Varuna concept and chapter 5 with related themes, the "bipartita minora". Chapter 6 deals with
PREFACE
xiii
the "tripartita minora". First an account is given of the war between the representatives of the first and second functions against the representatives of the third function; secondly of the narratives which later turned out to have an origin in Indo-European eschatology (according to Dumezil); thirdly of Dumezil's view of the goddesses. Within these chapters the material is divided into cultural units, within which I have alternated between a chronological and a thematic account in the interests of clarity. Finally, chapter 7 contains a discussion of the parameters to be used, followed by an evaluation. The present volume is a translation of my Leiden Ph.D. thesis (in Dutch): "Feiten, fouten en fabels. Ontstaan en ontwikkeling van Dumezils "ideologie tripartie" ". It owes its existence and completion to the stimulating assistance provided by my supervisors, Prof. L.L. Leertouwer and Prof. H.S. Versnel. I would also like to thank the others involved in the conferment of the doctor's degree (Prof. R.S.P. Beekes, Prof. J.C. Heesterman, Dr. J.M.C. Kroesen, Dr. J.G. Oosten en Prof. K. van der Toorn) for their contributions. I am particularly grateful for Henk Versnel's encouragement in editing the text for the English edition. This study is the fruit of years of work and it would be impossible to thank all those who have helped in one way or another. However, I cannot fail to mention my teacher, Prof. B. Hartmann. The cooperation with Peter Mason, the translator, combined business with pleasure. Noor Evertsen and Erik-Jan Starn put a lot of time into reading the original text and played an important part when I was writing it. The manuscript of the Dutch dissertation and of the present English edition were both converted into camera-ready copy by Sjoerd and Frans Schalekamp. Financial assistance was provided by the following bodies: Stichting Dr Hendrik Muller's Vaderlandsch Fonds; Stichting bet Scholten-Cordes Fonds and the M.A.O.C. Gravin Van Bylandt Stichting. The final word of thanks is to my wife Huiberti Lakeman, whose support on the emotional front made the progress of the research project possible. Leiden, 1990 Wouter Belier
ABBREVIATIONS AA ACIR AESC AHES AL
AO
ARB BGdSL BSL BSOAS CAn CPT CRev
cs
EC
FPC
GopB liTh Ill
JA JAOS JAS
Jer. JIBS
JMS
JP JRS MLP
MN
NC
NRF RB REA
REL
RES
RHA
RHR RP RPh
RV
ShB SHR SL SMSR
svs
TaitB TaitS TCAAS TCF
TQR
American Anthropologist Atti dell 'VIII Congresso Intemazionale de Storia delle Religioni Annales Economie Socittt Civilisations Annales d'Histoire Economique et Sociale Acta Linguistica Acta Orientalia Academie Royale de Belgique, Bulletin de Ia Classe des Lettres et des Sciences Morales et Politiques Beitriige zur Geschichte der Deutschen Sprache und Literatur Bulletin de Ia Soci~t~ de Linguistique de Paris Bulletin of the School of Oriental and African Studies Les Cahiers pour I'Analyse Cahiers "Pour un Temps" Classical Review Cahiers de Sud Etudes Celtiques Folklore Fellows Communications Gopatha Brahmana History and Theory Indo-Iranian Journal Journal Asiatique Journal ofthe American Oriental Society Journal ofAsian studies Jeremiah Journal of Indo-European Studies Journal ofMithraic Studies Journal de PlYChologie Journal ofRoman Studies Memoires de Ia Soci~t~ de linguistique de Paris Mitra/Varuna Nouvelle Clio La Nouvelle Revue Franfaise Religion och Bibel Revue des Etudes Armeniennes Revue des Etudes Latines Revue des Etudes Slaves Revue Hittite et Asianique Revue de l'Histoire des Religions Revue de Paris Recherches Philosophiques
Rig Veda
Shatapata Brahmana Studies in the History ofReligions (Supplements to: "Numen ") Studia Linguistica Studi e Materiali di Storia delle Religioni Sagabook of the Vzking Society Taittirfya Brtlhmana Taittiriya Samhita Transactions ofthe Connecticut Academy ofArts and Sciences Travaux du premier Congres international de Folklore The Quarterly Review
ABBREVIATIONS
VSamh ZfEth
VIJjasaneya SamhitiJ
Zeitschrift for Ethlwlogie
XV
CHAPTER ONE
INTRODUCITON What follows is a brief introduction to what is meant by the term "Indo-Europeans", to the theory of tripartition and the history of the reception of this theory. Indo-Europeans
After commenting on the term "Indo-European" I shall discuss the commonest responses to the question of the site of the original Indo-Europeans and the views of the nature of this Proto-Indo-European culture, concluding with a comparison between these responses and Dumezil's standpoint. The tenn "Indo-European"
The theory of tripartition was developed as a framework for phenomena which occur in those cultures in which an Indo-European language is spoken. The terms "Indo-European" and "Proto-Indo-European" refer to a group of languages and are used in this sense in historical linguistics. At the end of the eighteenth century similarities in vocabulary and structure came to light between the classical European languages and Sanskrit. The classic text bearing on the birth of the study of these similarities is in an address delivered to the Asiatic Society in 1786 by William Jones, in which he stated: The Sanskrit language, whatever may be its antiquity, is of a wonderful structure; more perfect than the Greek, more copious than the Latin, and more exquisitely refined than either, yet bearing to both of them a stronger affinity, both in the roots of verbs and in the forms of grammar, than could possibly have been produced by accident; so strong indeed that no philologer could examine them all three, without believing them to have sprung from some common source, which, perhaps no longer exists; there is a similar reason, though not quite so forcible, for supposing that both the Gothic and the Celtic, though blended with a very different idiom, had the same origin with the Sanskrit; and the old Persian might be added to the same family, if this were the place for discussing any questions concerning the antiquities ofPersia. 1
Later research led to an extension of this group of languages all belonging to the same family. It received the name of Indo-European from Th. Young in
1
Jones, 1807, p.24.
CHAPTER ONE
2
18132• The term Indo-Germanic was also used for this group for a time. It covers the following groups of languages: Hittite, Indo-Iranian, Tocharian, Greek, Armenian, Albanian, Italian, Celtic, Germanic, Baltic and Slavonic, as well as a few little-known languages such as Phrygian and Illyrian. The resemblance between these different languages can be explained in two ways: Figure 1: Two models for the resemblances between the Indo-European languages. Proto IE
a
b
d
Proto IE
direction unknown
IE languages
direction known
e
Genetic model
~odelofborrowing
-There was once a group of people who spoke a single language. For some reason they left their original settlement area and after a while they came to inhabit much of the old world. The language which they had once spoken changed and eventually split up into different languages. These languages are the Indo-European languages, which thus all derive from a Proto-IndoEuropean "Ursprache". This genetic model is the commonest. -Languages can borrow from one another to a large extent. Up to a certain point it is possible to determine the direction of the borrowing- who borrows from whom? -but if we continue backwards in time there comes a point when it is no longer possible to determine the direction of the borrowing. In the present case this period is called Proto-Indo-European. The similarities are thus the product of borrowing. Languages can thus be more or less Indo-European, depending on the extent of the borrowing. This model was developed by Trubetzkoy, but it has not received much following.
2 3
Young, 1813.
Trubetzkoy, 1939.
INTRODUCTION
3
Indo-European "Urheimat" and culture The possible existence of the "Ursprache" is not of essential importance for historical linguistics. This discipline is dominated by formal phonetic laws with a prognostic character. The postulated Proto-Indo-European "Ursprache" is thus the product of linguistic research. The genetic model is extremely attractive within this systematic arrangement of structure and sound because it is easy to visualise. However, it does not make much difference to our research if we use the term "Indo-European" in a more artificial way, as Trubetzkoy did. The science of culture does not have formal laws of this kind. It does not produce an Indo-European "Ursprache" as the result of research, but assumes it as its premise, for cultural and religious traditions cannot be formalised in the same way as language because of their more amorphous character. It is impossible to work in this field with a model of borrowing, so we have to assume a priori a genetic model for the Indo-European "Ursprache" and the corresponding "urverwandte" expressions in the field of religion and culture. The genetic model, whose reintroduction is credited to Dumezil by Scott Littleton\ is and has always been a necessary premise for any study in the less clearly defmed field of the science of culture. One of the first questions taken up by historians was that of the Indo-European "Urheimat". Where was the Proto-Indo-European culture situated? Although they are not logically inconceivable, a number of theories, such as C.B.G. Tilak's hypothesis of an "Urheimat" at the North Pole5, have received little or no support. Research in this field has been dominated by two hypotheses. The supporters of the first situated the "Urheimat" in Northern Europe, while those who adhered to the second located it in Southern Russia. It is the second of these alternatives which is most in favour at the moment6• It is by definition impossible to know anything about linguistic changes which may have taken place in prehistory. The most we can hope to fmd is changes in prehistoric artefacts. Pottery, the burial of the dead and the objects that are buried with them never stay the same and are subject to trends. The occurrence of sudden changes may indicate external factors such as the influence of centres of power which have become stronger or the immigration of new population groups. On the other hand, the presence of very gradual change or the absence of any change at all may be taken as indicators of the absence of immigration of any significance or major external factors. Montelius postulated such a continuity for Scandinavia, Denmark and Northern Germany7, basing his conclusion on the typological analysis of pottery. Kossina therefore concludes that if we fmd Indo-European languages here within a historical period, the
4
5 6 7
Littleton, 1974a. Tilak, 1903. For an overview see: Mallory, 1973. Montelius, 1900.
4
CHAPTER ONE
speakers of these languages must always have lived here8• In that case the Indo-European "Urheimat" has been found, and it is to be located in Northern Germany. The Indo-Europeans were the Germanic Aryans, a theory which was to have political implications later on. This location, or a similar one somewhere in Northern Europe, was the status quo at least until World War If9. The Southern Russia hypothesis, on the other hand, is based on a discontinuity in European history and an invasion by Indo-European cultures correlative with this discontinuity. The theory of invasion is no novelty, but it has found a fervent advocate in recent years in the person of Marija Gimbutas10• She has incorporated dramatic changes based on archaeology, particularly the transition from the copper to the bronze age, in her theory. Indo-European groups came to Europe in a number of waves. The first group reached Europe in the first half of the fourth millennium B.C. (a millennium earlier than in the theory of Schrader and Childe). In recent years Marija Gimbutas has built up a position of authority in circles connected with the Journal of Indo-European
Studies.
Both camps see Indo-European society as a society of nomadic conquerors who were extremely mobile thanks to the domestication of the horse and the chariot. It was a dominant elite which was going through a period of expansion. Renfrew's theory should also be mentioned besides these two other models 11 • He does not envisage a rapid migration by nomadic Indo-European groups, but a protracted process or a "wave of advance". The origin of agriculture resulted in an increased density of population. The group of agriculturalists gradually spread out from its original centre. Each generation moved a few kilometres further out. The local hunting cultures were overwhelmed by numbers on all sides. Renfrew identifies these agriculturalists with the speakers of the Indo-European languages. From 6000 B.C. this group spread from Asia Minor in a "wave of advance" to the West. They also expanded to the East in the same way, though it may be that nomadic Indo-European invasions from the steppes of Southern Russia in the second millennium B.C. have to be incorporated in the picture as well. Renfrew refers to this as a peaceful penetration by an egalitarian agricultural culture.
Dumezil's view of the "Urheimat" and Proto-Indo-European culture Dumezil has only incidentally tackled these questions relating to Proto-Indo-European culture and the causes of its expansion. In his opinion, it is a "demi-civilise[e]" culture 12, a term which he uses in opposition to "primitiP'.
8 9 10 11 12
Kossina, 1902. See e.g. Schrader, 1890 and Childe, 1926, 1950. Gimbutas, 1963, 1970, 1973, 1980. Renfrew, 1987. e.g. Dumezil, 1924a, p.235; 1938a, p.198.
INTRODUCTION
5
His teacher Meillet used the same opposition13 • It is possible to reconstruct the culture of these "demi-civilise[s]" by means of the "methode comparative" 14, comparing traditions within the Indo-European cultures. The "methode ethnologique/sociologique" enables us to compare this reconstruction with "demi-civilises" (or "non-civilises"15) societies which have still to be investigated. Thus Dumezil compares his reconstruction of the Indo-European ambrosia festival with the potlatch among the Kwakiutl, Tlinkit and Haida16• Or to take another example, he compares an Indo-European society, whose traces can still be detected in the myths of the Centaurs and the Gandharva, with societies which Frazer described among the Kwakiutl 17• This Proto-IndoEuropean culture is situated "somewhere" in the North. Dumezil speaks of "les peuples indo-europeens restes au nord" 18, referring to Germans and Celts who have remained in their "Urheimat"19• In 1941 he describes them as "une societe de conquerants ... mobiles", associated by ties of ethnic solidarity rather than by patriotic feelings 20• In the same year he presents a more detailed picture. The "Urheimat" is to be situated between the Hungarian plain and the Baltic. In the course of the third and second millennia B.C. waves of "troupes conquerantes" fanned out in all directions, a movement which is still going on today. Aujourd'hui, au de!a de luttes fratricides qui sont peut-ftre le dur enfantement d'un ordre stable, on ne voit sur Ia planete qu'un coin de terre ou pat grandir un appelant contre ce triomphe. Mais sans doute arriverait-il trop tard. 2
He is now also more explicit about the term "primitif': On ne croit plus que les Indo-Europeens aient ete des ; on sait que ni leur civilisation ni leur langue ne permettent d'atteindre un , un zero absolu; que l'une et l'autre portent au contraire Ia marque, Ia charge et le fruit d'un riche passe dans lequel notre vue ne peut guere remonter. 22
He repeats this in 1952 in remarking that Indo-European culture has a long religious development behind it, distinguishing it from forms, "qu'on appelle un peu vite elementaires ou meme primitives"23• A dominant elite expands and this
13 14
15 16
17 18 19
20 21 22 23
Meillet, 1922, p.49. Dumezil, 1924a, pp.287-8. ibidem, p.II. ibidem, p.273. Dumezil, 1929a, p.251, 258. Dumezil, 1924a, p.281. ibidem, p.280. Dumezil, 1941a, p56. Dumezil,194lb, p.387 = 194lc, p.ll-12. Dumezil, 194lb, p.389 = 194lc, p.14. Dumezil, 1952a, p.5.
CHAPTER ONE
6
"Herrenvolk"24 subjects the local population. Sometimes Dumezil assumes that the agriculturalists belong to this elite and join in the movement of expansion, while the local populations are enslaved or exterminated. At other times, however, he assumes that the elite are exclusively warriors and that the local populations are incorporated in the system as a class of farmers (see p.84, 180). This wave reached the Near East early in the second millennium B.C.25 • In 1949 Dumezil's picture is preceded by a remark stressing the vagueness of the model. II serait vain de pretendre dater les evenements ou meme, dans l'etat de nos connaissances, enJ'lacer avec precision, sur Ia carte, le point de depart. Mais le fait est 1~. In 1958 he formulates the same position in highly explicit terms. II ne sera pas une seule fois question, dans les pages qui suivent, de !'habitat des lndo-Europeens, des voies de leurs migrations, non plus que de leur civilisation materielle. Sur ces points fort debattus, Ia methode ici employee n'a pas de prise et, d'autre part, Ia solution n'en importe gu~re aux probl~mes ici ~-La que nous envisageons est celle de l'esprit.2 In 1968, however, Dumezil indicates that the supposition of a "Urkultur" which can be reconstructed until just before the moment of the "dispersion" is an oversimplification. He now goes no further than to state that there was something resembling a "communaute de langage"28, at least possessing a "minimum de civilisation commune" 29• There is another vague indication in 1987, when he describes the Indo-European people as: Un peuple plus ou mains unitaire, sur un domaine assez vaste pour qu'il y ait eu des differences dialectales dans Ia langue que taus utilisaient. Pour une raison inconnue, grace ~ Ia suprematie que constituaient le cheval de guerre et le char ~ deux roues, ils se sont repandus dans toutes les directions par vagues successives, jusqu'~ l'epuisement des reserves. 30
Dumezil, 1944, p.171. Dumezil, 1945, p.16. 26 Dumezil, 1949a, p.15. 27 Dumezil, 1958a, p5. 28 Dumezil, 1968, p.10. 24 25
29 30
ibidem.
Dumezil, 1987, p.llO.
IN1RODUCI10N
7
He also refers at this point to the accepted view: nowadays it is generally agreed that the "Urheimat" should be situated in Southern Russia. He is not attracted by the hypothesis of a site in Asia Minorl1• We can thus state that the question of the "Urheimat" never played an essential part for Dumezil. He goes along with the commonly held view, which does not conflict with his model of tripartition. Renfrew's mode~ however, is incompatible with Dumezil's theory, as Renfrew himself points out: There is thus no doubt that he (Dumezii,W.B.) is thinking in this passage of the first Indo-Europeans to reach Italy, who, on the model proposed here, were some of the first farmers at a date before 5000 BC. Even on the Kurgan invasion theory, as propounded by Maria Gimbutas, we are dealing with a date well in advance of 2000 BC, and consequently with a society, in terms of what we know of early bronze age Italy, reflecting at most a modest chiefdom organization. These comments are not mere pedantry. They bring into question one of the most essential foundations of the whole edifice of Dumezilian scholarship: its historical reality. What we see of analogous social institutions and similarly stratified societies in India, Rome and Gaul in the first century BC, or a little earlier, was simply not common to these societies or to their predecessors in 2000 BC, let alone in 4000
BC. 32
It is easy to understand Dumezil's rejection of the Asia Minor model in this
light.
Triparlition: a general introduction Before analysing the development of the theory of tripartition in detail in chapters 2 to 6, it is useful to present a brief summary ftrst. It is intended as a guide to the analysis rather than as a critical account. After looking at the period before 1930 I shall summarise two articles from 1930 and 1938 in order to outline the main points of Dumezil's theory.
The pre-1930 period Besides the question of where the "Urheimat" was situated, attempts were made at reconstructing more substantial aspects of Indo-European society. This was more the domain of linguists than archaeologists. The proto-lexicon was used as the basis for gaining insight into concepts which were characteristic of Indo-European religion and Indo-European law. Names of gods, religious rituals and dignitaries were compared. Comparative mythologists also tried to group religious phenomena under a common category, which was located in
31 ibidem, p.lll. 32
Renfrew, 1987, p.254.
8
CHAPTER ONE
cosmic phenomena (the sun, the moon) or in atmospheric phenomena (thunder and lightning), following the fashions of the nineteenth century. These endeavours did not provide much in the way of results. In his introduction to the study of Indo-European, Meillet remarks with circumspection: L'observation des ressemblances du sanskrit, du grec, etc. conduit A des conclusions pr&:ises. II n'en va pas de m!me de toutes les corncidences analogues que, A d'autres ~gards, pr~sentent deux populations; par exemple, en d~pit des ressemblances que !'on constate entre les religions des Hindous, des Iraniens, des Grecs, des Germains, etc., on n'a pu constituer un corps de doctrine de religion compar~e de ces divers peuples.33 The sub-title of Dumezil's 1924 dissertation, "Esquisse d'une etude de mythologie comparee indo-europeenne" was thus a polemic against the dominant view of the time. This investigation of a hypothetical ambrosia cycle was based on the assumption that mythological ensembles can be compared with one another as long as they are C9mplex enough and as long as the series of successive components is long and unpredictable. This dissertation is hardly relevant for the later theoretical development except as an exercise in the treatment of the sources. The same is true of a few later publications34• This development begins with a sudden insight that Dumezil received in the spring of 1938. It is in fact the elaboration of a problem in Indology which he had already discussed in 1930.
"La prehistoire indo-iranienne des castes" In 1930 Dumezil discussed the problem of the origin of the Indian caste system (referring to the varnas, not the jatis). Is the strict distinction between brahmans, kshatriyas, vaishyas and shudras, i.e. between priests, warriors, farmers and merchants, on the one hand, and the non-Aryan part of the population, on the other, a system that the Indo-European invaders found and borrowed in India? Or does the caste system date from a later period? Was this social division already characteristic of Indo-Iranian society in a more or less pronounced form? This 1930 article provides an answer to these questions. Dumezil's argument may be summarised as follows. The first direct reference to the caste system in the Vedas is relatively late. The first clear reference is in the Purusha hymn RV X,90. The question which then arises is: was there no caste system in the earlier period, or is the lacuna only a gap in our sources? Another possibility is that we are dealing with a stricter version of oppositions which were already to be found in an earlier
33 34
Meillet, 1922, p.13-14. e.g. Dumezil, 1924b; 1925a; 1925b; 1926.
INTRODUCTION
9
period. An important question in this connection is that of whether such a system existed in an Indo-Iranian period or not. Comparisons with Persian material had already led Fr. SpiegeP5 to conclude that there was a class system in the Indo-Iranian period. Dumezil adds material taken from the Ossetes, a Caucasian people who speak an Iranian language. References by Herodotus to the three sons of Targitaos and more recent folklore material on the three families of the Nartes, the more or less mythical heroes of the earliest times, form the evidence for a class system going back to the Indo-Iranian period, a system which was less rigorous than the Indian caste system.
"La prehistoire des flamines majeurs" It was now possible to ask another question: did this class system go even
further back in time to the Proto-Indo-European period? Dumezil's answer in 1938 formed the basis of what was to become the theory of tripartition. His argument can be summarised as follows. Vendryes had noted a large degree of similarity in religious terminology between the Indo-Iranian and Halo-Celtic groups - both Indo-European groups - despite the large distance between them36• Does their social structure reveal a similar resemblance? The Indo-Iranian group has a clear-cut system, viz. the one discussed in 1930. Ireland and Gaul also display a similar system, though it is somewhat diluted in their case. Gaul has its priests and warriors ("druids" and "equites"), Ireland has its warriors and farmers ("flaith" and "bo-airig"). At first sight Rome does not have anything comparable. However, while in India the brahman and the king (raj) are the summit of the social structure, their etymological equivalents in Rome, the flam en (dialis) and the rex (sacrorum) are the pinnacle of an ancient sacerdotal system. Festus (s.v. ordo sacerdotum) describes a system led by the rex and, below him, the flamines dialis, martialis and quirinalis, with the pontifex maximus following beneath the flamines. In this system the gods Dius (Jupiter), Mars and Quirinus had roles which could be described in functional terms. This triad occurs in contexts where the Roman people as a whole has to be represented. The flamines maiores represent the entire people just as the Indo-Iranian castes formed the whole of society. The Roman gods can be compared with the Indo-Iranian classes in terms of their functions: Jupiter - le dieu de Ia souverainete magico-religieuse - Brahmans Mars - le dieu de Ia guerre - Kshatriyas Quirinus - le dieu des citoyens romains - Vaishyas
35 Spiegel, 1878. 36
Vendryes, 1918.
10
CHAPTER ONE
If we assume a system with class distinction in Proto-Indo-European society, it can be observed in a stricter version in the Indian caste system. The archaic priestly hierarchy in the Roman republic preserves it as a fossilised relic; and both Irish and Celtic society have retained it in a fossilised form, which has further been reduced to two groups. Dumezil outlines the following picture of Proto-Indo-European society. There were three separate classes in this society. Though the distinction between them was not as strict as in the later Indian society, each of these classes had a clear-cut function of its own. The different classes can therefore be labelled in terms of a ftrst, second and third function. The ftrst class was concerned with government and religion, the second with military affairs, and the third with the production of food and with welfare. In Luther's terms, they are the Lehrstand, Wehrstand and Nahrstand. Indo-European society as a whole could be represented by the deities corresponding to these classes or functions. They were also aware of the need to cooperate with one another. Despite the fact that not all classes shared the same status and that there was a clear hierarchical distinction between them, they were still dependent on one another. This was expressed in an ancient Indo-European myth in which the members of the ftrst and second functions came into conflict with the members of the third function. After a war in which neither party scored a decisive victory, the representatives of the three functions were peacefully integrated. Various Indo-European cultures have myths based on this system of three classes, which demonstrate that people were aware of the system.
The system: a synopsis The rather abstract models sketched above will now be filled in with more detail, mainly following Dumezil's own 1958 synopsis37• First we shall consider the "sociale" tripartition. This social division appears to be an expression of a more general principle involving the three functions. This will be illustrated with a number of examples, concluding with the most important form of its expression: the pantheon. The "sociale" tripartition Post-Vedic Indian society was rigorously divided into four castes (varnas). The upper three are pure Aryan castes and are strictly separated from the fourth. These three classes comprise: the brahmans, the priests who study the holy scriptures and perform the sacrifices; the kshatriyas (or rajanyas), the warriors who defend the people; and the vaishyas, the producers of material goods. As
37
Dumezil, 1958a.
INTRODUCTION
11
we have seen, Dumezil considered this system to be ancient and Indo-European from the first. He bases this view on the following correspondences. In the Avesta there is a similar distinction within Iranian society. The series comprises: the priests (athaurvan/athravan), the warriors (rathae.shtar) and the farmers (vastryo.fshuyant). There is an exhaustive description of this society in the division which Jamshid, the legendary king of Persia, made of his kingdom, as recorded by Firdftsi. J amshid divided his kingdom among the asravan, the priests whom he set in the mountains to offer sacrifices; the artestar, who fought like lions and were responsible for the defence of the royal throne, while the vastryos were responsible for providing food. Dumezil considers this model as an instrument for analysing the key structures of the society rather than as a class system which was actually implemented. Another example is furnished by the Scythians, who lived to the north of the Black Sea. According to Herodotus, the first man to come to this previously uninhabited country was called Targitaos. He had three sons, Lipoxais, Arpoxais and Kolaxais. The latter acquires a number of golden objects: a plough, a yoke, an axe and a cup. He owes his rise to power to the possession of these four objects. The four Scythian tribes are the descendants of these three brothers. The objects which are essential for the exercise of power have a significance which can be accommodated within a tripartite division. The plough and the yoke are farming objects, the axe is a typically Scythian weapon and a symbol of the military class, and the cup refers to priestly functions during cultic ceremonial. The legends of the Ossetes contain recent evidence for a similar class system. The Nartes, the legendary heroes of old, are divided into three families. The Alaegatae are powerful through their intelligence, the Aexsaertaegkatae are distinguished by their bravery and courage, and the Boriatae are characterised as rich herdsmen. These three families are thus defmed in terms which are an expression of the three functions. According to written sources dating from soon after the conversion of Ireland to Christianity, Irish society appears to have had two classes, corresponding to the second and third functions: the military aristocracy (flaith) and the cattle owners (bo-airig). Caesar also mentions two classes in Gaul, the druids and the knights (equites). Thus the first function had disappeared in Ireland, while the third had disappeared in Gaul. Their combination, however, results in the ancient Indo-European model. Roman history reveals a division into patricians versus plebs. However, this may well be the result of a later development. Traditionally Rome was composed of three ethnic components: the Latin companions of Romulus and Remus, the Etruscan allies led by Lucumo, and the Sabines led by Titus Tatius, who had been enemies of Rome at first. These three groups are the Ramnes, Luceres and Titienses. The founders of each of these groups have different functions. Romulus is the first-function founder associated with the religious ceremonial, Lucumo is the second-function military specialist, and Titus Tatius is commonly described as the keeper of herds, which fits in well with the third function. The results can be illustrated as follows:
12
CHAPTER ONE
Figure 2: The "sociale" tripartition India
Iran
(Scythians) Firdusi
Avesta
Herodotus
1
Brahmans
Athaurvan
Asravan
Kolaxai:s (cup)
2
Kshatriyas
Rathae.shtar
Artestar
Lipoxai's (battle-axe)
3
Vaishyas
Vastryo.fshuyant
Vastryos
Arpoxals (plough/yoke)
Celts (Ossetes)
Gaul
1
Alaegatae
Druids
2
Aexsaertaegkatae
Equites
3
Boriatae
Romans Ireland Latins
Ramnes
Flaith
Etruscans
Luceres
Bo-airig
Sabines
Titienses
General principle of classification The question of whether Proto-Indo-European society was divided on a tripartite basis in actual fact is not the main problem for comparative studies. That consists of the principle of classification. Whether social tripartition was actually implemented in practice or not, it appears to be one of the forms of expression of an ideology, the "ideologie tripartite", which also occurs in other contexts. In an inscription in Persepolis, Darius (521-486 B.C.) asks Ahura Mazda to protect his kingdom against every imaginable form of evil. The menaces against which he asks for protection are formulated in a tripartite way. He asks for protection against the enemy army (second function), against a bad harvest
IN1RODUCTION
13
(third function) and against the lie (drauga), which fits in with the first function. Similarly, the ancient Vedic sacrificial texts mention a sacrifice to Agni accompanied by a prayer for protection against subjection (second function), incorrect sacrifices (first function) and bad food (third function). In another example, the introduction to the Senchus Mor, the ancient legislation of Ireland, we fmd the statement that the earth was destroyed three times. The first time was due to epidemic and starvation (third function), the second was through a war (second function) and the third was because oral agreements had been broken (first function). Tripartition in the pantheon Dumezil sees this system not only as a principle for the organisation of (reflection on) society and as a framework for the compilation of lists, but especially as an instrument for ordering the Indo-European pantheon. In the Indian texts we regularly come across invocations of Mitra and Varuna, the rulers of the universe, together with Indra, the warrior par excellence, and the Nasatya/Ashvin twins, who are associated with offspring, healing and the acquisition of goods. This triad crops up regularly, and in the same sequence. The oldest text in which it is to be found is the Mitanni treaty. After the various Indo-Iranian groups had split up, one of them left for the West and travelled along the Euphrates to found the Hurrian empire of Mitanni. The ruler of this empire, Mativaza, concluded a treaty with the Hittites in ca. 1380 B.C. In this treaty the gods Mitra, Varuna, Indara and the Nasatyas are invoked, apparently as representatives of the entire state. Dumezil was able to compare the Indian tripartite social system with that of Iran, but the comparison of the two pantheons proved more difficult. The relevant Iranian texts do not go back any further than the period of Zoroastrianism. Very little is known about this reformer, and the period in which he was active cannot be pinned down with an accuracy greater than four centuries. His teachings are clearly presented in the Gathiis, a part of the Avesta, which was written by him or by someone from his immediate surroundings. It was a religion of strict monotheism which did away with polytheism. Ahura Mazda is the only god. The fanatic monotheism of Zoroaster was somewhat diluted in the post-Gathic period. At this time we still come across Ahura Mazda, but there is also the god Mithra, as well as the demons Indra and the Naonhaithyas. The Gathiis do contain a group of beings besides Ahura Mazda. These are the Amesha Spenta, who resemble archangels. They are: Vohu Manah (good thought), Asha (order), Khshathra (force), Armaiti (pious thought), Haurvatat (integrity, health) and Ameretat (non-death, immortality). The significance of these beings had been the subject of controversy for a long time. It was supposed that they derived from older Indo-Iranian gods, but which ones? Dumezil's solution is to consider them as a
14
CHAPTER ONE
Figure 3: Tripartition in the pantheon
India
Iran
Rome
Gathic
post-Gathic Mithra/ AhuraMazda
1
Mitra/ Varuna
VohuManah/ Ash a
2
Indra
Khshathra
3
Ashvins/ Armaiti/ Nasatyas Haurvatat Ameretat
Germans
(gods)
Jupiter Indra
Mars
Naonhaithyas (demons)
Quirinus
Celts Tuatha De Danann
Caesar
1
Odhinn
Nuada/Lug!Dagda
Mercurius/Jupiter
2
ThOrr
ogma
Mars
3
Njordhr/ Freyr
close transformation of the Indian deities discussed above. The twin Nasatyas are here split into two separate entities. Bodily well-being, a concern of the third function, is clearly expressed in the figures of Haurvatat and Ameretat. Although they are not twins, they are rarely mentioned individually. We often fmd a goddess linked with the third function, like Sarasvati's connection with the Nasatyas in India. The earth is a symbol of nourishment and fertility, and is also the element used to characterise Armaiti. This female being fits in well with the third function. The essence of the social representatives of the second function in India, the kshatriya, is the kshatra, literally represented in Iran by Khshathra. The distinction between Mitra and Varuna will turn out to be a very important one. It also corresponds to the distinction between Vohu Manah and Asha. These entities are a "translation" of the gods Mitra, Varuna, Indra and
INTRODUCTION
15
the Nasatyas into terms which were acceptable for a monotheistic religion. The Roman series of Jupiter, Mars and Quirinus has already been discussed in connection with the 1938 article. The description of the ancient temple in Uppsala by Adam of Bremen provides a Germanic variant of this series. This temple contained three divine images, depicting Odhinn, Thorr and Freyr, who are all described by Adam of Bremen in functional terms. The Old Icelandic pantheon presents Odhinn as the head of the council of the gods, Thorr as the leading warrior who combats the giants, and Njordhr and Freyr, who are noted for their wealth. Although our information about the pre-Christian Celts is very limited, there are a number of points worthy of comment in this connection. In the interpretatio romana of the ·gods of Gaul, Caesar lists: Mercurius, the supreme god and the creator of all the arts; Jupiter, the ruler of the celestial beings; and Mars, who is the god of war. Apollo (the god of healing) and Minerva (who initiates into crafts) are also mentioned, but Dumezil considers that they belong outside the tripartite framework. Irish tradition contains a description of the people of the goddess Dana, the Tuatha De Danann. This account probably contains the last vestiges of the ancient Irish pantheon. The main figures in this group are: Nuada, the king, Lug Samildanach, the master of the arts, Dagda, the great druid, Ogma, the first champion, Dian Cecht, the healer, and Goibniu, the smith. This is a reduced pantheon, in which only the first and second functions are covered. This material can be arranged as in figure 3.
The discussion It is not striking that three kinds of reaction to Dumezil's publications can be noted. He was simply ignored, criticised or acclaimed. There is a remarkable degree of intensity in the reaction of both his supporters and his opponents. I shall only briefly touch upon these three reactions, in order to see whether Dumezil's work has already received critical analysis, and if so, what the results of this are. I do not aim to provide a detailed history of the reception of his work.
Indifference In a number of cases Dumezil was simply treated with indifference by other scholars in the field. Dumezil interpreted the complete absence of any serious reaction as an implicit rejection, as can be seen from his reaction to the discussion centred on the interpretation of a text by Propertius, for instance:
16
CHAPTER ONE
On a parfois ignor~ mes raisons et mes r~ponses, on ne les a pas discut~. Les seules contradictions in~dites que j'ai enre~tr~es ne soot pas du type dialectique, mais rMtorique et n'appellent pas de d~bat. It is difficult to convince those who accept Dumezil's interpretation of some of
the evidence but refuse to accept that it leads to a tripartite system: "comment contraindre a voir quelqu'un qui ne voit pas et qui, souvent, ne tient pas a voir?"39• Dumezil is particularly emotional in a reaction to K Helm's criticism40, where he expresses the feeling of not being accepted, and even of being ignored. "Je ne suis plus jeune moi-meme, mais les circonstances ont fait que, aux yeux de beaucoup, je reste un debutant, ignore, parfois < ignoriert > " 41 • Acclaim
We can divide the supporters of the theory into three functions: The pioneers A few scholars have had an important influence on the further theoretical development and elaboration of the tripartite model. A particularly important figure in this connection is E. Benveniste, who stressed the Indo-Iranian character of the social tripartition and indicated other tripartite categories in Indo-European texts42• S. Wikander was also important because of his publication on the tripartite structure of the Mahabbarata43• Another influential figure is L. Gershel, who has worked on the elaboration of the theory of a Roman "historicisation", the idea that (vestiges of) ancient Indo-European mythology can be found in Roman historiographyM. Academic respectability A few famous scholars have adopted a positive attitude to Dumezil's ideas and have adopted them in their own work: the Germanic experts De Vries and Turville-Petre, the Celtic authority Sjoestedt and the Iranian scholars Lommel, Widengren and Duchesne-Guillemin. Their support enhanced the status of the model, but it is not relevant to the present inquiry. It should be noted, however,
38 39
Dumezil, 1948a, p.137-138. Dumezil, 1949a, p.245. 40 Helm, 1955. 41 Dumezil, 1956a, p.173. 42 Benveniste, 1932, 1938 and 1945. 43 Wikander, 1947. 44 Gerschel, 1952; 1953.
INTRODUCTION
17
that Dumezil was and is supported by authoritative scholars without the status of the model being subjected to scrutiny. Elaboration One group of scholars took Dumezil's work as the starting point for their own researches. This group is concerned with the detection and elaboration of tripartite models in Indo-European cultures and cultures which display similar systems under Indo-European influence45 • The publications by Balog, Dubuisson, Obayashi, Sergent, Strutynski, Yoshida, Evans, Grisward and Briquet should also be mentioned in this connection. The most recent edition of Littleton's book contains a review of the activities of these scholars46• I shall not be concerned with these researchers inspired by Dumezil in the present work. Although their publications might be seen as evidence for the wide range of applicability of the model, this is not in fact the case, and they are thus not relevant to the present investigation. The most favourable response that the work of these scholars commands is respect for the creativity with which they have embraced the tripartite model and applied it in new ways. Their acceptance of the theory, moreover, took place when the theory was already in a late stage of development.
Criticism The criticisms which have been made can be divided into four kinds: criticism and rejection of the (secondary) interpretations of the Indo-European material; criticism of the conclusion, that although it is derived from a correct interpretation of the material, it must go back to a common Indo-European "Urkultur"; criticism by providing better models which are an improvement on Dumezil's model because they are operational on a wider scale; criticism of the implicit or explicit political implications, background or colouring of Dumezil's theoretical principles.
Criticism and rejection of (secondary) interpretations Various specialists have pointed out the inaccuracy of many of Dumezil's interpretations. Most of these criticisms were levelled by classicists. Dumezil's theory, which is based on personal deities which go back to the Indo-European period, proved to be extremely incompatible with the authoritative theories of 45 The work of this group can be found in the following publications: Hommages a Georges Dumezil, 1960; Puhvel, 1970; Cardona, 1970; Benoist, 1973; Larson, 1974; Riviere, 1979; Bonnet, 1981. 46 Littleton, 1982, pp.153-185, 202, 204-215.
18
CHAPTER ONE
those labelled as "primitivists" by Dumezil, such as Rose and Wagenvoort. They considered that early Roman history was a period of the worship of an impersonal mana (numen), and that personal deities were the products of a later development47• These "primitivists" were strongly influenced by Marett's ideas. The discussion was always conducted in emotional terms, but it was sometimes lacking in substance. Dumezil came up with an explanation for the resistance that he encountered from the quarter of specialists in Roman culture in 1947. He attributed it to: ...les belles fosses de leurs chantiers ne contenaient plus tout ce qu'il faut savoir, ni sans doute l'essentiel de ce qu'il faut savoir pour interpr6ter les premiers livres de Tite-Live et de Denys d'Halicarnasse ou le second livre duDe Republica. 48
Iranian scholars heavily criticised a marginal issue. Thieme and Dumezil engaged in a highly charged discussion of the interpretation of "ari-, arya-, Aryaman" 49• Gerschevitsch joined in later on and involved the tripartite model in his criticisms50• The interpretation of the families of the Nartes was criticised by Smith and Sperbe~ 1 • Criticism in this category is aimed not so much at the model itself as at some of its applications. The question of its applicability in certain specific situations does not necessarily entail consequences for the model itself. Gonda's critique seems to be more fundamental 52• He notes that the conclusions which Dumezil derives from his material are sometimes based on too little evidence, that Dumezil uses disputable etymologies, and that "he is sometimes carried away by his own hypotheses, constructions and formulations" 53• Gonda concludes with the remark that "Dumezil's theoretical framework is, to quote Brough once again, extremely elastic; the more capable it is of stretching to accommodate the facts, the less useful it becomes in explaining them so as to convince critical minds"54• However, the arguments used by Gonda are so general that it is a case of shooting from the hip rather than providing a detailed critique, although it must be admitted that he does score some hard-hitting points. More damaging is the fact that this general critique is illustrated by a very dated example, viz. the alleged correspondence between Ouranos and Varuna which Dumezil had investigated in 1934 and which has no relevance for the tripartite model. Another weakness in Gonda's
47 48 49 50
51 52 53 54
e.g. Wagenvoort, 1947. Dumezil, 1947a, p.29. Thieme, 1957. Gershevitsch, 1959. Smith and Sperber, 1971. Gonda, 1960. ibidem, p.4. ibidem, p.15.
IN1RODUCTION
19
critique is that he does not appear to be aware of developments within the tripartite model55 • Criticism of the conclusion Another sort of criticism was the charge that tripartite systems such as those postulated for the Indo-European cultures actually consist of very general, natural categories. Brough attempted to detect a tripartite system in non-Indo-European cultures in order to demonstrate the point. He did so by means of an investigation of tripartition in the Bible56• It has been more common for critics to refer to this investigation than to Dumezil's reaction, in which he shows that Brough has been compelled to interpret his material in a very forced way in order to arrive at a tripartite system57• Criticism by providing better models Frye has pointed out that those who would criticise Dumezil's model are obliged to come up with something better. The question is not whether the oldest sources are used (i.e. is the right material used?), but whether it works. The question should then be: does a different model account for a wider range of phenomena58? I do not agree with Frye on this. In contemporary scientific practice, theories are generally only rejected when more operational theories are available, but there is no need to turn this description into a prescription. Besides, Dumezil is not just responsible for the explanation of his material, but for the way in which the material has been collected as well. Kuiper took up this challenge to fmd a different structure in the Mahabharata. In Kuiper's opinion, a structure based on a centre and the four cardinal points is more adequate than Dumezil's tripartite model59• Criticism of the explicit or implicit political implications Another dimension has been added to the discussion in recent years. Momigliano, followed by Ginzburg, have pointed out the pro-Nazi character of Dumezil's 1939 publication on the Germanic gods60• This discussion also mentions the interest which "la nouvelle droite" has taken in the work of Dumezil61 • This discussion is certainly an interesting one, but it tells us more
55 56
See Dumezil's response in: 1977, pp.252ff. Brough, 1959. 57 Dumezil, 1959a. 58 Frye, 1960. 59 Kuiper, 1961. 60 Dumezil, 1939a. 61 Momigliano, 1983; Ginzburg, 1985.
20
CHAPTER ONE
about the academic preferences of the extreme Right than about the development of a theoretical model. Although the detection of order in Indo-European cultures may go hand in hand with a preference for order as well, it is difficult to verify arguments in a discussion of the role which such motives might have in the foundations of scientific models. I shall therefore not enter into this discussion. Conclusion
Dumezil's model is very challenging. If he is right, it provides an important insight into the way in which not only religion but also other aspects of the Indo-European cultures are structured. Furthermore, Dumezil's theory is one which not only accounts for correspondences between large volumes of material - and it is not confined to specialised secondary fields of research - but it can also be subjected to empirical scrutiny. Moreover, it means that scholars working in the various specialised sub-disciplines are bound to take an Indo-European component seriously. It is unwise to investigate the ancient Germanic, Celtic, Roman, Indian and Iranian traditions as if they were isolated entities. His theory also puts an end to the very important question of the reliability of various sources, particularly Snorri, Caesar and the Mahabharata. If Dumezil is right, these sources are very useful for reconstructing the culture of certain periods (pre-Christian Northern Europe) and strata of the population (the non-Vedic situation in ancient India) for which we do not have direct source material. The model is also highly incompatible with other models in a number of cases. If Dumezil's model is correct, the others are not. The discussion of this model has reached an impasse. On the one hand, there is a group of scholars who accept Dumezil's theory unquestioningly, while on the other hand there is a group which voices criticism of (parts of) the application and/or criticism based on very general statements and/or criticism based on arguments beyond the scope of science. This is a very regrettable situation, which fails to do justice to the importance of the model.
CHAPTER'IWO
METHODOLOGY Chapters 2 to 6 contain a description of the development of Dumezil's theory in a number of self-contained sections. These sections fall into two main groups. The first concerns his methodology and its implications. The second concerns his interpretation of the material. Dumezil explicitly and implicitly exploits a number of methodological concepts, a number of auxiliary hypotheses and a paradigmatic framework. This chapter deals with explicit references to methodology.
The methodological concepts The methodological concepts can be arranged according to four principles: the comparisons made are not typological but genetic; emphasis must be laid on the study of "ensembles" and not on the study of details; this approach makes it possible to establish the "centre propre" of the gods in particular; the object of the inquiry can be described as the "conception", "ideologie" which fmds expression in "fonctions".
Typological versus genetic Dumezil's method is a typically comparative one. On the basis of the postulated minimum of a common civilisation, he compares certain material within cultures in which a Indo-European language is spoken. As he stresses again and again, comparison may be done in two ways. The genetic approach, "Ia mythologie comparee, (au sens etroit du mot)"\ permits reconstruction. As we saw earlier (p. 4f.), this genetic reconstruction resulted in a picture of ProtoIndo-European culture which could be compared with other "demi-civilisees" cultures by means of the "methode ethnographique"2• He therefore calls this method the "raisonnement comparatif (au sens vaste "frazerienne")"3• At various points he argues for the use of "comparaison genetique" instead of "comparaison generate, typologique" 4• In his later work Dumezil constantly refers back to the following text on this question:
1
Dumezil, 1924a, p.287. ibidem. 3 Dumezil, 194ld, p.209. 4 e.g. Dumezil, 194ld, p.212; 1947a, p.13, 28; 1949b, p.9; 1951a, p.226; 1954a, p.7-8; 1979a, p.90-91. 2
22
CHAPTER TWO
Sans le moins du monde opposer les deux m~thodes, qui sont ~galement saines, ~galement l~gitimes et d'ailleurs compl~mentaires, nous ne faisons pas de Ia comparaison typologique mais de Ia comparaison genetique; avec toutes les adaptions que commande Ia diff~rence des mati~res, nous essayons d'obtenir sur le domaine indo-euro¢en, pour les faits religieux, ce que d'autres comparatistes ont obtenu pour les faits linguistiques : une image aussi pr&:ise que possible d'un syst~me pr~historique particulier, dont un certain nombre de syst~mes historiquement attest~ sont, pour une bonne part, Ia survivance.5 Dumezil quotes this passage in 194~ and he refers to it again in 19817•
"Ensembles" Ever since his earliest publication Dumezil has emphasised the need to relate details to the whole rather than studying them in isolation. This principle applies to three areas: myth as an entity: it must not be fragmented into isolated themes; the world of the gods as an entity: the gods must not be viewed separately, but in their relation to one another; religion as an entity: the various components of religion, and the relation between these components and other cultural aspects, must not be seen in isolation.
Myth as an entity In the introduction to his dissertation, Dumezil cites the following passage from VanGennep:
< Ce n'est pas Ia comparaison des th~mes consid~r~s isol~ment qui fournira Ia clef de tous les probl~mes que soul~ve l'6tude des litt6ratures dites populaires, mais seulement celle de certaines de leurs combinaisons bien d~finies..... >.8 He formulates his method in line with this quotation: Nous parlons de th~mes, de s6quences, de cycles, et non pas de dieux indo-euro¢ens....9 Les dieux passent, mais les th~mes demeurent. 10
5 6 7
8 9
Dumezil, 1943a, p.26. Dumezil, 1949a, p.42. Dumezil, 1981, p.20. Dumezil, 1924a, p.V.
ibidem, p.VI.
10 ibidem,
p.VII.
METIIODOLOGY
23
Looking back on his work in 1986, Dumezil emphasises precisely those passages in which his "souci de I' ensemble" was already present11 • He now speaks of the "structure" which emerged from this work12• He repeats this argument in his 1925 discussion of Slavic material, where he states: "Aucune necessite n'impose I' union, Ia succession de ces episodes en un meme cycle" 13 • He concludes from this that this sequence of episodes in the material under review must therefore have an Indo-European origin. The emphasis which Dumezil places on the study of myth as an entity represents a continuation of this line of thought, as the following citations show: .. .le mythe dans son ensemble et non dans un de ses details; ......14 Les mythes ne sont pas des puzzles, mais des organismes, qu'on ne peut decouper ~ tort et ~ travers. 15 He stresses a similar difference between the study of details and entities in a different context in 1940, where he notes that the ".. generate des deux legendes est evidemment plus importante que Ia poussiere de details, forcement variable, oil elle s'exprime" 16• When he refers to a "schema indo-europeen" later in the same work17, we may suppose that he implies an "ensemble" of the same kind. He works this idea out in more detail in 1941. More developed cultures have a mythological corpus, a "mythologie", with clearly defmed contours, but this does not mean that less developed cultures do not have such contours. Here too there is a coherence between different myths a ...sentiment de , suffit ~ constituer, au-dessus des mythes, une mythologie, un organisme dont on ne doit isoler les fragments qu'avec precaution. 18 Dumezil makes a comparison between the sequence and structure of the war between the Aesir and the Vanir and that between the Romans and the Sabines (see below, p. 177ff.). He concludes his analysis with methodological considerations: Puisque nous nous risquons a parler de methode, le lecteur voudra bien mesurer l'avantage que nous gagnons par le present essai. Ce n'est plus seulement un
11 12 13 14
15 16 17 18
Dumezil, 1987, p.122. ibidem, p.123. Dumezil, 1925a, p.217. Dumezil, 1937, p.156. ibidem, p.157. Dumezil, 1940, p.32. ibidem, p.93. Dumezil, 1941b, p.393. cf. Dumezil, 1936, p.236, where the same "sentiment" is described.
24
CHAPTER TWO fragment .... , ce ne soot pas non plus quelques fragments extraits des l~gendes...... . que nous consid~rons et comparons, dont nous comprenons le sens et constatons l'antiquit~. Ce soot ces deux l~gendes dans leur ensemble, dans leur organisation
systbnatique. 19
Dumezil considers the "science des contes" as a clear example of getting lost in details. This science breaks each myth down into tiny particles and then looks for similar particles in other cultures. He wittily pokes fun at this approach: II est amusant de transposer cette m~thode en termes linguistiques: elle ram~nerait toute l'~tude ~ un commentaire phon~tique. Devant l'accusatif pluriellatin deos, on dirait: Cette recherche peut avoir un petit interet: etendue de proche en proche, elle rev~lerait les sequences de sons, rares ou frequentes, admises par le latin. Pourtant sur deos, il y a des remarques plus importantes ~ faire. 20
In 1956 he tackles the objection that it is incorrect to reduce the events of a myth to their schematic character and to see this as their essence. But we must "en comprendre Ia le~on, laquelle coincide avec Ia marche de !'intrigue: avec le "21 • The world of the gods as an entity Dumezil elaborates this emphasis on the whole most fully in his discussion of the pantheon. The gods must not be considered in isolation, but in relation to one another. These relations and the way in which they are grouped are extremely important. Dumezil uses various terms for these entities and relations, such as: "symetrie structurale"22; "systeme"23 ; "systematique(ment)"24; "structure"25 ; "structurellement"26; "caractere structural"27; "these structuraliste"28; "groupement"29; "definition differentielle, classificatoire des
Dumezil, 1947a, p.289. Dumezil, 1948c, p.llO. Dumezil, 1956b, p.60. e.g. Dumezil, 1947a, p.44. e.g. Dumezil, 1946, p.62; 1939a,p.4; 1947a, p.46, p.54; 1949a, p.35; 1950a, p.19; 1951a, p.223; 1956b, p.19; 1979a, p.lO. 24 e.g. Dumezil, 1947b, p.124; 1951b, p.112. 25 e.g. Dumezil, 1947c, p.1350; 1947a, p.54; 1949a, p.36; 1950b, p.453; 1950a, p.19; 1951b, p.112; 1952a, p.14, 25, 81; 1954b, p.105; 1956c, p.S, 6; 1957a, p.13; 1958a, p.39, 91; 1959b, p.26; 1966, p.156; 1977, p.3940; 1979a, p.lO, 90; 1983, p.168; 1987, p.l17 w. 26 e.g. Dumezil, 1947a, p.77. 27 e.g. Dumezil, 1959c, p.408. 28 e.g. Dumezil, 1959b, p.17w. 29 e.g. Dumezil, 1947a, p.94; 1977, p.207. 19 20 21 22 23
METIIODOLOGY
25
dieux''30; "rapports entre les dieux''31• These terms are mutually interchangeable and all mean the same thing: a structured arrangement. In fact, what it amounts to is that the gods must not be considered in isolation from their tripartite connection. This corresponds to Dumezil's explicit use of these terms. Hence "dans un certain ordre" is used synonymously with the concept of "structurellement"32• This is connected with another methodological principle. The terms within a tripartite formula must be homogeneous: three characteristics of one deity or three qualities are permissible, but not combinations of the two33 • The term "structure" and its various derivatives have led to some degree of confusion. This is a general problem within the cultural sciences. Van Baal and Van Beek note: The use of the term structure is subject to much controversy in the social sciences..... In scientific usage the battle runs between two mutually exclusive views, the one claiming that the structure of an institution is part of its objective reality, open to observation, the other that it is a hidden power, a latent rational system present in the observed institution.34 The first of these usages goes back to Radcliffe-Brown, the second to Uvi-Strauss, but Dumezil's usage is of a different kind. Bolle's explanation of the term structuralism is not the same as Dumezil's either: The word structuralism has been used in so many ways as to make us nearly forget that in principle its meaning is clear. It means that below the surface of things we can find certain patterns or consistencies; or, more accurately, structuralism is the assumption that we can do so, in mythology and a great many other fields of inquiry. 35 The term has a highly specific meaning within the structuralist school of Uvi-Strauss. The structure of a myth is derived from the basic structure of the human mind and the ordering principle which gives the myth its meaning. For instance, in the Oedipus cycle the structure is not the narrative sequence (as in the previous section), nor the relation between the different actors (as in this section). The meaning which Uvi-Strauss assigns to "structure" refers to the ordering principle, viz. the opposition between an autochthonous origin or a biological one, with as a variant descent from one or two parents. This structure
30 31 32 33 34 35
e.g. Dumezil, 1944, p.25; 1949a, p.65. e.g. Dumezil, 1947a, p.ll; 1957a, p.13; 1958a, p.34, 46. Dumezil, 1947a, p.n. Dumezil, 1%1, p.280. Van Baal and VanBeek, 1985, p.214. Bolle, 1984, I, p.324.
26
CHAPTER TWO
is to be found in all the versions of the myth36• Dumezil and Levi-Strauss thus use the term "structure" in totally different ways. Levi-Strauss' claim37 that Dumezil studies history like a structuralist is thus incorrect. In 1973 Dumezil decided to abandon the term "structure" in order to put an end to this confusion. He refers to himself as a historian, not a philosopher who works with a priori schemes38 - apparently a jibe at the structuralists. The alternative terms which he uses in this work all conform to the meaning of an ordered arrangement described above. They are: "configuration"39, "cadre"40, "tableau"41 • However, he fails to carry out his intention of avoiding the term "structure" in the second edition of Mythe et Epopee I, published in 1974, and in the later works42 • Religion as an entity The sequence of the myth, the scheme and the pantheon are not the only important elements. In turn, they all form part of a wider system, religion. Mais il ne faut pas oublier qu'une religion...... est un systbne, un equilibre. Elle n'est pas faite de pieces et de morceaux assembl6s au hasard, avec des lacunes, des redondances et des disproportions scandaleuses. Si nous osions risquer apres tant d'autres une d6finition, toujours ext6rieure, nous dirions qu'une religion est une explication g6n6rale et coh6rente de l'univers soutenant et animant Ia vie de Ia soci6t6 et des individus. Si done on ne veut pas se m6prendre grossierement sur Ia forme, l'ampleur et Ia fonction propre de tel ou tel d'entre les rouages d'une religion, il est urgent dele situer avec pr6cision par rapport I) l'ensemble. 43 And in the preface of Eliade's Traite: Et ce qui paralt aujourd'hui frappant, ce qui appelle 1'6tude, ce n'est plus cette force diffuse et confuse .... ; ce sont au contraire les structures, les mecanismes, les equilibres constitutifs de toute religion et d6finis , discursivement ou symboliquement, dans toute th6ologie, dans toute mythologie, dans toute liturgie. On est venu - ou revenu -I) l'id6e qu'une religion est un systeme, diff6rent de Ia poussiere de ses 616ments; qu'elle est une pen~e articul6e, une explication du monde.44
36
Uvi-Strauss, 1963, p.206ff.
37 ibidem, p.290. 38
Dumezil, 1973a, p.14-15.
39 ibidem, p.237, 238, 293. 40 ibidem, p.231, 235, 248. 41
ibidem, p.248.
42
See also: Dumezil, 1979b, p.265; 1987, p.l19.
43
Dumezil, 194lb, p.397.
44 Dumezil, 1949b, p5.
METIIODOLOGY
27
However, the use of terms like "systeme" for a complex of these dimensions only plays a marginal role in Dumezil's work.
"Centre propre" The emphasis on the whole as Dumezil practises it is primarily a weapon with which to combat rival theories. In the case of Freyr, the generalistic approach enables us to distinguish between the "activites marginales, (et) le noyau de sa personnalite"45 • If we forget that religion forms a system, ...n'importe que) dieu etant plus ou moins amene a s'occuper de toutes Jes provinces de Ia vie humaine, on risque d'attribuer essentiellement a celui, quel qu'il soit, qu'on etudiera ce qui ne lui appartient qu'accidentellement; on le centrera sur Ia marge de son domaine ou meme au deJa et I'on meconnaitra au contraire sa destination fondamentale. 46
This entails that, although it may be possible to identify economic and agricultural aspects of Varuna, Odhinn and Mars, we must still ask ourselves whether they belong to their "centre propre"47• In particular, it is now possible to respond to the question of whether Mars was originally an agricultural god (Mars agraire). It is true that Mars is invoked by agriculturalists, but this does not mean that agriculture is his "centre propre". His agricultural help is only a subsidiary aspect of his "centre propre". His help consists of driving away demons which threaten the crops, but it is not permissible "a placer le centre de gravite du personnage hors de son domaine habitue} qui est Ia violence, Ia guerre, la victoire"48• Another statement of this point in relation to the discussion of 'Mars agraire' is as follows: a distinction must be made between the "mode de son action" and the "points d'application de son action" 49• This distinction is an extension of the one just mentioned. The way in which Mars acts is determined by his "centre propre"; agricultural assistance is merely an indirect application of this action. The emphasis on structure may help to throw new light on thorny problems. To take two examples from the Germanic field: is the combat between the Aesir and the Vanir (see p. 177) a reflection of actual historical events? Is Odhinn a latecomer who has displaced Tyr (see p.141)? These problems arise because the 'historians' lose sight of the whole and can no longer perceive the structure. The Aesir and the Vanir do not correspond to autochthonous and allochthonous groups respectively. Dumezil opposes the "these structuraliste"
45
Dumezil, 1939a, p.132. Dumezil, 1941b, p.397-398. ibidem, p.398, 399. 48 Dumezil, 194lc, p.84. 49 Dumezil, 1949a, p.Bl.
46 47
CHAP1ERTWO
28
to the invasion theorf 0• As for the relation between Odhinn and Tyr, it is determined structurally, not chronologicallf 1• The problem of the origin of the late arrival Odhinn thereby disappears, claim Dumezil and the structuralists52•
"ldeologie, conception, fonction" The object of study is indicated by such terms as "ideologie" or "conception", which can be formulated in terms of "fonctions". Especially the terms "ideologie" and "fonction" have misleading associations. They must not be confused with the meaning assigned to them within functionalist theory. The term "fonction" is used by Durkheim to mean: unconscious, unintentional side effects of certain acts and the like, leading to a maintenance of the social status quo. For instance, punishment can have the function of group integration and a reinforcement of feelings vis-a-vis the "conscience collective". The functionalism inaugurated by Durkheim has been developed particularly by Malinowski and Radcliffe-Brown. In Dumezil's work, however, "fonction" is synonymous with "activite". In 1938 he presents the "trois modes d'activites" as a synonym for the "trois fonctions sociales"53 • This is a synonym for "les principales activites sociales"54, "trois niveaux de l'activite des hommes arya"55, "activites"56, ""service" des classes"57 and "trois composantes actives" 58• The use of the term "fonction" in the sense of "activite" can also be found outside the tripartite framework. Thus he can assign to Aryaman the "fonction de mariage, de pourvoir aux mariages"59, and on (H)elernus he states that the "fonction de ce dieu [est]: il protege Ia feve de trois mois et generalement les plantes potageres de printemps"60• In all these instances "fonction" thus has the same meaning and connotations as "activite". A similar problem is raised by his use of the term "ideologie". Oosten comments on this term: "The word ideology usually refers to the way people conceive of an ideal organisation of their society"61 • But the term "ideologie" has a different connotation too. In Marxist thought ideology is an instrument of the ruling class to maintain its position. Giddens uses the term in his account of Marx thus:
50 Dumezil, 1959b, p.15ff.
51 52 53 54 55 56 57 58 59 60 61
Dumezil, 1946, p.61. Dumezil, 1952a, p.83. Dumezil, 1938a, p.197. Dumezil, 1941c, p.52. Dumezil, 1947b, p.123. Dumezil, 1948a, p.115, 134. Dumezil, 1948b, p.451. Dumezil, 1958b, p.429. Dumezil, 1949c, p.74, 76. Dumezil, 1975, p.225. Oosten, 1985, p.243.
METIIODOLOGY
29
The modern legal system and judiciary is a principal ideological support of the bourgeois state. But it is only the contemporary expression of the fact that, in all class societies, the dominant class develops or takes over ideological forms which legitimise its domination. 62 Dumezil uses the concept of ideology less as a technical term. Sometimes it seems to mean a philosophical system. When he refers to the "antique theorie de Ia tripartition des fonctions sociales et cosmiques"63, the terms appear to be interchangeable with "ideologie tripartie". Thus in 1948 he can state: Beaucoup de peuples ont une philosophie mythique fort avanree qui n'ont pas encore ou n'auront jamais de philosophie doctrinale. La mythologie prtcMe, prtpare souvent, en tous cas remplace l'ideologie et rend les mtmes services.64 In the triad of "Ia theologie, Ia mythologie et Ia philosophie de Ia tripartition cosmique et sociale"65 , the third term seems to be a synonym for "ideologie". Hence Dumezil can claim that the Romans dated the origin of their structure, their customs and even their ideology soon after the foundation of the ci~. Similarly, in 1952 he refers to the ancient ideological and mythological frameworks67 • Dumezil also wonders whether society was "effectivement" organised in a tripartite way, or whether this is merely an ideology68.This ideology was based on the "conception tripartite". In 1948 he states: "!'existence, a Ia base meme de l'ideologie de Ia plupart des peuples indo-europeens, d'une conception tripartite du monde et de Ia societe, ..."69• And in 1954: "La structure des trois fonctions cosmiques et sociales ... qui formait le cadre de l'ideologie et de Ia theologie anciennes ..." 70• The terms are used in the same way on two more occasions in the same year71 • In the latter passage the structure forms the framework for the "philosophie", which is here evidently used as a synonym for "ideologie". Compare also: "et une philosophie ou, si l'on prefere, une ideologie"n. There is also a weaker usage of the term "ideologie", as in "ideologie de rechange" 73 and "ideologie du vide" 7\ where it means no more than "ideas about".
62 63 64 65 66
Giddens, 1971, p.40-41. Dumezil, 1940, p.146. Dumezil, 194&, p.285-286. Dumezil, 1945, p.182. Dumezil, 1949a, p.l45. 67 Dumezil, 1952b, p.43. 68 Dumezil, 1947c, p.1354; cf. Dumezil, 1966, p.168, where he says of the tripartite framework: "qui n'est plus reste qu'ideologique". 69 Dumezil, 1948d, p.15. 70 Dumezil, 1954c, p.129. 71 Dumezil, 1954b, p.105 and 1954d, p.S. n Dumezil, 1979, p.SO and p.91. 13 Dumezil, 1958b, p.439, 441. 74 Dumezil, 1958c, p.120.
30
CHAPTER TWO
There is another usage of the term "ideologie" to mean a comprehensive system on which various forms of cultural expression are based. Thus in 1948 he remarks: " ..l'ideologie des trois fonctions ...... - gouvemant la theologie, les mythes, les formes religieuses, juridiques et meme litteraires,-...." 75 • In this case, ideology thus precedes the rest and constitutes the framework in which mythology and the other elements are to be found. Dumezil puts it in a similar way in 1950: "comme il est ordinaire, theologie et mythologie montrent en acte ce que l'ideologie contient en puissance"76• Here too theology and mythology are forms of expression of ideology77• Theology, mythology, sacred literature and a priestly organisation are all elements subject to "quelque chose de plus profond, qui les oriente, les groupe, en fait l'unite, et que je propose d'appeler, malgre d'autres usages du mot, l'ideologie ...."18• This ideology is thus a "conception et appreciation" of the overarching forces which activate the world and society and of their mutual relationship79 • Dumezil refers again in 1958 to the "idees dont vit un societe" on which mythology, theology and historiography are based80• Behind this diversity of rituals, myths, institutions and "history'' we can detect a "nombre limite d'idees directrices", or "le fond common que j'appelle l'ideologie"81 • On the one hand, "ideologie" is thus an expression of the "conception" on a par with theology and mythology. Ideology means philosophy, without this term having to refer to the ideal society. On the other hand, "ideologie" is the foundation from which theology and mythology are derived. In this case, "ideologie" and "conception" are interchangeable. This usage leads Littleton to treat "ideology'' and "world view" as synonyms82• The term "Gedankenwelt", which was to be used in the title of the German translation of L 'ideo/ogie tripartie des Indo-Europeens (1958) seems a better equivalent83• Since this term does not have Marxist connotations, its use renders the question of who might benefit from maintaining the ideology superfluous. The term "Gedankenwelt" can thus refer both to the more systematic representations within philosophical systems and to less systematic or unsystematic representations which are expressed in theology, mythology and "history''.
75 76 77 78 79 80 81 82 83
Dumezil, 1948a, p.155. Dumezil, 1950b, p.453. Compare also: Dumezil, 1958b, p.431. Dumezil, 1954a, p.7. ibidem. Dumezil, 1958a, p.74. Dumezil, 1981, p.24. Littleton, 1982, p.227. As announced in Dumezil, 1957a, p.8, n.1.
METHODOLOGY
31
Summary: The methodological concepts Dumhil makes comparisons in order to arrive at a genetic reconstruction of certain aspects of Indo-European culture. He therefore looks for con-espondences which point toward a common origin. Research of this kind, according to Dumezil, should not focus on series of isolated details. It is the entity as a whole which is central. This applies to the themes encountered in the myths, to the gods who fonn a part of the pantheon and are involved in relationships with one another and to religion as a system. The tenninology used for the relations in the pantheon (structure, etc.) led to misunderstandings. These tenns mean no more than "dans un certain ordre". Dumezil also considers it important to distinguish between the "centre propre" of gods (their original and essential function) and indirect applications of their activities. The aim of his inquiry is "ideologie", which can be defined in tenns of "fonctions". The tenns "ideologie" and "fonction", however, also led to misunderstandings. Dumezil uses the fonner in a variety of different ways. In his work, "ideologie" is often equivalent to "ideas about", and ''fonction" equivalent to "activity". The auxiliary hypotheses In linguistics it is necessary to posit the existence of phonetic laws of transformation between Proto-Indo-European and the various Indo-European languages which explain and/or systematise the difference between the original language and its derivatives. In a similar way, Dumezil recognises the need to provide an account of the different forms in which the tripartite conception is given form in the different cultures. Instead of phonological laws there are processes of transformation, or shifts ("glissements"), which give the tradition its typical form. Related to these are the "champs ideologiques" which we fmd in a number of cases. The latter describe the factors which determine the transformations "au sens mathematique et levi-straussien du mot" 84•
Auxiliary hypotheses preceding the origin of the tripartite model Before he had developed his tripartite model, Dumezil already resorted to certain processes of transformation which were necessary for the interpretation of mythical material. Some of these shifts will be dealt with in the following section. In India one has to take into the account the fact that the oldest material contains the least information. The Vedic texts only refer to material that is assumed to be familiar.
84
Dumezil, 1967, p.42.
32
CHAPTER TWO
On ne s'~tonnera pas qu'il faille parfois recourir ilia litt~rature ~pique post~rieure qui, sous le costume d'une religion franchement nouvelle, ale m~rite de raconter, en d~tail et de fa~n suivie, les l~gendes, les cycles qu'elle connait, et dont quelques-uns au mains ne sont pas d'invention r~cente. 85 The Iranian material is not very amenable either. Much ancient material was lost in the reforms of Zoroaster. The most that the Iranian material can do is to prove the Indo-Iranian character of the Indian material. The Greek material is not very suitable either, because other cultures have had such an influence on it86• He sums up: "Des ses premiers pas, l'histoire se heurte au mysticisme brahmanique, au moralisme de Zoroastre, au " 87• The Germanic, Celtic, Slavic and Latin traditions present problems of a different kind. The Germanic material, in this case the Scandinavian material on the history of religion, has been passed on at the periphery in Iceland. The Celtic and Slavic myths are preserved in the popular tales. In these the gods have become heroes and thus susceptible to the hold of Christianity. "Tres n5cents, ils representent cependant une tradition qui n'a pas ete rompue par une reforme ou une refonte religieuse" 88• The stories associated with the Tuatha De Danann and Vladimir and his court offer parallels with the ancient tales of the gods89• The Latin traditions do not bear any trace of reforms either, for .. .!'esprit de ces paysans et de ces soldats n'est pas speculatif, mais pratique, et Ia seule evolution qu'on puisse saisir il l'epoque ancienne, peut-etre en partie sous !'influence des Etrusques, c'est Ia multiplication des rites, des cer~monies, des pretres et des colleges de pretres.90 The introduction of the Greek myths here had the same effect as the advent of Christendom for the Celtic and Slavic traditions. Here too the old tales of the gods were transformed to become historical narratives associated with figures like Anna Perenna. In this way Latin mythology became human91 • Research on these traditions should lead to the formulation of general transformational laws, taking into account the technological, psychological, sociological and geographical circumstances92• In his Le crime des Lemniennes, Dumezil concludes that the narratives concerning the women of Lemnos are also historicised myths93 • In 1929 he goes into the Indian problem in more detail. The Vedic texts are certainly the oldest, but not the most informative.
85 86 87 88 89 90 91 92 93
Dumezil, 1924a, p.X. ibidem, p.XI, XII. ibidem, p.XII. ibidem, p.XIII. cf.: ibidem, p.159, 160, 197 ibidem, p.XIII. cf.: ibidem, p.134-135, 138, 139, 147, 157, 197; Dumezil, 1925a, p.215. Dumezil, 1924a, p.288. Dumezil, 1924b, p.40, 41, 54, 55, 57.
METIIODOLOGY
33
...seulement Ia notion
d' me semble mains claire; milieu societe v~dique, livres v~iques, soit; mais qui dit epoque dit totalit~, et )'expression semble inclure que les Veda refl~tent, laissent reconstituer toute Ia pen~e. Atous ses niveaux, des temps ou ils furent r~ig~. Cela est sllrement faux. 94
v~ique,
In the same year he returns to the transformation of myths into historical acts: Au fond, Ia m~fiance qui frappe les Fastes est mains dirig~ contre le ~te que contre Ia mythologie latine en g~n~ral. Cette notion fait sourire, parce que, d~ l'~poque d'Ovide, Ia mythologie latine n'avait plus gu~re que !'aspect d'une fausse histoire rapport~e aux premiers temps de Ia ville, aux cycles de Romulus ou de Numa. Mieux vaudrait certes une mythologie dans toute sa dignit~, mais celle-ci n'est pas n~gligable. II s'est simplement fait tr~s ttlt, dans le Latium, devant !'invasion de Ia culture grecque, un travail d'historicisation des mythes nationaux, comme !'Iran, I'Irlande, les Galles, Ia Scandinavie et maint autre pays l'ont pratiqu~ un peu plus tard, quand ·le christianisme ou l'islamisme y eurent recouvert les paganismes indigenes. Les mythologies d~chues, A Rome comme ailleurs, tournent au conte de f~e ou Al'histoire, les dieux y cedent Ia place aux g~nies ou aux rois, et leurs aventures, quand elles ne se dissolvent pas dans un folklore anonyme, constituent AIa Cit~ convertie un noble pa~. Saxo Grammaticus parte des princes Fro et Balderus, les r~cits ~piques de I'Irlande parlent des rois Tuatha De Danann, les Mabinogion du roi Manawyddan et du roi Pwyll, Firdousi du roi Jemsid et de l'usurpateur Zohak: et c'est lAde , par l'epopte, ou des hommes prestigieux font !'application et Ia
132 133 134 135 136 137 138 139
Dumezil, 1951b, p.llO. ibidem, p.l12. ibidem, p.114. Dumezil, 1952c, p.l49. Dumezil, 1952b, p.44. Dumezil, 1953a, p.l31 = 1970, p.l21. Dumezil, 1956b, p.15. ibidem, p.16.
40
CHAPTER TWO dtmonstration des principes que les dieux incarnent et des conduites qu'ils commandent. 140
After denying the historicity of the events described by Livy, Dumezil goes into the Sabine war in more detail (c). "Le recit de la premiere guerre de Rome est tres bien construit, mais evidemment construit"141. We shall come across a development of this construction later (see below, p. 177ff.), but if we understand it as a narrative, we can state that: .. .le rtcit de Ia constitution de Ia societe romaine compl~te est exactement parall~le recits, non plus , mais mythiques, connus chez d'autres peuples indo-europtens.... 142 ~des
Myth is no longer situated in another world, but here on earth, and humans have assumed the roles previously held by the gods. However, (c) "peut-etre meme ont-ils ainsi mieux joue le role d'exempla, d'encouragements, de justifications, qui est l'un des offices de Ia mythologie"143• Referring to the narratives of the Nartes, Dumezil remarks that this tendency to use epic or mythical material in the construction of ancient history is a universal phenomenon144. In 1968 Dumezil distinguishes between different kinds of historicisation. The transposition in the Mahabharata ("mythe savamment humanise sinon historicise" 145) is a process of direct translation of tales of the gods into the human world. Comparing this with other types of historicisation, he says: (alb) Dans son principe, l'entreprise n'est done pas difftrente de celle ..... par laquelle les annalistes romains, Saxo Grammaticus ant constitut l' des premiers rois de Rome, du Danemark, dans un cas ~ partir d'une structure de concepts, dans l'autre, particuli~rement proche du Mahabharata, ~ partir d'une structure de personnages divins.146 He then proceeds to quote from the work of Marcel Granet, who had earlier discovered the same process of historicisation in ancient China. History is written using constructive principles and schemes 147. China is a good example, but it is no exception: (c)
140 Dumezil, 1958a, p.74. 141 Dumezil, 1966, p.78. 142 ibidem. 143 ibidem, p.87. 144 ibidem, p.84. 145 Dumezil, 1968, p.21. 146 ibidem, p.239-240. 147 Granet, 1926.
METIIODOLOGY
41
Quand un peuple, ou plut~t quand les intellectuels d'un peuple soot amen~. par autonome ou A !'imitation de voisins prestigieux, A se repr~nter les , comment prorederaient-ils autrement? lis n'ont que des !~geodes ou des , auxquels ils donnent un sens d'ensemble (g~n~ralement le sens d'une cr~tion progressive de Ia r~lit~ sociale dont ils ont l'ex¢rience) en les distribuant, en les orientant dans les conceptuels, anciens ou r~cents, qui s'imposent Aeux. 148 r~flexion
Good examples can be found in the Hesiodic myth of the golden, silver, bronze and iron ages, in Indian conceptions of the succession of the four yugas, in the Iranian royal history of Firdiisi and in the Scandinavian royal history of Snorri and Saxo Grammaticus149• If the beginnings of Rome may be seen as a "succession organisee d'evenements significatifs; autrement dit une histoire construite" 150, it has its place within a broad tradition. The same is true of the twentieth-century view which sees a historical conflict in the conflict between the Aesir and the VanirlSl. There are clear correspondences between Indian mythology and Roman history, Dumezil claims in 1970. How are they to be explained? de toute Ia mythologie ...."167• In 1948 he notes: "Comme toute chose chez les Germains, c'est dans Ia guerre, c'est en fonction
159 160 161 162 163 164 165 166 167
Dumezil, 1939a, p.XII. De Bello Gallico, VI, 21. ibidem VI 22. ' ' Dumezil, 1939a, p.11. De Bello Gallico, VI, 21, 3. Dumezil, 1939a, p.1534. ibidem, p.155. ibidem, p.155-157. Dumezil, 1940, p.99.
METIIODOLOGY
45
de Ia guerre que Ia Souverainete est en grande partie con~ue.... " 168• This is in agreement with his remarks from 1949 and 1951:
Tyr, lui, avec le glissement guerrier qui marque toutes les repr~ntations germaniques... 169 LA (sc. in Scandinavia, W.B.) encore, dans des formules, dans des ltgendes, s'observe un groupement de dieux, qui avec des glissements, ...... propres A tout l'ensemble ~ermanique (comme Ia gtntralisation des prroccupations guerri~res) ... .t 0 Despite "petits ecarts locaux'', Dumezil claims in 1952171, the Germanic pantheon displays the same tripartite structure as that of other Indo-European cultures. Si I'on a bien compris les tlements et les articulations de cette structure tripartite, il me parait impossible de ne pas Ia reconnaitre dans les paganismes germaniques. Les fronti~res des fonctions y sont simplement, parfois, un peu deplarees. Comme Ia guerre, notamment, etait Ia grande affaire des anciens Germains, elle a souvent envahi le premier niveau .. 112 In 1958 Dumezil attaches far-reaching consequences to the concept of the "glissement guerrier" for the other levels (seep. 103). War has penetrated the field of the first function, a "gauchissement" which entails considerable consequences 173 • In an article from the same year, Dumezil refers to the same passage from Caesar. The Germans recognise only one kind of person: the warrior 174 • In 1959 the passage from Caesar is cited again, preceded by a concise characterisation: ...dans l'ideologie et dans Ia pratique des Germains, Ia guerre a tout envahi, tout colort. Quand its ne se battent pas, ceux dont cesar a donne Ia premi~re et saisissante esquisse ne songent qu'aux prochains combats... 175 Apart from a brief remark in 1968, ("Chez ces belliqueux Germains" 176), and a reference in 1971 to the RHR article of 1958m, Dumezil only returns to this
Dumezil, 1948b, p.450. Dumezil, 1949a, p.161. Dumezil, 1951a, p.230. Dumezil, 1952a, p.24. ibidem, p.23. Dumezil, 1958a, p.57-58. Dumezil, 1958d, p.l. Dumezil, 1959b, p.65-66. Dumezil, 1968, p.lOO. m Dumezil, 1971, p.88.
168 169 170 171 172 173 174 175 176
CHAPTER TWO
46
"glissement" on one other occasion. In 19n he deals with the characteristics of the Germanic religions. The first is:
Avant tout, Ia deuxi~me fonction, les besoins et Ia morale de Ia guerre ont pen6tr6 profond6ment dans Ia premi~re, Ia souveraine. lis ne l'ont ni sup~rim6e ni annex6e,....mais ils l'ont impr6gn6e et gauchie dans leur propre direction. 1 The passage from Caesar proves that this applies to the Southern Germans. It is also applicable to the Northern Germans, Dumezil adds. Dumezil found the same shift in Rome, as his work from 1939 already showed. In 1941 it leads him to set Mars, the god of the second function, in the foreground. In 1958 the "martialisation" of the Germanic Tyr is compared with that which enabled him to identify Quirinus and Mars 179• Summary: The Gennanic shift: the "glissement guerrier" The Gennanic material differs from that of other Indo-European areas. Above all, the divine representatives of the first function are marked by characteristics which are typical of the second function. Nevertheless, Dumezil considers the Gennanic material to be useful evidence in support of his theory, provided it is assumed that Caesar's labelling of the Southern Gennans as wannongers is co"ect and applies to both the Southern and the Northern Gennanic cultures. Dumezil is consistent in his application of this concept. This auxiliary hypothesis enables him to explain why the Gennanic material does not fit in with the Indo-European model.
The reforms of Zoroaster Before we can use the Iranian material for reconstruction, we have to know how it has been filtered. As was seen above, Dumezil used a model of moralising (see above, p. 32). At first he did not see much scope for using Iranian material:
Dans !'Iran, oil les faits soot plus embrouill6s et oil l'on sent d~s les premiers textes Ia volont6 des reformateurs, nous laissons aux specialistes le soin de prospecter eux-memes leur domaine... 180 ' After discussing the Proto-Indo-European tripartite system, Dumezil wonders about the Iranian material
178
179 180
Dumezil, 19TI, p.l84. Dumezil, 1958a, p.66. Dumezil, 1940, p.66.
METHODOLOGY
47
De cette mythologie des trois fonctions....... n'y a-t-il plus trace dans les constructions mazd~ennes? En particulier Zoroastre a-t-il Ia~ tout perdre de ce qui constituait le cadre Ie plus g~n~ral de Ia pen~e de ses .Xres?181
This turns out not to be the case, but we have to take account of an Iranian "champ ideologique". Dumezil delineates this in terms of two main components, the theological and the moral182• Theologically it is important that only Ahura Mazda is God. The ancient texts also reveal a strong opposition between good and evil. Then there is also an opposition between spirit and matter. There are clear ideas about the life after death. Zoroaster assumes an important position in relation to salvation. Morally there is a strong resistance to ritual. Furthermore, life is entirely dominated by the opposition between good and evil. Finally, the cow enjoys particular prestige. However, those who were responsible for the Zoroastrian and the later Mazdean reforms have distorted the ancient polytheist pantheon. This meant that the gods of old were deliberately made subject to the single God and were moralised 183 • Both reforms were implemented deliberately184• In his conclusion, Dumezil compares the Roman and Iranian mentalities. Roman mentality is political and national, Au contraire, dans le zoroastrisme pur, les formes politiques n'int~ressent Ia vie religieuse que par les obstacles qu'elles lui opposent ou les services qu'elles rendent. 185 Roman mentality has a historical framework, Le zoroastrisme au contraire, n'(:tant pas enferme dans son cadre de temps et d'espace, cherche ailleurs son appui: il pense non pas historiquement, mais tbeologiquement. 186
The ancient myths are thus no longer to be found in history, but the "armature philosophique" has been preserved in the accounts of the single God. Zoroaster's actions have their reasons: "c'est que Ia tradition polytheiste qu'il voulait transfigurer, sublimer, < monotheiser >, mais non pas detruire en tant que moyen d'analyser le reel....'' 187• In a comparison between Cicero and Zoroaster published in 1950, the "reforme moralisante" crops up again188•
181 182 183 184 185 186 187 188
Dumezil, 1945, p.55. ibidem, p.64-69. ibidem, p.86-87. ibidem, p.94, %. ibidem, p.187. ibidem, p.l88. Dumezil, 1947a, p.89. Dumezil, 1950b, p.453.
48
CHAP1ERTWO
Dumezil concludes that there is no other god besides God, and the knights have become crusaders189• The other deities of the polytheist system have been set aside, but .. .les rMormateurs appartenaient A Ia classe sacerdotale.... , et its n'ont pas voulu renoncer A ce qu'il y avait d'explicatif dans le systtme des fonctions dont its rejetaient les personnifications divines... 190
This analytical model to explain the work of the Creator has been replaced by the Amesha Spenta system (see p. 77ff.). The fact that the later Mazdean reforms demonised the gods of the second and third functions (see p. 75f.) indicates that those who implemented these reforms were in fact members of the first class of priests191 • Dumezil comments on the transformation of the gods into Amesha Spentas: La rtforme y a gagnt de conserver sans rien en perdre une analyse commode et nuanree du rtel visible ou invisible, et en m~me temps de corriger, de mettre au pas les idtologies des deux fonctions inftrieures et les conduites de leurs reprtsentants humains, des guerriers d'lndra notamment, en les alignant sur les postulats moraux de Ia premitre. 192
The reforms were carried out by priests, and the morality of the men representing the first function was also imposed on those representing the second and third functions. Belligerence and fertility cults were curbed 193• But why was the ancient pantheon implicitly preserved? The answer it that the priests were attached to this framework, and ...sans doute aussi parce que les hommes, les Arya Aqui s'adressait leur prtdication et qu'ils voulaient persuader ou contraindre, ttaient eux-m~mes attachts A cette forme de penste et done qu'il fallait leur fournir un substitut exact de ce qu'on leur enlevait .... 194
The ancient pantheon was preserved, on the one hand; the essence of the ancient gods was purged, on the other. This resulted in rather insignificant figures. This was part of the strategy of Zoroaster and his followers. It also explains why the study of the Amesha Spentas is so problematical. They recall the ancient gods, but they often appear to be mutually interchangeable 195 • 189 190 191 192
ibidem, p.454. ibidem, p.455. Dumezil, 1952, p.16. ibidem, p.20. 193 Dumezil, 1958, p.40. 194 ibidem, p.44. 195 ibidem.
METIIODOLOGY
49
Dumezil also refers to the possibility of persuading and convincing in 1965, when in a discussion of another case he states: On reconnaitra Ia un nouvel exemple de Ia mani~re beureuse dont Zoroastre s'est servi, pour pr~ciser sa doctrine ou en faciliter Ia propagation, des moyens que lui fournissait Ia tradition ....196 Gerschevitsch has criticised this model of the process of change. He viewed the strategy attributed to Zoroaster as an attack on his integrity197• In reply, Dumezil points out once again that Zoroaster came from a priestly background and grew up in a highly complex polytheist theology, so that: "il est naturel qu'il ait sauve le plus possible de son savoir en l'adaptant a son intuition monoth6iste" 196• Dumezil returned to Gerschevitsch's criticisms in 1977. Now199 he contrasts the latter's "roman recent" with the "exposes, classiques et bien informes" of Widengren200 and Duchesne-Guillemin201 • Dumezil points out that Zoroaster "n'a pas voulu perdre une philosophie dont i1 continuait d'apprecier la valeur, celle qu'exprimait leur multiplicite et leurs rapports" 202• Dumezil's conclusion is that Zoroaster's strategy should not be seen as an assault on his integrity; on the contrary, it deserves our admiration. Summary: The refomts of Zoroaster Initially it appeared difficult to include Iranian material in the reconstruction. Thanks to the auxiliary hypothesis of "the refomts of Zoroaster", in which the actions of Zoroaster and his followers came in for special emphasis, Dumezil discovered the postulated Indo-European situation "in translation". Firstly, this auxiliary hypothesis enabled him to look for the tripartite gods among the Amesta Spentas; secondly, it provided an explanation for the later demonisation of the gods of the second and third functions. Dumezil offers different explanations of the need for these "translations". In very general temts, he describes an Iranian "champ ideologique", an Iranian mentality, which is responsible for them. The key elements in this are monotheism and the opposition between good and evil. Dumezil also claims that the refomters who belonged to the priestly class did not want to abandon the old system. One reason for this was their desire to use it to convert those who were familiar with it. Furthemtore, the demonisation of the gods of the second and third functions provided a way of controlling the cults related to them. 196 197 196 199 200 201 202
Dumezil, 1965, p.48. Gerschevitsch, 1959. Dumezil, 1965, p.49, n.l. Dumezil, 1977, p.41, n.l. Widengren, 1954; 1955a; 1955b. Duchesne-Guillemin, 1%2. Dumezil, 1977, p.147.
50
CHAPTER TWO
The Celtic shift The Celtic material plays a marginal role in the reconstruction of the tripartite model. Consequently, the modalities of the shift are hardly elaborated. However, on a number of occasions Dumezil does make statements which are relevant in this connection. In 1939 he points out that Caesar mentions Mercury as the main deity, the Gallic counterpart to Lug Samildanach, "le dieu de taus les arts", "et cela tient sans doute ...... au grand developpement, a l'excellence, au prestige des arts et des techniques dans l'ancien monde celtique ..." 203 • In 1942 Dumezil indicates the correspondence between Rome and Ireland. Both cultures set their mythology in a historical setting. The Tuatha De Danann were a people who lived in Ireland before the invasion by the present settlers. After suffering defeat at the hands of the latter, the sons of Mil, they retreated to the barrows. All the same, there are important differences between these datable events and the history of Rome. The Irish live in a supernatural world. They do not oppose human and divine, and it is in fact difficult to disentangle the two. This penetration of nature by the supernatural is diametrically opposed to the Roman cosmovision204• Another aspect of the Celtic gods is their imperialism. By this Dumezil means that the gods have a clearly defmed functional area, but that they can also be seen to operate outside this area. This imperialism, ...qui pousse toute reprtsentation religieuse A outrepasser sa d~finition et A s'annexer le plus de mati~re possible. Les Irlandais, et sans doute d~jA les Gaulois, ant pou~ fort loin cette tendance. 205 In 1958 Dumezil points out that the Celtic pantheon is in fact unsuitable for reconstruction. The analogy between the Gallic gods mentioned by Caesar and the Tuatha De Danann creates more problems than it can solve206• The Indian shift We have already come across a number of elements bearing on the Indian shift in the preceding discussion, such as the opposition between the Roman and the Indian "champs ideologiques" (see above, p. 35, fig.4). In this case it is the Indian material which clarifies the Roman material, but Dumezil has never published a systematic elaboration of the Indian shift. All the same, there are a few random remarks of his which are pertinent. In 1947 he notes that, apart from the central position which Indra later came to occupy, the functional gods "ont subi le destin commun des representations vediques et ont connu des
203 204 205 206
Dumezil, 1939a, p.9. Dumezil, 1942, p.65-68. Repeated in 1949a, p.172ff., mentioned in 1966, p.123, n.1. Dumezil, 1944, p.26. Dumezil, 1958a, p59.
METHODOLOGY
51
assimilations, des liaisons, des repartitions contradictoires"207• In the same year Dumezil remarks that Vedic mythology is hardly any more functional than Greek mythology, and is certainly less so than Scandinavian mythology. The hymns with a functional structure are archaisms. Moreover, Indra, who became the king of the gods, has totally disrupted the system208• Disappearance of the functional structure In 1958, as we have seen, Dumezil indicated that the Celtic material did not lend itself for reconstruction. In the same section he makes the same claim for the Greek and Slavic material. Most of religion in Greece is not IndoEuropean, and so little is known about Slavic paganism that this field only lends itself for brilliant hypotheses209• He had already pointed out the unsuitability of the Greek material in 1924210• The Greek component came to play a larger part later on, as in the comparison between the Centaurs and Gandharvas, or that between Ouranos and Varuna. In 1940, however, Dumezil emphasises the total unsuitability of the Greek material: ...en Grece, !'objet de notre pr~sente investigation ne subsiste meme pas puisqu'aucune forme de mythologie ni de soci~t~ grecque ne s'articule plus sur le scMma indo-euro~en des ..211
In 1947 too he states that the tripartite structure is no longer present there."Les Grecs ont cesse tres tot de penser le monde, Ia societe divine et meme humaine, selon Ia hierarchie des trois fonctions primitives...."212• In an article from 1954 in which Dumezil inquires into the Greek tradition for expressions of tripartition, he begins by remarking that the Greek contribution to the formulation of problems and their solution is extremely marginal. The Indo-European component in the Greek tradition is minute 213 • This view of the Greek material has vanished to a large extent by 1987. Referring to the work of Vian, Yoshida and Sergent, Dumezil now considers that there is a considerable body of tripartite material214• The tripartite structure has also disappeared as a conscious framework in those cultures which still speak an Indo-European language. Only sporadic traces have remained. This is mainly due to the technological explosion which took place at the beginning of the present century, Dumezil notes215 • 207
208 209 210 211 212 213 214 215
Dumezil, 1947b, p.128. Dumezil, 1947a, p.157-158. Dumezil, 1958a, p58-59. Dumezil, 1924a, p.84ff. Dumezil, 1940, p.93. Dumezil, 1947a, p.l37, cf. p.221. Dumezil, 1954e, p.26; cf. 1968, p581. Dumezil, 1987, p.162. Dumezil, 1987, p.188f.
52
CHAPTER TWO
Summary: The Celtic shift; The Indian shift; Disappearance of the functional structure The idea of compiling "champs ideologiques" for the various cultures also led to an Indian "champ idiologique". Dumezil categorised Indian culture as one in which thinking was done in fables in a cosmic, philosophical, absolute, dogmatic, moral and mystical way. Other important factors are the Vedic "confusionnisme", which led to the disappearance of typical distinctions between the gods, and the central position which Indra later came to occupy. Dumezil also compiled a "champ ideologique" for Ireland and compared it with the Roman one, though he fonnulated it in a less explicit way. In both cultures mythology was placed in a historical setting, but Irish historical tradition displays a strong interpenetration of the natural and the supernatural. Dumezil also indicated the imperialist nature of the gods, by which they often exceed their functional boundaries and are difficult to identify. Other cultures proved less amenable as the bearers of evidence to support Dumezil's theory. These were the Greek and Slavic cultures, as well as the Celtic cultures at a later stage of Dumezil's theoretical development. Later on he claimed that the Greek material was amenable to tripartite analysis. Summary: The auxiliary hypotheses There were large differences between the variants of the tripartite model in the various Indo-European cultures. This was difficult to reconcile with a genetic comparison. In order to make the material amenable to this end, Dumezil posited a number of extremely general characterisations of the different cultures as auxiliary hypotheses, deriving them from the "champs ideologiques" in many cases. These characterisations made it possible to understand why the material which was actually found deviated from what the theoretical model would lead one to expect. At the same time they justified the attempts to compare material drawn from widely different traditions. Although these characterisations of very broadly defined cultural units are less emphatically based on the "champs ideologiques" in his later work, their presence is still implicitly assumed. Dumezil elaborated his auxiliary hypotheses with special reference to the Roman, Gennanic and Iranian material. The extremely important auxiliary hypothesis of "historicisation" acquired a number of different meanings in the course of time. The explanation of the Iranian "translation" also underwent changes. Dumezil remained consistent in his application of the Gennanic auxiliary hypothesis. The paradigmatic framework
This fmal section is concerned with two methodological questions. One of the classic problems in the history of religions is that of the connection between social relations and relations within the pantheon. Are ideas about the supernatural determined by culture and are they a reflection of
MElHODOLOGY
53
social relations? Or is the reverse true, and does our picture of the supernatural have a major impact on how we arrange our society? In both cases the question is the same: what is the relation between a society with a tripartite arrangement and the tripartite representations of the divine world? We shall return to these questions in the discussion of the qualities included. A different problem is that concerning the vision of the participants. Do participants in the culture under analysis recognise themselves in our conclusions? To put it less abstractly: are the tripartite expressions discovered by analysis conscious or unconscious products of the participants? Were they aware of the functional implications of their representations or not?
Culture/the supernatural Over the years Dumezil presents three different responses to the question of the relationship between the social data and the pantheon. The first two of these can be labelled synthetic statements. They refer to a reality, whether it exists now or has existed in the past. The third response consists of analytic statements. These refer to the manner of thought.
First response: primacy of social relations. There is a monocausal relation in which the social data are fully determinant. The world of the gods is thus a projection of social relations. The presence of a tripartite pantheon must thus correspond to a tripartite society. It is from this point of view that Dumezil says that Indra may be associated with thunder and lightning because of his heavenly battles, but: "l'indien lndra n'est pas l'orage personnifle, certes; il est Ia projection divine de Ia classe des guerriers" 216• The same goes for Quirinus. He is not a god who incarnates the grain, but a representative of the "classe sociale dont les soins sont consacres au grain... " 217• The implication of this for the study of the general frameworks and the main forms of expression of Indo-European religion is that: ...comme, chez les demi-civilists, Ia conception du monde et celle de Ia socittt, Ia hitrarchie des dieux et celle des hommes soot le plus souvent parall~les, cette recherche revient A definir, A Ia fois et indifftremment, comment les lndo-Europtens concevaient Ia division et l'harmonie de leur corps social et comment ils ajustaient les provinces de leurs principaux dieux.218
216 217 218
Dumezil, 194lb, p.391. Dumezil, 1944, p.78. Dumezil, 194lb, p.399.
54
CHAPTER1WO
Dumezil was later to change his mind on this score. This is already evident in 1948 in his analysis of the correspondences between the Germanic god Loki and the Ossetic Syrdon. Are the similarities due to comparable social organisations219? On this occasion he criticises this monocausal explanation. Elle a contre elle d'etre obscure, de faire uniquement appel aux rapports de causalit~ qui lient !'~tat social, ~nomique, culture! d'un peuple et les produits de son imagination. Ces rapports sont r~els, cette causalit~ joue. Mais dans des conditions et dans des limites qu'on ne peut pr~ciser. 220
Second response: primacy of social relations, but bearing in mind the highly conservative nature of ideology. There is a monocausal connection in which social data are fully determinant, but their effect is delayed. There can thus be a large discrepancy between the actual social relations and those which served in the past as the basis for the arrangement of the pantheon. One of Dumezil's methodological principles is therefore: Les mythes ne meurent pas toujours en meme temps que disparaissent, sous des influences diverses, les formes de vie politi3ue ou ~conomique, les rites religieux qu'ils avaient d'abord contribu~ Amaintenir. 1 In other words: "Par bonheur, les Iegendes sont moins vulnerables que les moeurs et souvent, veuves discretes ou joyeuses, elles leur survivent" 222• It is therefore possible to oppose the conservative "plan religieux'' to the innovatory "plan politique" 223 • And in 1973 Dumezil refers to the literary effect of a system of archaic representations, ...de plus en plus ~loign~ de Ia pratique et de l'actualit~ ou, plus pr~ci~ment, qui ne continuait Acommander et Aexpliquer Ia r~alit~ sociale, sous une forme bien alt~r~e d'ailleurs, que dans l'lnde, mai~ui partout restait compris, utilisable comme cadre de pen~e, comme philosophie. A similar statement dates from 1979, where he says:
219 220 221 222 223 224
Dumezil, 194&, p.247. ibidem, p.254. Dumezil, 1941b, p.392. Dumezil, 1941c, p.131. Dumezil, 1944, p.181. cf. 1949a, p.44. Dumezil, 1973a, p.7.
METIIODOLOGY
55
est souvent plus conservatrice, plus tenace que Ia pratique et, meme de consid~rables transformations des structures sociales, se maintient longtemps dans ses expressions libres, notamment dans Ia mythologie et dans Ia L'id~ologie
apr~
litt~rature. 225
Third response: primacy of the representations of the supernatural. Indo-Europeans tend to think in triple divisions. One of the forms in which this fmds expression is a tripartite social division. In this way Dumezil can begin by describing the "conception" which shaped the ideology, and then go on to state: "Cette conception tripartite devait inevitablement commander aussi Ia structure sociale ... " 226• He makes a similar remark in 1958. This time, however, Dumezil combines this response with the second one: Aussi bien n'est-ce pas le d~tail authentique, historique, de )'organisation sociale tripartie des lndo-Euro¢ens qui interesse le plus le comparatiste, mais le principe de classification, le type d'id~ologie qu'elle a suscit~ et dont, r~al~e ou souhait~. elle ne semble plus ttre qu'une expression parmi d'autres. 227 Social organisation is an expression of the principle of classification, although at an earlier stage the principle of classification must have been the result of social organisation. In 1966 Dumezil states that what is at issue is not the reconstruction of events, but the elucidation of "structures de pensee" 228• In 1987 Dumezil concurs with Eribon's formulation: "En somme.....vous eclairez plus les structures mentales que les faits" 229 •
Summary: Culture/the supernatural An important methodological issue is the question of the relation between social relations and ideas of the supernatural. Dumezil provides three responses to this question. In the first he claims that social relations detennine ideology,· in the second he makes the same claim, with the proviso that ideology is highly conservative and reflects archaic social relations; in his third response he claims that ideology detennines the representations of social relations. The first two types, which are closely related to one another, belong to the earlier period of Dumezil's work, but he continues to apply them, particularly the second, for a long time. The third response is clearly a later development.
225 Dumezil, 1979a, p.59-60.
Dumezil, 1952d, p.38. Dumezil, 1958a, p.18. 228 Dumezil, 1966, p.8. 229 Dumezil, 1987, p.176. 226 227
56
CHAPTER TWO
Conscious/unconscious In 1940 Dumezil defends himself against a number of unspecified attacks. One of them accuses him of adjusting the facts to fit his model. He rejects this charge. All he does is perceive, for: "les usagers des mythes, des rituels, des formules, avaient conscience du systeme dont notre seule tache aura ete de montrer l'ampleur et l'antiquite"230• However, this general statement soon comes to stand beside other points of view. For instance, the Romans of the historical period are supposed to have only partially understood the most explicit tripartite traditions231 • This can also be seen in Propertius: "Properce ne savait sans doute plus que les trois tribus avaient ete fonctionelles" 232• The same is true of Snorri and the authors of the Mahabharata233 • Dumezil was also criticised because it was alleged that the tripartite system is a universal phenomenon and not specifically Indo-European (see the explicit formulation of this position by Brough, above p. 19). This may well be the case, but according to Dumezil the difference lies in the fact that the Indo-European cultures were aware of this system, while the non-Indo-European cultures were not. It is precisely the fact that they were aware of this tripartite system that led to its successful observance, for example in the French monarchy (clerge, noblesse, tiers-etat), the Russian state (party, Red Army and workers) and the Nazi state (Partei, Wehrmacht and Arbeitsfront) 234• He repeats this defence in 1958 in a section on the three functions and the "nature des choses"235 , and again in 1979236• ...pour une societe, ressentir et satisfaire des besoins imperieux est une chose; les amener au clair de Ia conscience, reflechir sur eux, en faire une structure intellectuelle et un moule de pensee est tout autre chose. 237
His reiterated emphasis on the deliberate use of the functional ideology by the authors of the Roman annals is in closer agreement with the former criticism and defence238•
230 231 232 233 234 235 236 237 238
Dumezil, 1940, p.IX. Dumezil, 1941c, p.247. Dumezil, 1943b, p.273. Dumezil, 1948c, p.105. Dumezil, 1949a, p.242. Dumezil, 1958a, p.22-23. Dumezil, 1979a, p.79-81. Dumezil, 1958a, p.23. Dumezil, 1951c, p.318, 319, 320, 330; 1952b, p.44; 1958b, p.432, 434; 1975, p.253; 1982, p.166.
METHODOLOGY
57
Summary: Conscious/unconscious We may wonder whether the participants were aware of the tripartite framework in which they thought, according to Dumezil. He provides two answers to this question. On the one hand, in defending himself against the thesis of a universal diffusion of a tripartite ideology beyond the borders of the Indo-European cultural area, he emphasises the deliberate use of this framework by Indo-European cultures. On the other hand, in certain specific situations he presupposes an unconscious observance of the tripartite framework.
CHAPTER 1HREE
THE TRIPARTITE SYSTEM The following chapters contain a discussion of the application of the model. Dumezil detects relics of the tripartite system in the various Indo-European traditions: the world of the gods, social relations and historical traditions.
Tripartition in India According to Dumezil, we find relics of the Indo-European tripartition in India in the social system of classes, in the varnas and in the world of the gods.
Social tripaTtition The interpretation of the three Aryan varnas, on which Dumezil's model is based, has not changed in any fundamental way over the years. In 1938 Dumezil describes Indian society as virtually feudal. Hierarchy and its maintenance through the law of succession were important. In a comparison with Rome he states: "l'Inde vedique et postvedique amplifiait le pouvoir du raj, developpait le prestige des brahmanes et durcissait en castes les trois anciennes classes" 1• This view of the social division remains practically unchanged. The system of the three varnas is the original one, while the further extension and division into the numerous jatis are a part of the later "foisonnements" 2• India certainly took this model for society seriously, but did not invent it3. This is identical to his view in 1958. After providing a description of the three constituents of Indian society, he says: Ces groupes fonctionnels, hierarchises, sont en prmc1pe fermes chacun sur lui-mt!me par l'heredite, par I'endogamie et par un code rigoureux d'interdictions. Sous cette forme classique, il n'est pas douteux que le systeme ne soit une creation proprement indienne, posterieure au gros du Rgveda.... Mais une telle creation ne s'est pas faite de rien; elle n'a ete que le durcissement d'une doctrine et sans doute d'une pratique sociale preexistantes. 4
He had already remarked on the nature of this former social practice in 1948, in connection with the defence of his view of Roman social tripartition. He
1 2 3 4
Dumezil, 1938a, p.197. Dumezil, 1941c, p.42. ibidem, p.45. Dumezil, 1958a, p.7.
1HE 1RIPARTI1E SYS1EM
59
discusses the latter after noting on the castes: "Les castes sont sans doute une alteration apportee a l'organisation indo-europeenne, indo-iranienne, par }'influence d'indigenes non-arya"5• Now6, and again in 19497, he sees the world of gods and humans as subdivided into an aristocracy with a functional division and the corresponding tasks, on the one hand, and an amorphous mass, on the other. I shall return to this in my discussion of the Roman data (seep. 83f.).
Tripartition in the pantheon After having detected the existence of a tripartite social division in India, Dumezil wonders whether a corresponding theological tripartition can be found there. Does the pantheon display the same structure? After all: Les moins philosophes d'entre les peuples dits ne se repr~sentent pas leur etre social sans du meme coup, en syst~mes homologues et solidaires, se repr~senter les autres organismes r~els ou conceptuels de leur ex~rience et aussi l'univers qui les contient tous. 8
He had already made less systematic statements on the Indian gods. He saw the Maruts as companions of Indra, or as a group linked with the vaishyas. The gods of the second class were "personalities", like Indra, Varuna, Soma, Rudra, etc.9• The representations of the Indian pantheon become more systematic in 1941. We now meet Mitra and Varuna in the world of the gods as representatives of the first function, which can be broken down into two aspects, and Indra and Vrtrahan as representatives of the second function. The third function is more problematic: Quant ~ Ia F~condit~, ~ I'Abondance, ~ Ia Richesse, elles ont des patrons moins stables, plus nombreux, souvent collectifs.......qui ne peuvent etre d~finis aussi rapidement. 10
Dumezil considers that the gods who are summoned during the ritual foundation of the fire altar ahavaniya are archaic. The "vache d'abondance", Kamadhuk, is summoned to nourish every living thing. Then follows the prayer to Kamadhuk11, which Dumezil translates as follows: "Produis (comme lait)
5 6 7
8 9 10 11
Dumezil, 1948a, p.156. ibidem, p.155-161. Dumezil, 1949a, p.213-219. Dumezil, 1941c, p.57-58. Dumezil, 1938a, 1939, p.196, n.2. Dumezil, 1941c, p.59. ShB, VII,2,2,12 = VSamh, XII,72.
60
CHAPTER THREE
leurs desirs, vache d'abondance, a Mitra-Varuna, a Indra, aux deux Pushan, aux creatures et aux plantes" 12• He comments on this passage:
A~,
a
Comme Ia premiere classe par Mitra-Varuna et les Guerriers par lndra, les Eleveurs-Agriculteurs sont ici representes et par les deux Arvin, appeles aussi Ndsatya, tiers etat divin, medecins des dieux, donneurs de sante et de fecondite, et par Pflshan, dieu des troupeaux et specialiste de l'abondance. 13 Apart from the interpretation of Pushan, which a footnote tentatively connects with the fourth varna, that of the non-Aryan shUdras, no substantial change has been made to this interpretation. One consequence of this view is that the 1940 vision of the Ashvins as warring gods can no longer be maintained. In the first edition of Mitra-Varuna, Dumezil commented on Indra: "du mains figure-t-il en bonne place, a cote de ces autres divinites combattantes que sont les A~" 14 • This passage has been modified in the 1948 revised edition, now that the Ashvins have been incorporated into the tripartite system: "du mains figure-t-il en bonne place, a cote de ces divinites bienfaisantes noncombattantes, que sont les A62• Benveniste's objections to the functional interpretation of the "gene" lead Dumezil to remark: "Ia le~n objective qui s'en degage est que cette page d'Herodote n'est pas claire"63• It is wiser not to choose between an exclusively ethnic or an exclusively functional solution. " ...les informateurs d'Herodote et Ia tradition des Scythes eux-memes devaient combiner tant bien que malles deux conceptions..."64• But this looks like a case of lip-service, for later in the same book he only mentions "classes/etats sociales"65 • Similarly, it is the exclusively functional interpretation which is maintained in 194466, in 194867, in 195:z68 and in 195369• In 1962 Dumezil claims that he had doubts about an ethnic or a functional solution70, but such doubts are absent from this period. The volte-face comes in 1958. Now Dumezil remarks on the Herodotean passage: II me parait aujourd'hui certain qu'il faut, avec E. Benveniste, rendre "g6nos" par : les Scythes comptent quatre tribus, dont une est Ia tribu chef. Mais toutes ant, r6elle ou id6ale, Ia me me structure.71 He also subscribes to Benveniste's view in 196072• But in 1962 he changes his mind again73• Now Dumezil looks for a way of widening the argument to support the functional analysis, which nevertheless remains confmed to a fossilised tradition or an ideal concept, without the need for social realia to correspond to it. The arguments he uses are: Paralatai may be connected with paradhata, the nickname of Haoshyanha, the ftrst of a series of three functionally qualifted Iranian kings; Traspies and Katiaroi are based, among others, on "aspa" and "cahra", horse and meadow. We shall see that the third function can be broken down into the opposition "steer-horse", which are thus
60 61 62 63
64 6S 66 67 68
69 70 71 72 73
VII, 8,18-19. Dumezil, 1953b, p.264. Dumezil, 1960, p.141. Dumezil, 1949c, p.153. ibidem. ibidem, e.g. p.203, 220. Dumezil, 1944, p.168. Dumezil, 1948b, p.451. Dumezil, 1952d, p.40. Dumezil, 1953b, p.263f. Dumezil, 1%2a, p.l89. Dumezil, 1958a, p.9-10. Dumezil, 1960, p.141, n.S. Dumezil, 1%2a, p.187-202.
TilE 1RIPARTITE SYSTEM
71
represented by these two ''gene". The Aukhatai are left with the second function, and the association with "aojah-", "force fysique" 74, a typical notion for the second function, reinforces this interpretation. The three groups thus have a functional and not an ethnic colouring, but they are not comparable with the Indo-Iranian castes/classes. In 1968 Dumezil comes back to this article again, but he is now uncertain about the interpretation of the names of the Katiaroi and the Aukhatai. The four "gene" are still exclusively "types ideaux''. Now he states: Je suis port6 A admettre que les d6signations des quatre "gene" sont rest6es purement s¢culatives, sans correspondre....Aune r6alit6 sociale, ni d'ailleurs A une r6alit6 g6ographique. 75
In 1979 he offers a similar interpretation along functional lines in which the "gene" are not a social reality. La premi~re (triad of brothers, W.B.), celle des fils de Targitaos,.. pr6sente les prototypes... des desservants humains des trois fonctions A l'int6rieur de toute soci6t6 scythique. 76
Dumezil therefore sees no need to change anything in the "jugement bien pondere" of Xazanov, who had remarked: "il ( = Dumezil, W.B.) ne les considere que comme un model ideal ... , eloignee de Ia pratique reelle" 77• Finally, in 1983 he opposes the "divisions fonctionelles" of Kola.xals and his brothers to the "support territorial" divided among the sons of Kola.xals78• Summary: Scythian tripartition Herodotus mentions three sons of Targitaos. The youngest of these becomes king because he received four objects. Dumezil interprets these four objects in a tripartite way. A confirmatory argument by Benveniste ( 1938), according to which the four objects co"espond to three functions, is given a hesitant reception at first (1941), but is enthusiastically welcomed later (1960). The three sons are the ancestors of four "gene". Are these "gene" ethnic or functional groups? Dumezil provides different answers to this question over the years, backing them up with various arguments. In the first interpretation the "gene" are conceived as functional groups, which served to prove a division of Indo-Iranian society into
74 15 76 77 78
ibidem, p.202. Dumezil, 1968, p.452. Dumezil, 1979c, p.178. ibidem, p.200. Dumezil, 1983, p.79.
72
CHAPTER 1HREE
classes (1930). Despite doubts on this position in 1941, Dumezil maintains it until 1958. In 1958 and 1960 Dumezil interprets the "gene" in an exclusively ethnic way. From 1962 the "gene" are interpreted functionally again, but a greater distance from social realia is now assumed. They are ideal types (1968), far divorced from actual practice (1979). In 1962 Dumezillooks for philological arguments. At first he finds them for all four "gene'~ but in 1968 he retracts these arguments to a large extent. In the discussion of the relation between culture and the supernatural (above p. 53ff.) we encountered three different responses by Dumezil. The first, the primacy of social relations, coincides with his first interpretation here. His third response, the primacy of the ideological representations, is also a supposition of his third interpretation here. The structure of the "gene" is now an expression of thought that takes place within a tripartite framework. Tripartition among the families of theNartes In 1930 Dumezil also resorted to material from the Caucasian Ossetes. This is a people who speak an Iranian language. The popular legends are concerned with the Nartes, who are subdivided into a number of families, three of which are the most important. Dumezil cites Tuganov's account of this subdivision as follows: "Les Boriatli possedaient beaucoup de betail, les Aliigatii etaient puissants par l'intelligence, les Ahsiirtiigkatii se distinguaient par Ia vaillance et Ia force" 79• Dumezil remarks that this division is not connected with the ancient tripartition. The Boriatae are just as bellicose as the Aexsaertaegkatae. Drinking parties take place among the Alaegatae. It is difficult to show the relation with intelligence here either. In a footnote, however, he refers to an article that was to appear in/A, 1930. In this article he claims: C'est exactement !'~tat de chases indo-iranien sous sa forme Ia plus claire et Ia plus simple, l'~levage n'ayant meme pas encore red~ sa place A !'agriculture. La seule diff~rence est que les divisions des Nartes sont non plus des classes (endogames etc.), mais des families (exogames, etc.): diff~rence grave, assur~ment, mais explicable par Ia diversit~ des temps et des moeurs; comment les Osses, qui depuis longtemps ne connaissent plus de classes dans leur pratique sociale en attribueraient-ils A leur Mros? Le souvenir des a surv~cu, mais s'est adapt~ aux cadres sociaux modernes. 80
This account is striking because Dumezil claims that the families are hard to distinguish from one another in the tales. Tuganov's argument is supported by the fact that the drinking parties take place among the Alaegatae. This feature can be explained by the Brahmanic organisers of the Soma orgies. This family is not involved in the conflicts of the other two either. Although many heroes can
79 Dumezil, 1930a, p.19. 80
Dumezil, 1930b, p.126.
TilE TRIPARTITE SYSTEM
73
be found among the Boriatae, it is the Aexsaertaegkatae who are the real warriors in the tales. Many conflicts break out between Aexsaertaegkatae and Boriatae after a member of the first family kills the cattle and steals the horses from the second: "forme naturelle d'un conflit entre "guerriers" et "eleveurs" "81 • Dumezil thus concludes that the Ossetic families are not Indo-Iranian classes, but nevertheless they do recall an earlier social reality. Other Caucasian peoples have borrowed the legends of the Nartes, but not the tripartite division. "c'etait done bien Ia une tradition specifiquement osse, scythique, aryenne" 82• Dumezil has not substantially modified this interpretation over the years. The social analysis, summarised in Tuganov's classification, has continually served as additional proof of an Indo-Iranian origin for a division by classes into three groups83• However, he does try to add new arguments gradually to Tuganov's classification and the remarks in JA from 1930. In 1945 he analyses the names of the families. The name of the family which is associated with the third function varies in the different traditions: Boriatae, Boratae, Bariatae, Baratae. Dumezil finds the latter variant in the name of the Ossetic Pluto: Bara-styr, "great Bara". This does not surprise him, "Ia troisieme fonction et Ia troisieme classe etant souvent en rapports avec Ia Terre, le sous-sol et le monde infernal"84• The Ossetic lag, "man", comes from *aryaka, and similarly Alaeg- (ae = a) can be derived from *aryaka, with a long initial vowel. In the light of this possibility Dumezil says: "dans des petits enfants" .171 The Quirinalia, the feast of Quirinus, can now be explained too. Although nothing is known about this feast, it coincides with the Fornacalia, a feast in which the storage of the grain plays a central part. Dumezil puts the word "coincide" between inverted commas because he does not regard this connection as fortuitous 172• Besides this material connected with the cult, there are also sources which identify Mars and Quirinus: "Mars enim cum saevit Gradivus dicitur, cum tranquillus est Quirinus" 113 ; and "Quirinus autem est Mars qui praeest paci et intra civitatem colitur, nam belli Mars extra civitatem templum habuit" 174 • Mars and Quirinus are identified here. In 1938 Dumezil comments as follows: "c'est }'ancien Quirinus autonome, avant son assimilation a Mars, qui etait " 175 • The identification of the two gods is a reflection of a social process. The men who till the soil as Quirites in times of peace are identical with the men who defend their country as Milites in times of war. Quirites and Quirinus have the same roots, "d'ailleurs obscur" 176• Dumezil is not satisfied with this solution in 1944 because it is "tres verbale" 177• He now postulates a development to explain Quirinus. At first Quirinus was the god of the third function, as the activities of his flamen show, but this was modified. In the first place, peace does not exclude war; on the contrary: Nous sommes simplement aiguilles sur une conception ancienne, peut-etre primitive, de Ia paix vigilante et annee consideree comme domaine de Quirinus, conception qui enrichit, sans y contredire, Ia figure deja acquise pour ce dieu. 178
This aspect of Quirinus grew in importance as Rome came increasingly to stand at the head of an imperium. He was no longer the god of the Romans as Quirites who had carried out their duty as Milites. In the Imperium Romanum
171 172 113 174 175
Dumezil, 1966, p.268. Dumezil, 1938a, p.195; 1941c, p.90. Servius, in Aeneidem, 1,292. ibidem VI 860. ' ' Dumezil, 1938a, p.195. 176 Dumezil, 1941c, p.89. 177 Dumezil, 1944, p.204. 178 ibidem, p.201.
TilE TRIPARTITE SYSTEM
93
with Rome and the provincia, Quirinus became the god of the latter. The Pax Romana was accomplished in the provincia by soldiers, Paix ntcessairement militaire, mains en ce sens que, comme toute paix, elle a ses origines dans Ia guerre, que parce qu'elle se maintient grace~ un appareil guerrier considtrable. 179 Seen in this light, Mars and Quirinus are variants of the vision of war. "Mars et Quirinus ne s'opposent plus seulement comme laguerre et la paix, mais comme la guerre destructrice et une sorte de guerre conservatrice"1PJ). As we saw in the preceding section (see p. 83f.), in 1948 Dumezil devised a theory which was also pertinent to the Quirites. A close connection gradually developed between Quirites and Tities, which led to an equivalence between Quirites and the third function. There was also a fusion of Quirites and the plebs in the opposition between patricians and plebs. As the god of the mass which could be mobilised, Quirinus already had an "aspect aire-pacifique" at the same time as his "aspect agricole" from very early on181 • The first aspect is not subordinate to the second. Dumezil had already pointed out the collective character of the Quirinalia in 1944182• He developed this idea further in 1949183 • Now he views the agricultural aspect of Quirinus as secondary. Quirinus is compared with Aryaman, one of the Adityas, the ruling gods in India, who is the god of the "masse arya". Aryaman is also linked with the third function. As god of the "masse romaine", Quirinus is not only connected with the third function, but he has even taken the place of more specialised gods in that function. Besides the equivalence between Quirinus and Mars, there is also that with Romulus. Tradition has it that Romulus appeared after his death with the message that he should not be lamented, but that he should be worshipped as a god under the name Quirinus 184• Dumezil was taken aback by this association in 19411ss. After all, Romulus belongs to the first function and his deity is Jupiter. The explanation at this point is that as god of the third function Quirin us can also be god of the world of the dead. In Iranian tradition, for instance, we also fmd that Yima, the first king, becomes the god of the underworld after his death. This explanation is in agreement with the interpretation of the activities of the flamen quirinalis during the Larentalia. In 1944, however, Dumezil notes that there is not a single text in which Romulus or Quirinus are seen as "roi des morts" 186, which calls for a change to his interpretation. He now concentrates on the evolution of the
179 liM)
181 182 183 184
1ss 186
ibidem, p.205. ibidem, p.206. Dumezil, 1948a, p.169, n.2. Dumezil, 1944, p.197. Dumezil, 1949c, p.163-164. Plutarch us, Romulus, 28. Dumezil, 1941c, p.183. Dumezil, 1944, p.204 (This problem had already been indicated in 1941c, p.186).
94
CHAPTER TIIREE
career of Romulus (this is discussed explicitly in the following section). At this point Dumezil identifies the following phases: (a) The first phase is that of Romulus and Remus. It is a purely religious phase of the foundation of the city. (b) After the killing of Remus for a religious contravention comes the phase in which the second function is in the foreground. Romulus is now assisted by Lucumo in his war against the Sabines. (c) This is followed by the joint government with Titus Tatius. This is the phase of the third function. (d) These three phases are concluded with the order to found the Imperium Romanum on modesty and courage. There is a divine form for his appearance in this apotheosis ready to hand. He appears as the god of constructive warfare and the Pax Romana, Quirinus. In 1949 the main emphasis is on Romulus' role as representative of the Roman mass. De plus, on con~it que cette essence de ait favoris6 !'assimilation de Quirinus ~ Romulus, cr~ateur de Rome. 187 ...on con~it que Romulus, devenu le premier et unique chef de Ia communaut~ romaine rassembl~e (et que! qu' ait ~t~, dans Ia de Rome, son apport fonctionnel) ait pu s'assimiler ~ Quirinus, non pas en tant que dieu de Ia troisieme fonction, mais dans sa valeur premiere de patron du corps social romain. 188 Changes take place in the different stages of Romulus' career. In 1966 they have consequences for the relation between Romulus and Quirinus. Now the phase of Romulus and Remus is classified as a third function phase. Dumezil describes the sequence Romulus I, II and III: ...il est, en association avec son jumeau, le berger, tier et noble certes, mais le berger. Puis ... Romulus devient homme de guerre ... et ... Ia variante ~ trois races l'associe au technicien ~trusque de Ia guerre, Lucumon. Entin apres le syn~cisme ... il est le roi .. 189 Once Quirinus had been identified naturally with Romulus I, his character was automatically expanded. Quirinus followed the career of Romulus, which enabled him to be identified with Mars. trouv~, dans Ia l~gende, fils de Mars, ce qui, restant litt~raire n'~tait pas grave; mais il a aussi volontiers revetu l'uniforme guerrier de Romulus II, qui contredisait sa nature. 190
II s'est
187 188 189 190
Dumezil, 1949a, p.227. Dumezil, 1949c, p.l66. Dumezil, 1966, p.255. ibidem, p.256.
TIIE TRIPARTITE SYSTEM
95
Summary: Jupiter, Mars and Quirinus Jupiter, Mars and Quirinus are very important gods for the reconstruction of the Indo-European tripartition from 1938 onward. Dumezil makes few additions to the evidence that was available then, but he concentrates on a number of problems. One of these is the fact that in the case of Jupiter and Mars there are a number of references to activities which fall outside their functional framework. Dumezil suggests various solutions over the years. With respect to the interpretation of Jupiter, the first is the idea that this transgression of borders is a nonnal consequence of Jupiter's universalist evolution (1944; 1949). At a later stage Dumezil denies these transgressions (1966). The discussion concerning the interpretation of Mars contains the following features: although Mars is the god of war, he is also invoked for fertility, but it is important to distinguish between primary and secondary characteristics (1938; 1941; 1944). In 1944 and in a later development dating from 1966 Dumezil emphasises the confonnity of Mars' activities to his function. The third problem is even more thorny. Quirinus was initially seen as an originally agricultural god, secondarily identified with Mars as a result of the equivalence between Quirites and Milites (1938; 1941). Dumezil then elaborated a theory of the Pax Romana and the idea of the Romanum Imperium to explain this equivalence (1944). Later on Quirinus is seen as the god of the Roman mass, a Roman Aryaman, in whom the agricultural aspects only play a secondary role (1949, cf. 1966). The figure of Acca Larentia reflects the way that Dumezil changed his mind about the third function. At first ( 1941) it is her grave which is the main feature. Dumezil postulates a relation between the underworld and the third function. Later on (1944) he supposes that she was originally a fertility goddess. Subsequently (1949) she is a "genie" offertility, sensuality, wealth and generosity. In 1952 she is the incarnation of sensuality, riches and generosity as the "courtisane-donatrice". Her role as the "mother" of Romulus and Remus can be understood now that the third function is no longer confined to agriculture ( 1966). Although Dumezil's arguments and defences undergo considerable changes, his interpretation of Jupiter, Mars and Quirinus as the gods of the first, second and third function respectively remained unchanged. Romulus and his companions and the Roman kingship cycle
We saw that in dealing with the Roman traditions one has to take the effect of historicisation into account. This has led to two cycles in the history of Rome. The first concerns the career of Romulus, his successive engagement with Remus, Lucumo, Titus Tatius and his fmal position as unique sovereign; in the second Romulus stands at the head of a series of successive kings. His first successor is Numa. Both kings are seen as representatives of the first function in 1939. As we shall see, Dumezil divides the first function into two complementary poles: the "dedoublement indo-europeen de Ia souverainete'' results in "le Majestueux" and his complementary opposite, "l'lnspire
96
CHAPTER TIIREE
furieux'' 191 • The majestic Mitra with his laws and Varuna with his magic display this opposition. On Romulus and Numa Dumezil says: .. .Ia mythologie historique romaine juxtapose de meme ses deux fondateurs, le Luperque Romulus (dont les gardes sont dits les Celeres, les ), dont toute Ia vie est faite de puissantes violences- et le ltgislateur pacifique Numa (dont le premier soin est de dissoudre les Celeres).wz. This first move is further elaborated in 1940. In India the Mitra-Varuna opposition is reflected in two contrasting dynasties 193• These are the "dynastie solaire" of the "mithraic" Manu, the legislator king, and the "dynastie lunaire" of Purftravas, king of the Gandharvas. His grandson Nahusha is struck dead by lightning after having insulted the religious dignitaries. The traditions relating to king Purftravas reveal a "morale des Gandharva" on which Dumezil remarks: ...elle est semblable ~ celle des premiers Luperques, de Romulus et de ses rudes compagnons, menant en ce monde Ia vie d'un autre monde, brigands, violents, indociles aux regles comme aux avertissements. 194 The Roman kingship cycle is not a dynasty, but Dumezil does distinguish a "serie pieuse et serie belliqueuse" 195 • Romulus and the third king, Tullus Hostilius, form the second series. Tullus is even more warlike than Romulus and he is killed in the same way as Nahusha. Numa and Manu are comparable, and the first concern of Ancus Marcius, the grandson of Numa and the fourth king, is to restore the cult which had been seriously neglected under Tullus Hostilius. Dumezil concludes: L' romaine a done distribut en regnes successifs les provinces secondaires de Ia souverainett, celles qui ne col'ncident pas avec les deux provinces antithttiques exprimtes dtj~ successivement par les regnes de Romulus et de Numa. Ces provinces secondaires, l'Inde vtdique, qui pensait par dieux plut6t que par Mros, les distribuait plus simplement entre les Aditya, autour du couple essentiel de Varuna en de Mitra. 196 Only the first function is now represented by the successive kings. In 1942 this leads Dumezil to wonder:
191
Dumezil, 1939a, p.39.
192 ibidem. 193
Dumezil, 1940, p.27-42.
194 ibidem, p.64. 195 ibidem, p.65-66. 196 ibidem, p.131.
1HE TRIPARTITE SYSTEM
97
Est-il vraisemblable que Ia mythologie historique des Romains soit arretee en si bon chemin? Que, apr~s avoir si minitieusement analyst dans Romulus et dans Numa Ia premi~re fonction, et dans Romulus et dans Tatius les rapports de Ia premi~re et de Ia troisi~me sinon Ia troisi~me en elle-meme, elle ait ntgligt de caracttriser et de justifier Ia seconde fonction, celle de Ia force guerri~re? Cette lacune est a priori improbable. 197
This lacuna is filled by Tullus Hostilius, the third king, who was described by Livy as "ferocior etiam Romulo" 196• The cycle is extended further in 1947. Anchises and Aeneas predict the future. The future kings of Rome are presented in a brief passage in Virgil's Aeneid199: Romulus and the auspicia, Numa and the sacra, followed by Tullus Hostilius, the king of the second function who occurs in a phrase dominated by the anna. Dumezil concludes from this: II est done naturel de penser que, au del~ de Tullus, l'intention et le proredt de Virgile resteront les memes et que les personnages suivants seront aussi dtfinis par quelques mots exprimant ce que Ia tradition considtrait comme le ressort central de leur action. 200 The terms jactantior ("le trop ambitieux'') and populares aurae ("Ia faveur populaire") would then apply to Ancus. Unlike the ftrst three, however, they bear a negative connotation. According to Dumezil, both terms can best be viewed in the light of an account of the succession like that to be found in Dionysius201 • Here with the assistance of a group of men Ancus forges a conspiracy against Tullus and murders him and his children. He then sets ftre to the house and proceeds to give the blame to Jupiter. Ainsi il existait une tradition suivant laquelle Ancus ttait un ambitieux qui, fort de Ia faveur d'un grand nombre de Romains, aurait acquis le trOne par un meurtre doublt d'une imposture sacril~ge. 202 Dumezil then supposes that the ftrst four kings represent the three functions. As for Ancus, he wonders: ...l'oeuvre d'Ancus correspond-elle ~ ce que nous attendons non pas simplement d'un roi psychologiquement , mais d'un roi qui serait comme le symbole de Ia troisi~me fonction sociale telle qu'elle avait tvolut ~ Rome, le
197 Dumezil, 1942, p.79. 196 1,22. 199 VI, m-&23. 200 Dumezil, 1947a, p.165. 201 11,35. 202 Dumezil, 1947a, p.l72.
CHAPTER 1HREE
98
symbole de ce que les Romains de Ia fin du lYe si~cle sentaient dans le riche concept qu'ils subordonnaient encore aux concepts et ?203
This "Quirinus concept", as we saw earlier (seep. 92f.), consisted of a depiction of Quirinus as the "Mars qui praeest paci". The pax is here the Pax Romana, and Quirinus acts as the defender of the Imperium Romanum. The wars are defensive and constructive and are waged with insight. Like Romulus, he instills bravery and modesty. Ancus can be accommodated easily within this concept. He wages wars, in which the opponent is in principle not eliminated, but incorporated in the Imperium; wars which have a beneficial economic effect. Moreover, he is the king who was responsible for the population of the Aventine, the plebeian hill par excellence. In addition, there are a few other fragmentary items which link Ancus with the third function. During his reign the rich Demaratus, who was later to become Tarquinius Priscus, visits Rome. This rich man is also the first king to be elected by the people as such204• The affair with Acca Larentia also takes place during the rule of Ancus. As we saw in the previous section (p. 91f.), her cult belonged to the third function. Ancus lived on the site of the later aedes Larum. The Lares, associated with the soil, fit in perfectly with the third function. Furthermore, among the cults initiated by Ancus is that of Venus Calva, and despite the complexity of Venus, she is certainly connected with such matters as sensuality and fertility. This functional division of Romulus, Numa, Tullus Hostilius and Ancus also explains the apparent duplications. The representative of the first function with the aggressive aspect has a preference for the second function, while the representative of the first function with the mild, legislative aspect has a preference for the third function. The analogies between Romulus and Tullus Hostilius, on the one hand, and those between Numa and Ancus, on the other, were thus predictable. It remains an open question why the kingship cycle consists of seven kings. Dumezil's reply is that the fourth, filth and sixth kings display duplication and can thus be seen as equivalents of one another. This tendency to form groups is a general feature of the third function. We also see very close, almost fraternal bonds between both Ancus Marcius and Tarquinius Priscus and between the latter and Servius. We already came across this association between Romulus and Titus Tatius. These dual entities are also typical of the third function, whose representatives are often twins. Dumezil launched this view of the kingship cycle in 1947 and maintained it until 1956. After 1947 the filth and sixth kings disappear, and he no longer mentions the close association between the fourth and the filth and between the ftfth and the sixth. This inevitably means a change in the second edition of Mitra-Vantna. The citation from 1940 quoted above now reads:
203 ibidem, p.175. 204
ibidem, p.l90.
1HE 1RIPARTI1E SYS1EM
99
L' romaine a done distribu6 en r~gnes successifs soil les provinces secondaires de Ia souverainet6, celles qui ne corncident pas avec les deux provinces antith6tiques exprim6es d6j~ successivement par les r~gnes de Romulus et de Numa, soil les incidences des deux fonctions inferieures sur Ia souverainete.Z!JS In 1950 Dumezil emphasises the systematic nature of the cycle with the four kings: "Ces quatre rois forment un systeme qui n'est pas une vue de l'esprit, mais que les Romains comprenaient, affrrmaient, admiraient en tant que systeme"206• In 1956 Dumezil goes on to note that the interpretation of Romulus, Numa and Tullus Hostilius has not been very controversial, but that the interpretation of Ancus Marcius has sparked off discussion. "II en a ete autrement pour le quatrieme roi, Ancus Marcius, et en partie, me semble-t-il aujourd'hui, a bon droit" 207• Now the career of Romulus and the kingship cycle are linked together. Romulus and Remus are the representatives of the third function. The sequence up to the historical boundary of Ancus is: third function, i.e. the twins; frrst function (aggressive), Romulus alone; frrst function (legislative), Noma; second function, Tullus Hostilius208• This view is also presented in the "bilan" of 1958209• In 1966 Romulus and Remus are also the representatives of the third function, but now more stress is laid on the career of Romulus. The twin phase is followed by the phase of Romulus and Lucumo, the second function. Afterwards Romulus reigns, frrst with Titus Tatius, later by himself: this is the aggressive aspect of the first function. This is followed by the phase of Noma's rule: this is the legislative aspect of the frrst function 210• Another change is made to this scheme in 1968. This time Dumezil virtually returns to the 1947 system. Referring to his 1947 analysis, he has the following to say about the material relating to Ancus: ... toutes ces 16gendes, quelles qu'en soient les origines, silrement diverses, n'ont en facteur commun que de relever toutes- richesse et g6n6rosit6, s&luction et volupt6, sant6 - de Ia < troisi~me fonction >.211 Romulus and Remus no longer appear to be connected with the third function. Romulus and his companions generally represent the frrst function212• There is a reference to the discussion of Ancus Marcius in the reprint of the 1956 work published in 1969, but Dumezil no longer entertains any doubts as to the
Z!J5 206 207 208 209
210 211 212
Dumezil, 1948d, p.192, italics mine, W.B. Dumezil, 1950b, p.458. Dumezil, 1956b, p.20. ibidem, p.20-21. Dumezil, 1958, p.SO. Dumezil, 1966, p.255. Dumezil, 1968, p.281. ibidem, p.291.
100
CHAPTER TIIREE
possibility of a functional interpretation. Another change is that the career of Romulus and the kingship cycle are no longer combined213•
Summary: Romulus and his companions and the Roman kingship cycle Dumezil used the historical data on Rome as the material for two scenarios. In the beginning of the history of Rome, Romulus collaborated successively with Remus, Lucumo and Titus Tatius. Afterwards he wielded sole power. Dumezil interprets this scenario in tenns of tripartition. The Roman kingship cycle contains the following kings: Romulus, Numa, Tullus Hosti/ius, Ancus Marcius and others. Dumezil also interprets this scenario in tenns of tripartition. The following figure presents a schematic view of these interpretations of the career of Romulus and the kingship cycle: Figure 8: Career of Romulus and the kingship cycle (I, II and III stand for first, second and third function; 1a, 1b, 2 and 3 stand for the first function in its two aspects, the second and the third function)
Romulus+ comp. Remus Luc. 1939 1940 1942 1944 1947 1949 1956 1966 1968 1969
I I I III III I
II II II II II
Tatius Rom. 1a 1a 1a III 1 III 1a 1a III 1a I 1a III 1a 1a
Numa Tullus Anc. Anc. +etc 1b 1b 1a 1b 1b 2 1b 1b 1b 1b 1b 1b
2 2 2
3 3
2 2
3 3
3
A number of interpretations are already established here, but other interpretations are more problematic. The functional qualification of the "Lucumo phase" and of Tullus Hostilius, the opposition between an aggressive and a legislative aspect of the first function, as expressed by Romulus and Numa, are present from early on.
213
Dumezil, 1969b, p.15-16.
TilE 1RIPARTITE SYS1EM
101
The first phase in the career of Romulus is more problematic. In the first instance we saw that Romulus and Remus were assigned the first function, to be followed by an interpretation of the Dioscurian twins within the third function, and later still by the replacement of Romulus and his companions in the first function. The "Titus Tatius phase" seems to be settled, although Dumezil has no qualms about interpreting it in tenns of the first function in 1966. In the kingship cycle it is the interpretation of Ancus which is particularly awkward. Dumezil responded to criticism on this score, and Ancus disappeared from the functional field between 1956 and 1968. On the one hand, the historicisation appears to have a conservative effect. The division of the first function has disappeared at the theological level and Jupiter has assumed all its aspects, but it continues to exist in the historicised version. On the other hand, the historicisation is an accurate reflection of the theological evolution; Ancus Marcius follows in the footsteps of Quirinus. Tripartite Gennan material Odhinn, Thorr and Freyr
In this section I will begin with Dumezil's difficulties regarding the Germanic traditions, before discussing the way in which the "glissement militaire" acquires its shape. I then proceed to examine Dumezil's use of the gods of Uppsala, the curse of Egill and the sons of Mannus, and conclude this section with the remaining evidence and the "bilan". The problem The Germanic material on tripartition relates in the ftrst instance to the world of the gods. Dumezil began investigating this in 1939, when he followed Vendryes in indicating a problem. In the course of his study of the Indo-European religious vocabulary, Vendryes came across a correspondence between the Indo-Iranian and the Italo-Celtic branches. He explained this similarity in terms of the presence of a priestly class in all four groups214• Since the Germanic peoples did not have priests, as Caesar had indicated215 , they did not share this common vocabulary either. Dumezil pointed out the problem that this raises for studying this material: "La part d'incertitude, d'hypothese inveriftable, est done ici plus grande que sur les domains voisins"216• The society of the gods was not protected by a priestly class. It was subjected to processes of combination and levelling out, as we saw above, which made it as difficult to
214 215 216
Vendcyes, 1918. De Bello Gallico, VI,21. Dumezil, 1939a, p.8.
102
CHAPTER TIIREE
classify as the human world. We do not fmd here the social and functional distinctions which were to be found among the Romans and in India under the influence of the Indo-European tradition. Nor do we fmd a clear-cut world of the gods comparable to the Roman or Indian pantheon. All the same, Dumezil implicitly assumes that this structure can be found in the triad comprising Odhinn, Thorr and Freyr. Otherwise his defence of what he sees as Caesar's sharp-sighted perception would be devoid of content. Dumezil comments on the triad of Sol, Vulcanus and Luna which Caesar records: Odhinn n'est pas si mal . sinon exprimt, par le mot latin Sol; Vulcanus ne rtpond pas mal au Thl}rr armt du marteau; ~ur NjOrdhr, qui a ttt d'abord une dtesse.... .l'ttiquette lunaire n'est pas ... bizarre.. 17
But initially the functional analysis here has an organisational function: it is used as a peg on which to hang the various themes, as Dumezil himself states in the preface: Chaque chapitre est une petite monographic, et !'ensemble peut se repartir sous trois rubriques: mythes de Ia souverainitt magique, mythes de Ia force militaire et des conquetes, mythes de Ia vitalitt, de Ia ftconditt, de Ia richesse.218
A number of familiar and more recent insights are contained under these headings, but Dumezil does not apply the tripartite model as such. Later on, however, he does touch on important themes, such as the distinction within the first function - the Mitra-Varuna opposition - and the influence of the devotion to warfare which Caesar had already mentioned. The "glissement militaire" The emphasis on the military aspect which Dumezil assumes to be a characteristic of the Germanic cultures helps him to explain the link between Odhinn and war219• This influence is also found in Odhinn's berserkir. Dumezil compares Odhinn with Varuna and the bcrserkir with the Gandharvas, "peuple de Varuna", who in turn resemble the Maruts, "societes a armes", the companions of Indra. The berserkir thus represents a combination of groups from the first and second functions.
217 218 219
ibidem, p.12. ibidem, p.xii-xiii. ibidem, p.153-155.
TilE TRIPARTITE SYSTEM
103
II ne semble pas qu'on trouve, dans le monde germanique, une trace aussi neue des anciens distinctions: meutes d'hommes-fauves et troupes de guerriers-champions s'y sont fondues en un type de socittt magico-militaire sptcifiquement germanique. 220 In 1952 Dumezil points out two shifts, though without relating them to one another at this stage. The concentration on warfare has transformed Tyr into Mars in the interpretatio romana and Odhinn into a god devoted to war. Moreover, some people or social groups have seen the importance of the activities of Thorr for fertility. These two shifts are associated with one another in 1958 in a relation of cause and effect. Odhinn has taken over the warrior function from ThOrr, and the latter has then assumed the fertility function of Freyr "compensatoirement"221 • Freyr was thus left with no alternative but to assume that aspect of the third function which is related to human reproduction. Adam of Bremen characterises Freyr as the god who must be invoked during the marriage ceremony. Dumezil also assumes this link between the two shifts in his interpretation of the Rigsthula222• We shall see below (p. 109f.) that he saw in it an account of the Germanic social classes. They were symbolised in the sons of Rigr: Jarl, Karl and Thraell. These were under the authority of the first king, the young Konr, Konr ungr, the youngest son of JarI. Jarl himself, who performs the second function as a warrior, has the colour white. Karl, who performs the third function as a farmer, is red. The slave Thraell is black. This colour symbolism is Indo-European, but a shift has taken place in this situation, for normally white is the colour of the first function, red of the second, and black (or green, blue) of the third. Dumezil illustrates this in a table (see figure 9), and comments on it: Le tableau ... montrera comment Ie dtbordement d'Odhinn sur Ia fonction
et celui de Th(')rr sur Ia fonction de ftconditt peuvent expliquer cette du blanc et du rouge. 223
guerri~re,
The gods ofUppsala This militarisation is not adduced by Dumezil in the context of his discussion of a text from Adam of Bremen224 • This describes a group of images in a temple in ancient Uppsala consisting of Thorr, Wodan and Fricco. Thorr must be invoked in times of famine, Wodan in times of war and Fricco on the occasion of a marriage. Is ThOrr here a third function god? Dumezil points out the importance of the distinction between the deity's marginal activities and the
220 221 222 223 224
ibidem, p.155. Dumezil, 1958a, p.57. Dumezil, 1958d. ibidem, p.8. IV,26. See: Dumezil, 1939a, p.130-131.
104
CHAPTER TIIREE
Figure 9: The Germanic shift and the Rigsthula.225 Etatindoeuropeen
Theologie scandinave
blanc-magie
Odhinn
Rigsthula Konrungr
-{magie guerre
rouge-force, guerre
ThOrr
Jarl- blanc
- [ combat atmosferique pluie fecondante
-
blue (etc.) -fecondite
Karl-rouge
Freyr - - fecondite terrestre Thraell - noir
"noyau de sa personalite" 226, as he had done in the discussion of Mars "agraire" versus Mars "bellator". The fact that the rain caused by the god of thunder is important for fertility does not make him a fertility god. Apart from this text, the evidence for the existence of the triad is limited to the tripartite interpretation of the curse of Egill and the similar interpretation of the sons of Mannus. In 1940 Dumezil sums this material up as follows: En particulier, dans diverses formules, dans des groupements divins, dans Ia division gtntrale de Ia mythologie, on reconnait Ia grande tripartition des fonctions cosmiques et sociales. 227
And similarly in 1941: "Les anciens Scandinaves notamment distinguent trois religieux, qui s'expriment en mainte circonstance par des triades"228• The group of gods from the temple of Uppsala is now cited as an example of a triad of this kind (without mention of any kind of shift). In 1944 Dumezil labels this group of gods as the "triade type de Ia mythologie ancienne" 229• 225 Dumezil, 1958d, p.8, figure 9. Dumezil, 1939a, p.132. Dumezil, 1940, p.94. 228 Dumezil, 1941c, p.155. 229 Dumezil, 1944, p.166. 226 227
TilE 1RIPAR1ITE SYSTEM
105
The curse of Egill In a footnote Dumezil mentions that Egill curses king Erik of Norway when he is forced by Erik to leave the country230• Egill does so whilst invoking Odhinn, Njordhr, Freyr and the Landass. If we interpret the Landass as Thorr we come up with the tripartite series, in which Njordhr and Freyr naturally stand together for the third function. Dumezil interprets them in this way without any discussion231 • In Dumezil's translation, the strophe runs: Puissent les dieux (blind) chasser le roi; les dieux (godh) lui feraient ainsi payer le rapt de mes biens! Que les dieux (rogn) et Odhinn s'irritent! · Ase du pays (Landoss), fais que l'oppresseur du peuple doive fuir de ses terres! Que Freyr et NjOrdhr harssent le fl~au des hommes qui ravage les sanctuaires! 232 With a reference to F. Jonsson's acceptance of the Landass = Thorr interpretation, this text is elevated to the status of the main functional formula in 1946233 • Dumezil also provides a new item of proof. In the Skirnisma.I the giantess Gerdhr is addressed by Skirnir. The gods are angry with her because of her refusal to marry Freyr. Skirnir says to her234: Irrite contre toi est Odhinn, irrite contre toi est le plus distingue des Ases (=Th~rr), Freyr te harra, Fille toute mauvaise, qui as provoque Ia vive col~re des dieux. 235 It was his interpretation of this text which in 1985 provoked Dumezil's virulent reaction236 to the criticisms by R.I. Page 237• Page had accused Dumezil of working with suppositions rather than facts, using the curse of Egill to substantiate his claims. Instead of taking lines 5 and 6 and lines 7 and 8 together, Dumezil now takes lines 5 and 8 and lines 6 and 7 together in the second part of the strophe, in support of the argument mentioned below as c. (He had already done so in the English translation of Les dieux des Gennains: Gods of the ancient Northmen 238.) The translation now runs: Faites fuir l'oppresseur du peuple hors de ses terres (afjordhum), Freyr et Njordhr! Que l~se du land (Landliss) ex~cre l'oppresseur des hommes qui viole le lieu sacre (ve)!239
230
231 232 233 234 235 236
237 238 239
Egilssaga, ch.44. Dumezil, 1939a, p.32, n.l. Dumezil, 1945, p.53. Dumezil, 1946, p.63. Skirnismill, 33. ibidem. Dumezil, 1985, p.278-298. Page, 1978-1979, p.66-67. Dumezil, 1973b. Dumezil, 1985, p.280.
106
CHAP1ER 1HREE
The arguments for identifying this Landass as ThOrr are as follows: (a) We should expect a god of the same order to stand next to Odhinn, Njordhr and Freyr. (b) The existence of the triad Odhinn, ThOrr and Freyr is an established fact, according to Dumezil: La triade Odhinn Th6rr Freyr est bien ~tabli, comme structure, par !'analyse du culte upsalien et par des expressions formulaires: elle couvre par ses trois termes Ia totalit~ des fonctions n~cessaires AIa vie d'une soci~t~: totalit~ qu'il est opportun de mobiliser dans Ia circonstance pr~sente. 240
(c) The linking of lines 5 and 8 recalls the final scene in the Lokasenna, where Loki has profaned the holy spot and ThOrr acts as the defender of sacred places at the end of the poem.
The sons of Mannus Dumezil referred in another footnote in 1939 to the passage in the Germania in which Tacitus recounts that a certain Mannus had three sons, who were the ancestors of the (H)erminones, the Istraeones and the Ingvaeones241 • Without a doubt, Dumezil asserts without arguing the case any further, these were *Tiwaz, *Thunraz and *Ingwaz242• He provides the arguments in 1941, referring to De Vries' interpretation of 1935243 • De Vries wanted to read the sons of Mannus as: Irmin, Istraz and Inguaz, i.e. Wodan, Tiwaz and FrO-Ingwaz. Ermin (Irmin) must have had a meaning that was generally understood. One of its occurrences, De Vries states, is in Odhinn's nickname iormunr, although he does not link Irmin and Odhinn (Wodan) together. Dumezil uses this passage: MJan de Vries a, en effet, montr~ que deux des noms des ~ponymes sup~ par les noms des descendants, *Inguwaz et *Ennenaz coincident l'un avec le second nom du scandinave Freyr (Yngvi), !'autre avec un des surnoms d'Odhinn (ltJrmunr), c'est-A-dire d~signent deux des membres de Ia triade fonctionnelle d'Upsal. 244
Dumezil notes that it cannot be demonstrated that the name *Istuwaz or *Istraz can be identified with ThOrr. However, in 1940 (and literally again in the second edition of 1948), he adduces the argument that adjectives with the sufftx *-raz = P.I.E. *-ro- are common in the "formation frequente dans les noms des dieux combattants et forts: Indra, Rudra, *Thunraz lui-meme"245 • In 1971
240 ibidem, p.283. 241 Gennania, 2.
242 243 244 245
Dumezil, 1939a, p.12. De Vries, 1935, p.212-216. Dumezil, 1941c, p.215. Dumezil, 1940, p.95 = 1948d; p.143.
TilE 1RIPARTITE SYSTEM
107
another linguistic argument is adduced for the identification of the ancestor of the lstaeuones with ThOrr . ...quant au radical de Istaeuones (Piine: Istri(an)ones), s'il ne se retrouve pas dans un surnom du dieu fort et combattant Th6rr, il dtrive d'une racine qui, en indo-iranienne comme en vieil-islandais, a produit des mots techniques de Ia fonction guerri~re: en sanscrit, isirli, , tpith~te des Marut......; en vieil-islandais enfin, le verbe eiskra, qui caracttrise l'ttat de fureur des berserkir, guerriers A pouvoir surnaturels.246 Other evidence and "bilan" We saw earlier (seep. 62) that in 1945 Dumezil gave a tripartite interpretation to the nature of the objects made by the Rbhus as evidence of the tripartite nature of the Indian pantheon. The functional nature of these objects was demonstrated by a Germanic paralleF47• Snorri relates that one day Loki cut off the golden hair of Sif, the wife of ThOrr. Thorr would have killed him if Loki had not managed to get the svartalfar to make two golden plaits which grew like real hair. The svartalfar also made the ship Skidhbladhnir for Freyr and the spear Gungnir for Odhinn. Loki then challenged two other dwarfs to make objects of the same kind, staking his life that they would not be able to do so. They made the ring Draupnir for Odhinn, the hammer Mjollnir for ThOrr and the wild boar Gullinborsti for Freyr. The gods judged that the second series was the best, and Loki was to pay with his life. However, he claimed that he had not risked his neck, so that the dwarfs must keep away from that. The winning party therefore sucked off Loki's lips. Dumezil follows De Vries248 in assuming that the second list is the oldest. This narrative therefore illustrates both the functional gods and the functional objects. Dumezil comments on this group of gods: ...en tous cas le caract~re fonctionnel de ces derniers ne fait pas de doute, attestt qu'il est par tout l'ensemble des mythes de chacun et par plusieurs traditions ou rituels ou its sont organiquement groupts.249 He also mentions beer libations at which ThOrr, Odhinn and Freyja (or Freyr and sometimes the Ases in general) were invoked250• Dumezil had already used Tacitus' account251 in 1940. Tacitus mentions Mercury, Mars and Hercules, i.e. Odhinn, Tyr and ThOrr. The Germanic peoples do not have a god for the third
246
Dumezil, 1971, p.256.
247 Snorri, Skaldskaparmal, 33. 248 De Vries, 1933, p.90. 249
Dumezil, 1945, p.SO.
250 ibidem, p.48. 251 Germania, ch.9.
108
CHAPTER TIIREE
function because of their scorn of agriculture252• In 1952 Dumezil adds to the group253 • Besides Mercury, Mars and Hercules Tacitus also mentions Isis 254, Nerthus255 and the twin Alcis256• Dumezil can therefore sum up: Mais ces petits ~carts locaux (sc. the militairisation, W.B.) ne d~figurent pas Ia structure tripartite de Ia mythologie. Claire d~jA dans La Germanie de Tacite, elle ne l'est pas moins en Scandinavie, ou Ia triade r~gne au temple de Vieii-Upsal comme dans des formules de toast sacrificiels ou de mal~diction, et aussi dans des mythes, tels que ceux des joyaux divins ou de Ia bataille eschatologique.257 In this form the corpus of evidence for the existence of a functional triad consisting of Odhinn, Thorr and Freyr is thus considerable. In 1958 Dumezil first rejects historical theories in which Odhinn, Tyr and ThOrr are treated as belonging to different chronological levels and then goes on to indicate the firm structure of the triad: "Nous nous en tiendrons aux faits. Et d'abord a !'existence meme de Ia triade comme telle" 258• He does not adduce any new arguments, but he is more explicit about his cryptic reference in 1952 to the eschatological struggle. He was referring to Voluspa 53-56, where Odhinn, Freyr and ThOrr each engage in their final combat in the last struggle. The 1939 study of Germanic mythology was published two years after the handbook by De Vries259• The full revision of the material in Les dieux des Gennains in 1959 came two years after the complete revision of the handbook260• The tripartite structure of the Germanic pantheon is now the main theme, so that Dumezil begins by summing up the arguments for a functional interpretation of the group comprising Odhinn, ThOrr and Freyr261 • The passage from Adam of Bremen and the curse of Egill crop up again. The latter is now supported by the strophe from the Skirnismal and by a curse of Hallfredhr V andhraedhaskald, who calls down on himself the curse of Freyr and Freyja, Grimnir ( = Odhinn) and Thorr in one of the strophes. Dumezil also refers to A. Bang's collection of formulae 262, in which the group comprising Odhinn, ThOrr and Frigga occurs twice. The group of gods occurs in the mythology in the tale of the objects made by the dwarfs and in the eschatological struggle. The dead are also divided between Odhinn, ThOrr and Freyja. In the Harbardhsljodh (str. 24) Odhinn receives the nobles and ThOrr the commoners. In the Grimnismal (str. 14) 252 253 254 255 256 257 258 259 260 261 262
Dumezil, 1940, p.95. Dumezil, 1952a, p.26. Germania, 9. ibidem, 40. ibidem, 43. Dumezil, 1952a, p.24. Dumezil, 1958a, p.54. De Vries, 1935 and 1937. De Vries, 1956 and 1957. Dumezil, 1959b, p.S-8. Bang, 1901.
TilE TRIPARTITE SYSTEM
109
Odhinn and Freyja divide those who have fallen in battle between them. In the Egilssaga (78, 19) the women go to Freyja. Dumezil demonstrates that this is not confmed to the North Germanic material by referring to the texts by Caesar and Tacitus referred to earlier. He also refers to the Vatican renunciation formula 263, in which Thunar, Uuoten and Saxnot are renounced: Thorr, Odhinn and a god of the Saxon mass. Summary: Odhinn, Thorr and Freyr Dumezil was already occupied with Gennanic mythology at an early stage. At first he used the tripartite model as a peg on which to hang the themes for discussion (1939). Use of the auxiliary ''glissement militaire" hypothesis enabled him to account for deviations from the Indo-European model. At first the shift only concerned unexpected aspects of the representatives of the first function, but later Thorr appeared to have been expelled from the second function as a result of this shift. In tum, ThOrr assumed certain aspects of the third function from Freyr ( 1958). At first Dumezil did not have much evidence to support the existence of a tripartite Gennanic pantheon. He based his case on the group of gods from Uppsala, the curse of Egill and the sons of Mannus. Dumezil tried to strengthen the tripartite interpretation of the last two items on a number of occasions. Later on Dumezil adduces additional evidence. The interpretation of the curse of Egill is elaborated in 1946, 1959 and 1985, and that of the sons of Mannus in 1971. In 1945 Dumezil indicates the tripartite nature of the three objects fashioned by the dwarfs. He also refers to libations in which the three functions can be distinguished. In 1952 the material is expanded with the addition of the deities mentioned by Tacitus. After a repetition of these arguments, Dumezil makes slight additions to them in 1959. He refers to the fonnulae compiled by A. Bang. Furthennore, in three different poems (the Harbardhsljodh, the Grimnismdl and a poem in the Egilssaga) the dead are divided between the gods of the three functions. Finally, he refers to the Vatican fonnula of renunciation. Social tripartition
Dumezil's statement of the problem of the lack of social tripartition among the Germanic peoples comes relatively late, in 1958. ...alors que les Celtes... presentent, dans !'organisation sociale une formule presque superposable a Ia structure indo-iranienne ... , mais, dans Ia theologie un tableau complexe ... , tout au contraire les Germains professent une theolo~ trifonctionnelle claire..., mais ne divisent pas leur societes selon les trois fonctions.
263 Dumezil, 1959a, p.27-29. 264 Dumezil, 1958d, p.l.
CHAPTER TIIREE
110
The interpretation of the Rigsthula puts an end to this shortcoming. In this Edda poem Heimdallr travels through the world incognito as Rigr. He fathers a son, Thraell (slave), by the great-grandparents Ai and Edda. Thraell's children all have pejorative names. Subsequently he stays with the grandparents Afi and Amma, where he fathers Karl (free farmer). Karl's children have names associated with the life of farming. Then he comes to the house of Fadhir and Modhir (father and mother), where he fathers Jarl (noble). Rigr stays there to bring up his son. Jarl's youngest son is called Konr. He becomes the ftrst king (the young Konr = Konr ungr = Konungr). The life of Jarl and his sons and father, as well as their military exercises, bear comparison with the life of the Indian kshatriyas. The description of Karl corresponds to the agrarian definition of the Indian vaishyas. The pejorative treatment of Thraell marks him as an outsider to the established order, like the position of the shudras vis-a-vis the Aryan order. There is no priestly class, but this is compensated by the magical character of Konr ungr's initiation in the runes. Thus the ftrst two terms in the Germanic groups of adalingus, liber, servus (Angles), edhilingi, frilingi, lazzi (Saxons), nobilis, ingenuus, servus (Danes) present the second and third members of the Indian social classes. Dumezil illustrates this with a table265 :
Figure 10: Tripartition in the Rigsthula
Jarl Karl
brahmanes kshatriya vai\;ya
(adalingus) (liber)
Thraell
\;Udra
(servus)
Celtic tripartition We fmd tripartite survivals in the Celtic t1aditions in the social relations and in the group of mythic heroes, the Tuatha De Danann.
Social tripartition In 1938 Dumezil put forward a model of interpretation in which the material on Celtic society could be accommodatedU6. Two groups were combined. Caesar
265 ibidem, p.3. U6
Dumezil, 1938a, p.191f.
111
TIIE TRIPARTITE SYSTEM
fmds two social classes in Gaul267, the druids and the knights, the representatives of the first and second functions respectively. As for the rest of the social organisation, Dumezil notes: "le reste de Ia societe, ruine et demoralise par les dettes, ne comportait plus d'organisation". 268 In the later medieval Irish texts the druids occur together with other professions, and no longer form a separate class. However, the texts reveal the existence of a military elite, the flaith, and a group of free cattle-farmers, the bo-airig, i.e. the airig (cf. Indo-Iranian arya-) of bo ( = cattle). If we combine the material from the island and mainland stemming from before the collapse of Celtic culture brought on by Romans and Christians, we obtain the following picture: Figure 11: Celtic social tripartition Caesar first function second function third function
druides equites
Medieval Irish texts flaith bo-airig
There is a slight alteration in the evidence adduced to support this interpretation in 1958. The association of airig with arya is now dropped and airig is interpreted to mean everyone "qui participe aux assemblees (airecht)"269• However, Dumezil indicates that this is a minor change: "Mais peu importe: le tableau triparti celtique recouvre exactement le tableau, reel ou ideal, des societes indo-iraniennes" 270• In the same year Dumezil also examines the absence of a first function class in Ireland in more detail271 • He claims that the "fonction artisane" has been responsible for a development among the Celts. Handicrafts are held in very high esteem. This led either to an extension of the first level- "< craftsmanship, druidry, jurisprudence> ,zn are associated with one another - or to a generalisation of the term "handicraft" to include the druids and the warriors. They are listed in the series: smith, doctor, bronze-worker, carpenter, top fighter, wizard, cupbearer, druid, exorcist, sorceress and Dagda himself. The series consisting of druids, flaith and bo-airig is thus not disrupted under the _influence of Christianity, but it is transformed in a typically Celtic way.
267 De Bello Gallico, VI, 13.
Dumezil, 1938a, p.l92. 269 Dumezil, 1958a, p.12. 268
270 ibidem. 271 Dumezil, 1958e. 2n ibidem, p.723.
112
CHAPTER THREE
The Tuatha De Danann
The Celtic pantheon does not display a clear-cut tripartition. There are hardly any direct sources on the gods in the tradition. For the mainland there are sporadic remarks in the classical authors. The Tuatha De Danann plays a major role in the historiography of Ireland. This mythical group inhabited Ireland before the sons of Mil, i.e. human beings, conquered the island. The Tuatha De Danann retreated to the barrows. It is generally assumed that the narratives of the Tuatha De Danann contain the ancient myths. Dumezil selects five figures from this group273: Lug Samildanach, "le dieu de taus les arts", Dagda, "le grand dieu et le druide par excellence", Ogma, "le premier champion", Dian Cecht, "le medicin divin" and Goibniu, "le forgeron". One of the results of his interpretation is that Caesar is a good source of information about the mainland. Caesar lists five gods and characterises them as follows 274: Mercury, the inventor of all the arts, Jupiter, the divine sovereign, Mars, the leader in war, Apollo, the god of healing, and Minerva, the goddess of handicraft. The Irish and Gallic pentads are equivalent, Dumezil claims. A Ia presence et a Ia place de l'lnventeur pr~s, et aussi avec une absence curieuse de I'Eieveur-Agriculteur, cette hierarchie n'est pas eloignee de ce qu'on entrevoit des anciens syst~mes sociaux chez les Indo-Iraniens....... chez les Irlandais et chez les Gaulois eux-memes. 275
It is anomalous that there is a social representative of the third function, the
bo-airig, but not one at the level of the pantheon. Dumezil resolves this anomaly in 1941276• The Tuatha De Danann fight two battles at Magh Tuiredh, the first against the previous inhabitants of Ireland, the second against the Fomoire. Dumezil compares the second of these battles with other conflicts in which the first and second functions join forces against the third. The Tuatha De Danann is thus a combination of the first two functions, while their opponents, the Fomoire, represent the third function (see p. 189f.). As we have seen, the Indian and Germanic functional groups were associated with a series of functional objects. The same goes for the Tuatha De Danannm. There was Dagda's magic cauldron with a never-ending supply of food, Lug's unconquerable spear, the king Nuada's sword that never missed its mark and the stone Fal (without a named owner) which uttered a cry whenever a legitimate king of Ireland stood upon it. The rest of the tripartite tradition can be described as marginal.
273 274 275 276
m
Dumezil, 1939a, p.9. ibidem, p.9-10. ibidem, p.lO. Dumezil, 194lc, p.167ff. ibidem, p.226-232.
1HE TRIPARTITE SYSTEM
113
Greek tripartition
Dumezil did not use much Greek material to support his theory. In 1940, in the context of a different investigation, he remarks: " ...en Grece, !'objet de notre presente investigation ne subsiste meme pas, puisqu'aucune forme de mythologie ni de societe grecque ne s'articule plus sur le schema indo-europeen des < trois fonctions (ou classes) sociales > •>27&. He returns to this verdict in 1941. Plato, Plutarch and Strabo distinguish a number of groups. Plato divides the ancient Athenians into the class of priests, the class of handworkers, herdsmen and farmers, and the class of warriors 279• Strabo280 and Plutarch281 refer to the Hoplitai, the representatives of the second function; the Ergadeis, the artisans; the Geleontes, the farmers; and the Aigikoreis, who are herdsmen in Plutarch and religious magistrates in Strabo. The Dorian tripartition into Hylleis, Pamphyloi and Dymanes is also given a functional interpretation282• The Hylleis, the descendants of Hyllos, represent the second function and are associated with Heracles, "en tant que heros, combattant, auxiliaire ....." 283• The Pamphyloi, whose name begins with the preftx pan-, which recalls vishva, vaishya, comprise the third class of the masses. The Dymanes, composed of the religious preftx du- and the sufftx -man/-maon (cf. brah-man), are the priestly class. These arguments lead to the conclusion: "tout cela est conforme a ce que nous attendons si ces < tribus > , comme les tribus romaines, derivent des classes indo-europeennes."284 These Ionian and Dorian tripartitions form the background to the tripartition in Plato's republic. Plato had been preceded by "les plus vieux penseurs indo-europeens qui avaient mis en parallele l'ordre triple du monde et de Ia societe"285 • The interpretation of the three Dorian tribes is abandoned in 1954. " ..elles soot probablement d'un autre type" 286• However, Dumezil adheres to his interpretation of the three Ionian tribes. They are divided "par leur role dans l'organisme social"287• Dumezil does not pay any further attention to the tripartition of Greek society or the Greek cosmos apart from the statement in 1985 that the tribes can be deftned in functional terms, "au moins theoriquement" 288• Summary: Celtic tripartition; Greek tripartition The Celtic and Greek tripartite traditions turned out to be marginal. Dumezil made no additions to the Celtic material after 1941. Part of the Greek material from 1941 is later abandoned. 278 279 280 281 282 283 284 285 286 287 288
Dumezil, 1940, p.93. Timaeus, 24a. VIII, 7, 1. Solon, XXIII, 4. Dumezil, 1941c, p.254.2S7. ibidem, p.2S5. ibidem, p.2S7. ibidem, p.260. Dumezil, 1954e, p.2S; cf. 1954a, p.60. Dumezil, 1958a, p.16. Dumezil, 1985, p.203.
CHAPIER FOUR
THE MITRA-VARUNA CONCEPT Chapters 4 and 5 deal with Dumezil's account of the bipartition of the first function. I shall begin with a discussion of the concept of bipartition itself, before moving on to some related themes, the "bipartita minora". A mythological theme which is related to it- that of the one-eyed or one-handed being - is fust dealt with. This is followed by a review of other concepts and earlier theories which are significant in this connection. Finally, we shall see that the application of the concept of bipartition was not confmed to the first function.
Introduction As we saw above (p. 9f.), the theory of Indo-European tripartition dates from 1938. At that time the theory was a first attempt to connect the Indo-Iranian social material with the Roman theological data. The brahmans formed the top rung in a social hierarchy. The flamines maiores are in the service of gods which can be connected with this Indo-Iranian hierarchy. As has been shown, the theory was steadily developed from 1938 on. A similar theoretical development applies in the case of the concept discussed in the present chapter. Dumezil uses it to explain the fact that the first function often has two occupants. This function often tends to attract pairs, such as Mitra and Varuna, Numa and Romulus, or Tyr and Odhinn. In a series of lectures dating from 1938 and 1939, which were published in extenso in 1940\ Dumezil supposed that these pairs had a complementary character. However, this was not the first time that Dumezil tackled this field of investigation. In 1929 he had already published the results of a study of the relations between Centaurs, Gandharvas and Luperci; in 1934 the results of a study of the correspondence between Ouranos and Varuna; in 1935 those of a study of the similarities between brahman and flamen; and in 1938 those of a study of the opposition between iunior and senior. These investigations, which precede the origin of the Mitra/Varuna concept and do not imply the theory of tripartition, all contributed to the formation of this concept. In 1940 all this preliminary research led to the (direct) crystallisation of a theory of bipartition within the first function. It is the development of this concept which we shall address first. Two questions led Dumezil to undertake his 1940 investigation:
1
Dumezil, 1940.
TilE MI1RA-VARUNA CONCEPT
115
L'importance du sujet nous ~tait apparue en 1934, au cours d'une conversation avec Sylvain Uvi. Ce grand esprit bienveillant avait accueilli notre Ouranos-Varuna, mais il soulevait une difficult~: Au d~but de 1938, discutant A Ia Soci~t~ Ernest Renan Ia communication oil nous avions confront~ Ia hi~rarchie romaine des trois flamines majeurs et Ia tripartition brahmanique de Ia soci~t~ ... MJean Bayet tirait de !'appellation meme du flamen dialis une difficult~ semblable: Le lecteur verra bient6t que ces deux questions sont toute Ia question. 2
It was in seeking answers to these questions that Dumezil came up with the opposition between Mitra and Varuna. It is difficult to indicate precisely what this opposition implies, Dumezil remarks in 1940. He compares it with that between yin and yang. De meme l'analogie du yang et yin nous d~charge de l'embarras de d~finir exactement, par sa matiere, notre couple indo-euro~en: il ~chappe A Ia d~finition parce que, lui aussi, il est essentiellement un mode de pe~e, un principe forme! de classification. Tout au plus peut-on l'echantilloner, dire par exemple que l'un des deux termes (Varuna, etc.) recouvre ce qui est inspir~, impr~visible, fr~n~tique, rapide, magique, terrible, sombre, exigeant, totalitaire, iunior, etc., tandis que !'autre (Mitra, etc.) recouvre ce qui est r~gl~, exact, majestueux, lent, juridique, bienveillant, clair, li~ral, distributif, senior, etc. Mais il est vain de pr~tendre partir d'un ~l~ment de ces pour en d~duire les autres.3 The problem of definition was to dig the investigation from the beginning to the end. In the "bilan" on the first function gods which was published in 19n, Dumezil claims that the distinction is an intuitive one rather than one which can be defined with precision. Du bilan qui pr~cede r~sulte une impression d'unit~, de coh~rence entre les formules si diverses qui pr~tendent mat~rialiser le contraste et Ia compl~mentarit~ des deux dieux. Ne craignons pas le mot : ~tant donn~ leurs caracteres, on qu'il n'est pas possible qu'ils ~changent leurs positions, sauf dans les quelques circonstances particulieres oil ils sont livr~s aux subtilit~s des casuistes et dont on a rencontr~ plus haut des exemples. 4 The resemblance to yang and yin also reveals what is special about the MN concept. The categories of yang and yin can be used to describe all fields. The MN opposition, on the other hand, can only serve as a description of the bipartition of the first function, at least in principle. This is also relevant to the demonstration of the genetic nature of the tripartition itself, since we encounter
2 3 4
ibidem, p.X,XI. ibidem, p.144-145. Dumezil, 1977, p.n.
CHAPTER FOUR
116
tripartite groups with a specific character. We shall now examine the forms which the MN concept has assumed in the various fields. Mitra- Varuna The first qualification of Mitra and Varuna in terms of the concept dates from 1939: ...a rote de Varuna et de sa magie, l'lnde vedique (et deja le monde indo-iranien) honorait Mitra et ses lois; les deux divinites complementaires etaient etroitement associees, et par consequent opposees (par exemple, disent les textes liturgiques, comme le jour et Ia nuit, comme Ia droite et Ia gauche....).5
In 1940 Dumezil goes into this contrast in more detail. He begins by citing the views of Bergaigne6, who had claimed that Mitra stood for the "ami" aspect and Varuna for the "terrible" aspect of the "dieux souverains"7• Dumezil stresses this aspect with reference to Varuna, referring to his study of Ouranos and Varuna from 1934 (see p.163f.). On Mitra he also cites the work of Meillet8, who saw the Indo-Iranian Mitra as the "contrat personifie"9• Meillet claimed that these two ways of looking at Mitra (as "ami" and as "contrat personifie") were irreconcilable, but Dumezil disagrees with him on this score. Levi had described the opposition between Mitra and Varuna as "!'intelligence - la volonte", "la decision -l'acte" and "Ia lune decroissante -Ia lune croissante" 10• He remarked: "L'ecart de ces interpretations en demontre la fantaisie" 11 • Dumezil disagrees. In his opinion, they reveal a typical difference. Les deux autres (sc. "l'intelligence - Ia volonte", "Ia decision - l'acte", W.B.) formules en rejoignent beaucoup d' autres et ce faisceau de definit excellemment deux manieres de diriger et de considerer le monde. 12
The material for this spectrum of ideas is derived from the Shatapatha Brahmana and the Taittiriya Brahmana. Dumezil also uses the qualifications mentioned here in 194813, 195214, 195815 and 195916• The same material recurs in 5 6 1 8
9 10 11 12 13 14 15 16
Dumezil, 1939a, p.39. Bergaigne, 1878-1883. Dumezil, 1940, p.45. Meillet, 1907. Dumezil, 1940, p.45. Uvi, 1898. Dumezil, 1940, p.48. ibidem, p.48-49. Dumezil, 1948b, p.448. Dumezil, 1952a, p.40ff. Dumezil, 1958a, p.62ff. Dumezil, 1959b, p.59~1.
TilE MI1RA-VARUNA CONCEPT
117
196817 and in 1977 18, but now with references to the texts. The Mitra-Varuna opposition can be found in ShB, IV,1,2-3 as brahman-kshatriya; in IV,1,4,8 as milk-soma. IV,5,1,6 contains the remark that the correct offering is for Mitra and the incorrect offering for Varuna; in V,3,2,5, Mitra is the recipient of what falls by itself and Varuna of what is chopped off with the axe; in V,3,2,6 Mitra receives the natural butter, while Varuna receives the beaten butter; in V,3,2,8 Mitra receives what is steamed and Varuna what is roasted above the ftre; in TaitB 1,7,10 1 the day is Mitra's and the night is Varuna's, and they are opposed to one another as right (in both senses of the word) to left. Dumezil's conclusion is: ...toutes ces expressions doubles d~finissent des points homologues sur deux plans....: Mitra est le souverain sous son aspect raisonnant, clair, r~gl~, calme, bienveillant, sacerdotal; Varuna est le souverain sous son aspect assaillant, sombre, inspir~, violent, terrible, guerrier. 19 Dumezil also records the remark that Mitra-Varuna are opposed to one another as "ce monde-ci" - "!'autre monde" and as white-black, but the latter opposition is particularly important for the opposition between Jupiter and Dius Fidius (see p. 135f.) 20• The opposition also helps to explain Mitra's aversion to the killing of Vrtra, although it is necessary for the soma offering. In the succeeding period Dumezil repeats this opposition in brief summaries. " ...un couple divin: celui du magicien Varona et du juriste Mitra'm. Or in the similar phrasing of 1940: Varuna est le souverain magicien, cr~ateur, terrible, violent, prestigieux, sombre, agressif; Mitra est le souverain juriste, conservateur, abordable, r~gl~, majestueux, lumineux, sacerdota1. 22 Dumezil provides a modified version of the opposition in 1945 in dealing with the distinction between the Amesha Spentas Vohu Manah and Asha (see below, p. 126ff.). He retains as the most typical of the oppositions listed in 1940 that from the ShB, XII,9,2,12, in which Mitra is opposed to Varuna as "ce monde-ci" to "!'autre monde". The extended series, marked by the large number of epithets, is now a specifically Indian way of representing the opposition.
17 18 19
20 21
22
Dumezil, 1968, p.147ff. Dumezil, 1977, p55ff. Dumezil, 1940, p.49. ibidem, p53. Dumezil, 1941c, p.69. Dumezil, 1942, p.73.
118
CHAP1ER FOUR
...peut~tre n'avons-nous pas pris garde que !'opposition ainsi d~finie n'~tait qu'un cas particulier, plutOt meme une con~uence tr~ fr~uente et tr~ d~elop~ d'une opposition plus g~n~rale et plus profonde. 23 Once again Dumezil cites Bergaigne, who had broken the opposition down into three aspects24: Mitra as "forme terrestre" of fire and soma, V aruna as "forme celeste"; Mitra as "l'Agni ou le Soma invisible" and Varuna as "l'Agni ou le Soma manifeste sous forme d'eclair ou de soleil"; Mitra as the "pretre qui honore" and Varuna as the "dieu qui est honore". These oppositions yield the following definition: ...tout concept et tout acte religieux peuvent etre compris de deux mani~res, ou en deux parties, dont l'une est plus ~loign~e et !'autre plus rapprocMe de l'homme; r~guli~rement Ia Eremi~re sera sous le patronage de Varona, Ia seconde sous le patronage de Mitra. scind~
The conclusion in the second edition of Mitra-Varuna is slightly different. In 1940 one of the points on which the Yang-Yin opposition was not comparable with the Mitra-Varuna opposition was that Yang and Yin were opposed to one another as male and female, while Mitra and Varuna were not. Still, Dumezil points out that there is an opposition between a "Moi, spectateur qu'il appelle Purusha, "le principe male"" and a ""nature", actrice multiforme, la prakriti" in the Sankhya, one of the philosophical systems26• Purusha and Prakriti are compared with Mitra and V aruna in the Mahabharata, so that Mitra is equated with the male principle and V aruna with the female. If this is compared with the distinction between Ouranos and Gaia, however, in which Ouranos is the male, heavenly figure who represents "!'autre monde", while Gaia is the female, earthly figure who represents "ce monde-ci", one might expect the opposite to be the case. Dumezil therefore remarks: ...mais
peut~tre !'auteur de ces assimilations signifiait-il cr~ateur, etc., plutOt que le caract~re ici
actif, calme et entit~s. 27
par I~ les aspects passif et male, I~ femelle des deux
Though this connection took Dumezil by surprise in 1940, by 1948 he regards it as a logical consequence of material from the Shatapatha Brahmana28• Now he cites ShB IV,4,19: "Mitra ejacule sa semence dans Varuna"29• The opposition
23 24 25 26 27 28 29
Dumezil, 1945, p.106. ibidem, p.107-109. ibidem, p.llO. Dumezil, 1940, p.145-146. ibidem, p.146. Dumezil, 1948) comme le brahman et le ksathra (nous dirions A peu pr~s: et le ).31 Dumezil sees this as providing the historical perspective for the later Indian philosophical reflections on Purusha and Prakriti. In 1948 and 1949 both types of qualification occur side by side: both in the series of epithets and in the short description contained in the 1945 definition. In 1952 Dumezil provides a different account, which he continues to use up to
30 31
ibidem. ibidem, p.208-209.
120
CHAPTER FOUR
Figure 13: Mitra- Varuna (1958 and 1968).
1. Quant a leurs domains
Varuna l'immense ensemble (del (del
l'autre monde la nuit
2. Quant aux modes d'action 3. Quant aux caracteres
(soma/feu invisible grand sorcier, maya violent, soudain (impitoyable (inquietant noeuds, saisit avec violence (ShB V,4,5,12 bien sacrifie (ShB IV,5,1,6; coupe a Ia hache roti au feu soma eniverant la volonte (Ia lune croissante (le feu qui deja flambe
4. Parmi les fonction autres que la leur propre
(affinite avec lndra, GopB II,1,22 1968) violence conquerante kshatra
Mitra proche de l'homme terre, RV,IV,3,51968) emplacements humains du sacrifice, RV,IX,77 5 1968) ce monde-ci (ShB, XII,9,2,12) le jour (Sayana at RV 1,141,9; TaitB 1,7,10,1; TaitS VI,4,8) soma/feu visible 1958) "contrat" bienveillant, rassurant, progressif, doux1958) "amical", RV,II1,59: confiance 1968) ne fait mal a personne ShB V,3,2,7) mal sacrifie TaitB 1,6,5,5) casse de soi meme (ShB V,3,2,5) cuit a la vapeur (ShB V,3,2,8) le lait (ShB IV,1,4,8) !'intelligence la lune decroissante ShB 11,4,4,18 1968) le feu en train de s'eteindre ShB II,3,2,10 and 121968)
prosperite pastorale et Ia paix brahman (ShB IV,1,4,2,3, 11,5,2,34 etc.)
1HE MITRA-VARUNA CONCEPT
121
and including the "bilan" of 1977. The qualification of Mitra and Varuna is now divided up into a number of categories, although there is some room for manoeuvre in the number of categories and in the distribution of qualities within these categories. In the 1952 scheme Varuna and Mitra are categorised by means of a number of key terms32 (see figure 12). In 1957 Dumezil reiterates this scheme without any changes33, but the 1958 version is modified34• He repeats this modified version in 1968, this time referring to the texts on which it is based35 • The texts referred to in the scheme below date from 1968. The differences between the 1958 and 1968 versions are indicated in parentheses, see figure 13. In 1976 Dumezil presents a summary of the contents of the "bilan" which was published in the following year. He considers this "bilan" as a personal rounding off of his research: "to present them in a form, which, though by no means definitive, will for me be final". 36 This summary contains two corrections to the 1952list. On the first point, under Mitra, soma must be replaced by "lait". This change should probably also be effected in the 1958 list and the brief presentation in 195937 • The other correction is the deletion ofthe statement that Varuna and Mitra can be seen as the deity and his priest (under the fourth point of the 1952list). The 1977list (figure 14) is a variant of the preceding one38• The lists dating from 1952, 1958 ( = 1968) and 1977 contain a number of variants. In the 1958 list "Ia nuit" and "le jour" belong to the heading "Quant a leurs domaines", and in the 1977 list the same terms belong to the heading "les affinites cosmiques", while in the 1952 list "Ia nuit" and "le jour" are grouped under the heading "Quant a l'humeur". The title of the category "Quant aux moyens d'action" is used in 1952 and 1977, while the 1952 title for this category is "Quant aux modes d'action". The third category from 1952, "Quant a l'humeur", is the same as the category "Quant aux caracteres" from 1958 (with the exception of the pair "la nuit" and "le jour"). On the other hand, the items from this category are distributed between the categories "les caracteres" and "les modes d'action" in 1977. The latter category is thus not identical to the category with the same name in the 1958 list. Categories 4 and 5 from 1952 are merged to form a single category in 1958 (category 4) and in 1977 (category e).
Summary:Mitra - Vanma Dumezil claimed that the first function was divided into two complementary parts. The account of this bipartition was largely based on the opposition between Mitra Dumezil, 1952a, p.42. Dumezil, 1957a, p.ll-12. Dumezil, 1958a, p.63-64. 35 Dumezil, 1968, p.147-149. 36 Dumezil, 1976, p.26. 37 Dumezil, 1959b, p.59-61. 38 Dumezil, 1977, p.61-n. 32 33 34
CHAPTER FOUR
122
Figure 14: Mitra- Varuna (1977). Varuna
Mitra
b) les moyens d'action
justicier, rigoureux redoutable, RV,I,25 maya,liens,noeuds
bienveillant, amical, aimable, RV,III,59 "contrat", yatayaijana "(se) mettre a l'endroit appropie, occuper son emplacement naturel", "par rapport aux autres"
c) les modes d'action 1. en matiere concrete
action violente, soudain coupe a la hache
douce, dans les voies de la nature casse de soi meme creme sur le lait cuit a la vapeur plantes, sur un terrain laboure accueillante
Distinction des deux dieux a) les caracteres
ShB V,3,2,6: ShB V,3,3,8: 2. en matiere morale ou rituelle 3. en matiere psychologique
4. en vertu de sa nature d) les affinites cosmiques e) les affinites sociales et theologiques
1. 2. 3. 1. 2. 3.
beurre barate roti au feu plantes, sur une terre non travaillee punitive action qui fait
= kartar
force agissante = daksa Viraj (etrange, puissante epouse de Varuna) I'autre monde la nuit, noir feu: menace kshatra, roi terrestre avec Indra, deux. fonct. roi du "peuple des Gadharva" Y ama,les morts l'eau mobile et mysterieux
reflexion quicon~it
=
abhigantar force deliberante = kratu sraddha (confiance tranquille) cemonde-ci le jour, blanc feu: apaisement brahman et son chapelain RV,VII,82 avec Vasu, trois. fonct.
les vivants la terre ferme
TIIE MITRA-VARUNA CONCEPT
123
and Varona. Dumezil initially tried to define the gods by means of a series of epithets. Later on, in 1945, the opposition was reduced to the most basic features of the components. From 1952 on he presents extensive lists, whose subdivisions vary over the years. We can state that the opposition which was intuitively grasped in 1938/1939 did not undergo any far-reaching changes. This opposition is so flexible that it can sustain widely differing presentations. Appreciable shifts took place in these presentations. Between 1940 and 1977 Dumezil also gradually presented the textual material on which he based the opposition, which was only intuitively posited at first. This material was already the implicit basis for the treatment of the opposition. Mithra - Ahura Mazda
We have seen which changes, according to Dumezil, were brought about in the original Iranian pantheon as a result of the reforms of Zoroaster. The gods Mithra and Ahura Mazda returned in the post-Zoroastrian period, in opposition to the demons Indara and Naonhaithyas. The gods of the first function thus retained their divine status, while the gods of the second and third functions were assigned the status of demons. The question then arises of whether these later figures of Mithra and Ahura Mazda can be described in terms of the MN concept. In 1940 Dumezil treats Ahura Mazda and Varona in the same way when he remarks on the Mithra-Ahura formula in the Avesta: "elle associe Mithra a un Ahura qui n'est pas encore l'Ahura Mazdah de l'histoire, mais qui rejoint l'Asura type des hymnes vediques, Varuna" 39• Dumezil therefore considers doubts as to the possible identification of Varona and Ahura Mazda as without foundation. Ahura-Mazdah est-ill'h~ritier de cet , et par con~quent l'homologue du grand Asura v~dique Varuna? Cette hypothtse, longtemps admise sans discussion, a ~t~ par Ia suite fort discut~e,l) tort, croyons-nous. 40
Nevertheless, in 1953 Dumezil adopts a slightly different view of this identification. The reforms have retained Mithra and Airyaman from the Indo-Iranian pantheon at the level of the first function, but nothing of the "moitie Varona" has been retained. All the same, Dumezil claims that Zoroaster has borrowed features from Varona and applied them to Ahura Mazda41 • But in 1958 Dumezil describes Ahura Mazda again without making any reference to these problems: "sans doute lui-meme sublimation de I'Asura majeur, de celui que l'Inde appelle Varuna"42• In 1977 he offers another Dumezil, 1940, p.44. ibidem, p.67. Dumezil, 1953c, p5ff. 42 Dumezil, 1958a, p.40.
39 40 41
124
CHAPTER FOUR
possibility besides this one. In the Rig Veda an unspecified Asura is sometimes mentioned as holding a position above the gods, not as one of them but as the only God. It could be a further development of this tendency toward henotheism which resulted in the position of Ahura Mazda in Iran. All the same, Dumezil prefers the identification of Ahura Mazda with Varuna to this alternative43 • Besides the question of whether Ahura Mazda may be identified as a continuation of an Indo-Iranian Varuna, there is also the question of whether he can be accommodated within the MN concept. In 1940 Dumezil investigates the opposition between Ahura Mazda and the return of Mithra as yazata44• There were festivals in honour both of Ahura Mazda and of Mithra in Persia. The New Year Festival, which celebrates the creation, is in honour of Ahura Mazda, while one of the aspects of Mithra's festival is its eschatological import. The festivals thus oppose "Ia puissance creatrice" to "Ia perfection immobilisee", so that Ahura Mazda and Varuna are not only related genetically, but typologically too. Ces syst~mes de representation antithetiques, attaches par une tradition vivace aux deux fetes complementaires d'Ahura-Mazdah et de Mithra- aux deux equinoxesnous paraissent confirmer que, avant Ia reforme, le couple Mithra-Ahura avait meme signification, meme double orientation, meme equilibre, que le couple vedique Mitra-Varuna...45
The correspondence between Mitra and Mithra is thus postulated here, and they are both situated within the MN concept. However, a complication arises at this point. Mitra, who is associated with the third function and strives for peace, appears to have a warlike Iranian counterpart. Mithra is an aggressive god, and the figure of him wielding his club, the "vazra", is closely related to Verethraghna and recalls Indra, wielding his "vajra", who is closely associated or identified with Vrtrahan. The latter identification is an Indian development, according to Dumezil46• The warlike figure of Indra-Vrtrahan pushed other gods into the background in the divided kingdoms of India. In Iran the most important elements of the religion are divided among Ahura Mazda, his Amesha Spentas and Mithra. The latter has not entirely absorbed Verethraghna, but he has certainly taken over many of his features. It is quite likely, Dumezil considers, that it is not Mithra who has made gains, but Mitra who has sustained losses. Within this perspective, the Indo-Iranian Mitra must have had a broader definition. Likewise in Rome lightning is in the hands of Jupiter and Dius Fidius, not in those of Mars. Rome and Iran both display the same connection between the weapon and the first function. In India, on the
43 44
45 46
Dumezil, 1977, p.115-117. Dumezil, 1940, p.66-69. ibidem, p.69. ibidem, p.91.
TilE MITRA-VARUNA CONCEPT
125
other hand, we fmd the same image as in the Germanic world, where the god who wields the hammer of thunder is the god of the second function, ThOrr. While Dumezil's 1940 thesis was no more than the claim that Mithra had assumed certain characteristics of Indra-Vrtrahan, in 1953 he states: Mais Ie veritable substitut d'lndra, le vrai patron de Ia fonction guerritre, est autre (sc. not Verethraghna, W.B.): c'est Mithra qui manie Ie vazra, comme l'lndra vedique manie Ia vajra "Ia foudre", et Verethraghna n'est plus qu'un de ses officiersY As we saw earlier, Dumezil set the demonisation of the gods of the second and third functions in an ideological framework. In 1953 he situates the assumption of the second function by Mithra within the same framework. L'intention de Ia reforme apparait ici clairement: il n'y a de guerres acceptables qu'au nom, selon Ies principes et dans !'interet de Ia societe fidtle, de I'Eglise; Ia guerre et Ies guerriers ne se justifient que comme croisade et croises. 48 Dumezil makes similar claims in 1959. At the same time he retracts the 1940 proposal to contrast Iran and Rome with India and the Germanic world. Mithra has simply borrowed the warlike aspect from Indra49• In 19n Dumezil provides a different motive for the demonisation of Indra and the assumption of his function by Mithra . .. .tout se passe comme si Ies thCologiens reformateurs, appartenant eux-memes ~ Ia fonction, avaient juge plus sur d'annexer Ia force guerri~re au domaine propre de Ia souverainite religieuse. 50
premi~re
Moreover, Mithra has also assumed a number of characteristics of the Indo-Iranian Varuna, such as spying on humans. Dumezil sums up: "C'est dire que le Mitra et le Varuna vediques, en meme temps qu'un Indra corrige, rencontrent dans l'iranien Mithra des parties qui leur correspondent"51 •
Summary:Mithra - Ahura Mazda According to Dumezi/ (1940), the gods Mithra and Ahura Mazda assume the places of Mitra and Varona in the post-Zoroastrian period. The identification of
47 Dumezil, 1953c, 48 ibidem, p.8. 49
p.7.
Dumezil, 1959b, p.66-67. p.119.
50 Dumezil, 1977, 51 ibidem, p.l19.
126
CHAPTER FOUR
Ahura Mazda with Va!Una is uncritically accepted in 1940 and 1958, but it is treated more circumspectly in 1953. In 1977 Dumezil suggests an alternative: the identification of Ahura Mazda with an unspecified God mentioned in the Rig Veda who occupied a henotheistic position. The identification of Mitra and Mithra IUns up against the problem that Mitra is a god ofpeace, which fits in with the conceptual framework, while Mithra is a god of war, which does not fit in with it. Dumezil does not explain this aggressive character in 1953 by assuming an Iranian "glissement militaire", but by resorting to the effects of the refonns of Zoroaster, resulting in a demonisation of the second and third functions. In 1940 he suggested a contrast between Iran and Rome, on the one hand, and India and the Gennanic world, on the other. This contrast hinged on the question of whether it was the representative of the first or of the second function who was anned. In 1959 he retracted it again. In 1977 Dumezil assumes con-espondences between Mithra, on the one hand, and Mitra, Va!Una and a modified version of Indra, on the other. The Iranian case thus puts heavy pressure on the M!V concept. It would be unrecognisable as such without the auxiliary hypothesis of "the refonns of Zoroaster". Vohu Manah -Asha
We have already seen (above, p. 77ff.) that the tripartite group of gods did not occur alone in a later version. Their places were taken by the Amesha Spentas in the Gathic period. The first two in this group are Vohu Manah and Asha, who take the place of Mitra and Varuna respectively. Varuna is closely connected with order, Rta. In Iran this order is "Asha". As far as the interpretation of Vohu Manah is concerned, Dumezil initially follows Christensen's interpretation52, in which there was a correspondence between the Gathic Vohu Manah and the Avestan Mithra. Both are responsible for the herd and both play a part in the last judgement. Christensen saw the connection between Ahura Mazda and Vohu Manah as a continuation of the connection between Varuna and Mitra, and thus detached Vohu Manah from Asha. Dumezil objects to this. The two names occur together on a large number of occasions: 106 times, with or without the other Amesha Spentas. Apart from the 11 occasions on which Vohu Manah acts alone and the 30 occasions on which Asha does so, they occur by themselves 19 times in connection with one of the other Amesha Spentas. "Au total done, dans les stances oil ils figurent, Vohu Manah et Asha sont reunis 106 fois et separes 19: ce rapport de 5 a 1, est demonstratif'' 53• However, the statistical argument is a formal one. "11 faut en outre verifier si leur association ne recouvre pas une dynamique interne, une tension conceptuelle analogue a celle qui anime le couple des dieux vediques"54• We have seen how the MN concept was reduced to a compact 52 Christensen, 1926. 53 Dumezil, 1945, p.104. 54 ibidem, p.105.
TilE MITRA-VARUNA CONCEPT
127
formula in 1945 (see p.l17). This was centred on the opposition between "ce monde-ci" and "l'autre monde", and the Vedic opposition between Mitra and Varuna was seen as an extension of it. The series of epithets which was used to defme the MN concept before 1945 is not applicable to Vohu Manah and Asha. The new defmition, however, offers better prospects. Two texts are important in this connection: Yasna 29 and Yasna 44. In Yasna 44 Ahura Mazda is questioned on a number of matters connected with the world and with religion. The relevant strophes are 3 and 4: 3. Void ce que jete demande, dis-le moi en verite, oAhura: Qui done a ete par generation le pere premier d'Asha? Qui done a cree le chemin du soleil et des etoiles? Qui est celui par qui Ia lune croit, ensuite decroit? Ces choses-la, o Mazdiih, je desire les savoir, et d'autres. 4. Void ce que jete demande, dis-le moi en verite, oAhura: Qui done a flxe Ia terre en dessous et Ia nuee pour qu'elles ne tombent pas? Qui (a cree) les eaux et les plantes? Qui done a lie vitesse au vent et aux nuages? Qui done, oMazdiih, (a ete) le createur de Vohu Manah? 55 Strophe 3 is set in a distant scenario near the sun, the stars and the moon. Strophe 4, on the other hand, is set in a nearby scenario with the earth, waters, plants and the atmosphere with wind and clouds. Asha is mentioned in strophe 3, Vohu Manah in strophe 4. La conclusion s'impose: le p est objectif, absolu, independant de l'homme, en regard des relatifs et humains eunoia (Vohu Manah), .. "68 •
Summary: Vohu Manah -Asha In 1945 Dumezil situates the first two Amesha Spentas within the M!V concept too. He makes hardly any changes later on to the argumen~ apart from a few 63 64 65 66 67 68
Nyberg, 1938. Dumezil, 1945, p.133. ibidem. Dumezil, 19TI, p.136. Dumezil, 1945, p.131. Dumezil, 1977, p.138.
CHAPTER FOUR
130
additional points in 1977. The argument is based on two key texts: Yasna 29 and 44. Dumezil reduced the opposition within the M/V concept to what he saw as a basic opposition between "/'autre monde" and "ce monde-ci" in order to perceive a bipartition within these texts. Romulus- Numa So far we have been considering the binary opposition within the first function at the level of the pantheon. The focus of the Roman application of the MN is situated elsewhere: it is not the Jupiter-Dius Fidius opposition (see p. 135ff.) which is central, but that between Romulus and Numa. Mais c'est surtout dans leur qu'ils Ia d6veloppent, et longuement, analytiquement, intelligemment: en effet les r~gnes des deux premiers rois, les personnages de Romulus et de Numa, leurs caract~res, tout le d6tail de leurs gestes et de leur institutions s'opposent avec une rigueur qui ne peut etre fortuite, et dans le sens oil s'opposent Varuna en Mitra. 69 Numa is the successor to Romulus. This involves them in a chronological sequence, with consequences for the way in which they are described. Mitra and Varuna could be described and opposed to one another synchronically in the Brahmanic literature, making the opposition a clear-cut one. Romulus and Numa, however, are hardly described in this way, if at all. It is not possible to draw up a list of typical characteristics linked to the concept until both Romulus and Numa have been studied in their entirety. Hence Dumezil begins his analysis of Romulus and Numa as follows: Or il n'est pas contestable que les vies, et non seulement les oeuvres mais les figures meme de Numa en de Romulus, par-dessus des Msitations de d6tail qu'on peut et qu'on doit relever, ant et6 con~ues dans toute Ia tradition comme rigoureusement antith6tiques, et que cette antith~se recouvre, par beaucoup des points ou elle s'exprime, l'antith~se rituelle et conceptuelle analys6e au chapitre pr6redent.70 This is followed by a study of the differences between Romulus and Numa, in which the "Lives" written by Plutarch play a prominent part. Dumezil's use of the sources is as follows. For Numa he cites 24 fragments from Plutarch, 11 from Livy, 2 from Dionysius of Halicarnassus, 1 from Gellius, 1 from Ovid (both of whom repeat the material contained in Plutarch) and 1 from Virgil. For Romulus he cites 16 fragments from Plutarch, 5 from Livy and 1 from Ovid
69
70
Dumezil, 1942, p.73. Dumezil, 1940, p.29.
THE MITRA-VARUNA CONCEPT
131
(both of whom repeat the material contained in Plutarch). The material is organised in terms of a number of characteristic oppositions71 • Romulus is bent on power
Numa tries to keep the peace
Romulus is amoral and passionate; he kills his brother, associates with all kinds of riff-raff and rapes the Sabine
Numa is dispassionate, lacking in ambition and devoted to right
virgins
Romulus is a religious fraud; he devises Consus and the Consualia as a cover for the rape of the virgins
Numa is punctilious in religious and cult affairs
Romulus is without ties
Numa is the founder of a conventional family
Romulus belongs to "I' autre monde"
Numa belongs to "ce monde-ci"
Romulus founds a cult of Jupiter Feretrius and Jupiter Stator.
Numa founds a cult of Fides.
All these items display the opposition: "Ia tumulte, passion, imperialisme d'un iunior dechaine; ici, serenite, exactitude, moderation d'un senior sacerdotal"72• Or as Dumezil puts it in his 1942 summary: Plus efficacement que Jupiter et Fides, plus concr~tement du mains, ces figures humaines illustraient et justifiaient aux yeux de tout le peuple, de gl!nl!ration en gl!nl!ration, les deux aspects contraires et pourtant ~galement bons et necessaires de !'administration de l'Etat et du monde: cr~ation et organisation, enthousiasme et droit, violence et exactitude, celeritas et gravitas ... 73
Numa and Romulus form the most pregnant example of the bipartition of the ftrst function in Indo-European mythology and philosophy next to Mitra and Varuna. We saw that Dumezil tended more and more to interpret the ftrst kings of Rome in line with the Jupiter Mars Quirinus model over the years (see above, p. 100f.}. This tendency was less pronounced at ftrst. In 1940 Romulus and Tullus Hostilius were representatives of one aspect of the ftrst function, while Numa and Ancius Marcius represented its other aspect. As Purftravas founded the dynasty of the moon and Manu that of the sun, so Romulus and Numa each head two lists of mutually opposed kings. Dumezil refers to this binary model in 1943, despite the change in the functional interpretation of
ibidem, p.29-33. 72 ibidem, p.32.
71
73
Dumezil, 1942, p.73-74.
132
CHAPTER FOUR
Tullus as representative of the second function in 1942. In 1945 the abbreviated version of the opposition as that between "I'autre monde" and "ce monde-ci" is also considered to be applicable to Romulus and Numa. Dumezil bases his argument on this score on Plutarch7\ who puts the following words into Numa's mouth: On attribue a Romulus Ia gloire d'etre ne d'un dieu; on ne cesse de dire qu'il a ete sauve et nourri dans son enfance par une protection particuliere de Ia divinite; moi au contraire, je suis d'une race martelle, j'ai ete nourri et eleve par des hommes que . vous connatssez ...75
Dumezil concludes: "Plutot que de a , !'opposition est ici d' < humain > a < surhumain >, de < terrestre > a < supraterrestre > "76• This also throws light on a passage from Cicero77, which Dumezil had not referred to before ("nous n'avions su qu'en faire" 78). Piganiol had brought it to his attention. In this text, Romulus is the initiator of the auspicia and Numa of the sacra. The auspicia consist of receiving and interpreting signs sent below from above, while the sacra are the cults which proceed in the reverse direction. In 1947 Dumezil adds a passage from VirgiF9 to the material. The ftrst kings pass under review in the vision of Anchises. The opposition which has now become so significant between "auspicia" and "sacra" is applied to Romulus and Numa80• Deux substantifs dominent les deux premieres formules: auspicia pour Romulus, sacra pour Numa. Ce sont presque des termes techniques. Ciceron, lui aussi, dans le De natura deorum, III, 2, designe Romulus et Numa comme les fondateurs l'un de Ia science des auspicia, !'autre de Ia science des sacra, et l'on sait toute !'immense substructure de cette distinction. 81
As we saw earlier, the Jupiter Mars Quirinus concept was applied to the fullest
extent to the Roman kings in 1947. On this occasion Dumezil's opponents include E. Pais82• Pais had detected a number of groups in the material whose terms coincided and within which variation was possible, such as the group consisting of Romulus, Tullus Hostilius and Servius, or the group consisting of Titus Tatius, Numa and Ancus Marcius. Pais also saw the two Tarquinii as a duplication. Similarly, all the women could be reduced to a single concept. Dumezil objects: 74 75 76 77 78
Numa,5. Dumezil, 1945, p.107. ibidem. De natura deorum, 111,2. Dumezil, 1945, p.llO. 79 Aeneis, VI, m-784, 808-823. 80 Dumezil, 1947a, p.161ff. 81 ibidem, p.163-164. 82 Pais, 1915.
TilE MITRA-VARUNA CONCEPT
133
L'histoire royale ne se r~duit pas A deux ou trois types et il n'y a pas de double emploi. Les d~finitions diff~rentielles des rois en apparence les plus analogues sont ai~es A obtenir et leur juxtaposition se justifie, apparait meme en un sens comme n~ssaire. 83
Dumezil thinks that the origin of this confusion can be explained. In 1945 he had already stated that there was an affinity between the first and the third functions within the Mitra side, on the one hand, and between the first and second functions within the Varuna side, on the otherM. This was necessary in order to grasp the connection between Vohu Manah, Mithra and the cow. Among the Germanic peoples, as we saw, Odhinn was closely associated with war. Now Dumezil uses this affinity to explain that the aggressive aspect of Romulus is also an aspect of Tullus Hostilius, and that the resemblance between Numa and Ancus Marcius is an inevitable consequence of their functionality. After adding yet another text85 to the material in 1950, Dumezil defends his point of view: Ce n'est pas moi, c'est Tite-Live, c'est Denys d' Halicarnasse, c'est Plutarche, c'est toute Ia tradition qui s'ing~nie A opposer point par JOint, sur taus les points imaginables, Romulus et Numa.
In 1958 he organises the material in an even more systematic way. This time the characteristics are divided into four categories. They are the same as the ones which he used for the Indian material. He describes the exhaustive nature of the opposition in his introduction. Dans Ia tradition annalistique, les deux fondateurs de Rome, Romulus et Numa, forment une antithese aussi r~guliere, aussi d~velop¢e, et de meme sens, que celle de Varuna et de Mitra dans Ia litt~rature vMique: tout s'oppose dans leur caractere, dans leur fondations, dans leur histoire, mais d'une opposition sans hostilit~, Numa compl~tant !'oeuvre de Romulus, donnant A l'id~ologie royale de Rome son second ~le, aussi n~cessaire que le premier.87 This is followed by the four areas in which this finds expression. As far as the oppositions are concerned, these are88: -domaine - the main points here are the "auspicia" versus "sacra" oppositions and the passage from Plutarch89•
83
84 85 86
87 88 89
Dumezil, 1947a, p.194. See too: Dumezil, 1941c, p.178ff., where Dumezil mentions Tatius' affinity with Numa. F1orus, 1,8. Dumezil, 1950b, p.459. Dumezil, 1958a, p.S0-81. ibidem, p.Sl-82. Numa, 5, 4-5.
134
CHAPTER FOUR
-modes d'action - Dumezil refers to the foundation of the cult of Jupiter Feretrius and Jupiter Stator by Romulus and of the cult of Fides by Numa (see p.131). -caracteres - Romulus is the wild leader of the Celeres, while Numa is the dispassionate figure who dissolves this company and he is the founder/organiser of the flaminate. -affinites - Romulus has affinities with the second function and Numa with the third. The annalists may have added to the material, Mais il est clair que ces innovations memes se sont conformtes A une donnte traditionelle, dont !'intention ttait d'illustrer deux types de roi, deux modes de souverainit~: ceux memes que l'lnde couvre des noms de Varuna et de Mitra. 90 The organisation of the material in the "bilan" of 1977 is the same as that which he used for Mitra and Varuna. Dumezil rearranges the material without adding any new material91 • Caracteres covers the regni cupido of Romulus, his irreligious attitude during the rape of the Sabine virgins, his conquests and his pride in them, and his tyrannical features which caused the senators to view him with distrust. Numa is a dispassionate man with a conventional married life. He refused the crown at first. After his death the senators bore him to his grave. Under les moyens d'action we fmd Romulus' connections with the Celeres and the lightning attacks. Numa disbands the Celeres. This category also includes the subdivision between auspicia and sacra. Under les modes d'action we fmd two ways of resolving disputes: weapons are Romulus' answer, bargaining is Numa's method. Numa also appoints the Fetiales to regulate the violence of war. In the juridical area it is Romulus who carries out punishments, while Numa tries to win the opponent over by reason. Although Romulus is not anti-religious, Numa is the devotee par excellence. Under les natures (this heading is called "les affinites cosmiques" in the case of Mitra and Varona) we see that Romulus is the son of a deity, while Numa's origins are normal. Romulus does not found a gens, while Numa founds four. Romulus and Remus are the prototypes of the two groups of Luperci, while Numa is the founder and organiser of the flaminate. As far as les affinites sociales are concerned, Romulus is associated with the second function and Numa with the third. Although there is no day-night opposition to be found in the legends of Numa and Romulus, nor is there any senior-junior opposition in the relations between Mitra and Varuna, Dumezil nevertheless remarks: 90 91
Durnezil, 1958a, p.83. Durnezil, 1977, p.159-165.
THE MITRA-VARUNA CONCEPT
135
Toutes ces formules de l'antith~ compl~mentaire que constituent Romulus et Numa recouvrent des aspects de )'opposition collaborante de Varuna et de Mitra, avec les diff~rences, les r&luctions qu'imposaient et Ia condition humaine des personnages, meme du demi-dieu, et leur insertion dans le temps et dans l'espace romains, dans I'< histoire > romaine.92 Summary:Romulus- Numa The opposition between Romulus and Numa is very important as evidence for the Indo-European M/V concept from 1940 on. In the meantime Dumezil has only made a few sporadic additions to the material presented at that time. As was the case with the Mitra-Varnna opposition, we find a large degree of flexibility in the presentation of the concept and in the stress laid on particular texts. Jupiter- Dius Fidius/Fides The opposition between Jupiter and Dius Fidius/Fides is important in 1940 in two respects. It is connected with the evidence for the opposition between Romulus and Numa; and it fits in with the MN concept. After his victory over Acra, the king of Caeninum, Romulus establishes a cult of Jupiter Feretrius. Culte royal, culte ou Jupiter est bien le meme Jupiter que dans Ia vieille triade
hi~rarchi~e Jupiter-Mars-Quirinus, c'est-il-dire le dieu de Ia tete de l'Etat, le dieu
du regnum ....93
The other cult founded by Romulus is that of Jupiter Stator, who inspires the Romans with new courage in a desperate situation and thereby leads them to victory. "Jupiter Feretrius, Jupiter Stator, c'est dans les deux cas Jupiter royal, violent, triomphant; Jupiter Stator est en outre un grand magicien" 94• Numa establishes a cult of Fides, because: d'Halicarnasse ~crit (Antiquit~s Romaines, 11,75): ..." 133• The opposition between Ouranos and Zeus contains features of the MN concept, but the fact that they are successive instead of synchronic sets them in a different theological framework. Perhaps the presence of Zeus and the absence of Ouranos echo the opposition between "l'autre monde" and "ce monde-ci", but the Aegean component is too large to suppose that the opposition goes back to the MN concept134• Dumezil refers to this opposition again in 1948. He admits that Zeus is a complex deity with Aegean influence, but all the same he stands in sharp contrast to Ouranos, whose name Dumezil derives from *(O)wor-w-eno (cf. the derivation of Varuna from *Wor-n-no). A travers beaucoup de variantes, les poetes et mythographes grecs Iaissent paraitre que le a introduit dans le monde des principes de regularite, de droit, de justice, qui l'opposent a ce qu'il etait sous le regne de son premier souverain, le terrible procreateur des monstres et le tyran a Ia prise redoutable, Ouranos .135
Dumezil has not made any further use of this opposition. Summary: Chapter 4 THE MITRA-VARUNA CONCEPT Part of the proof that the tripartite strnctures were not universal, but typical of the Indo-European traditions, was supplied by the characteristic bipartition of the first function. The Indian material (Mitra - Varnna) and the Roman "historical" material (Numa -Romulus) were ofparticular importance in the definition of this bipartition. These two variants displayed the most extended fonnulation of the concept. This fonnulation turned out to be extremely flexible. The Iranian
132 133 134 135
Theogony, 173ff. Dumezil, 1940, p.55. ibidem, p.93. Dumezil, 1948b, p.448.
TIIE MITRA-VARUNA CONCEPT
145
bipartitions (Vohu Manah -Asha; Mithra -Ahura Mazda) could only be seen as representatives of the concept if its definition was considerably reduced ("/'autre monde" versus "ce monde-ci") or if the refonns of Zoroaster were assumed to have been responsible for a serious distortion. There was very little evidence to support a Roman opposition between Jupiter and Dius Fidius/Fides. Dumezil explained the paucity of the evidence by the imperialist character of Jupiter, who assumed all the other aspects of the first function. Dumezil dropped one of the Gennanic variants early on (the opposition between Odhinn and U/lr/Mithotyn). There is little evidence to support the second - the opposition between Odhinn and Tyr- and besides it can only be understood from the perspective of the "g/issement militaire". Dumezil makes occasional reference to the Greek material. Apart from that, he does not find traces of the M!V concept in the remaining Indo-European cultures.
CHAPTER FIVE
BIPARTITAMINORA "Le Borgne"- "Le Manchot"
The bipartition of the first function resulted not only in the articulation in conceptual terms, Dumezil claims, but also in a Indo-European myth. In this myth the wild, savage aspect is represented by a one-eyed god (or hero), while the mild, juridical aspect is represented by a one-handed god (or hero). In the German material these are the gods Odhinn and Tyr; in the Roman traditions they are the heroes Horatius Codes and Mucius Scaevola. We shall first concentrate on these two oppositions before moving on to the Celtic, Indian, Greek and Iranian material. Odhinn-Tyr
Tyr and Odhinn are opposed to one another in terms of functional qualities deriving from the MN concept. There is also a connection between these qualities and the related mutilations1• Odhinn became one-eyed after he had pledged one eye for a drink from the well of the omniscient Mimir. He thereby acquired the ability to see into the future, an ability which is related to his partial blindness. Tyr became one-handed after he had placed his hand in the jaws of the wolf Fenrir as a pledge. The wolf then bit his hand off. SnorrF narrated that the gods had foreseen that the wolf Fenrir would eat them once he became full-grown. That is why they tried to bind him. Their attempts were in vain until Odhinn arrived with a magic rope. During a game in which the wolf allows himself to be tied up and then tries to break loose again, the wolf is encouraged to allow himself to be tied up with this magic rope in the end. However, he is not sure that the gods will release him if he does not manage to do it himself. He asks for the hand of one of the gods in his jaws as a surety. Tyr offers his hand. The wolf is tied up, fails to escape, and bites Tyr's hand off. Odhinn and Tyr are mutilated. Dumezil considers that these mutilations are clearly connected with their function. The causal relation is clearer in the case of Odhinn than in that of Tyr. Ainsi, pour Odhinn, Ia mutilation et Ia fonction sont bien solidaires: Ia mutilation a ~t~ un paiement, Ia disgrace qui s'en suit est un certificat habilitant le dieu A sa fonction de magicien.3
Dumezil, 1940, p.111-116. Gylfaginning, 35. 3 Dumezil, 1940, p.112. 1 2
BIPARTITAMINORA
147
It is possible that Snorri's account is based on a tradition in which the relation between Tyr's mutilation and function was more obvious.
Mais, meme dans l'ttat attestt de Ia tradition, ce qui est en d~ de l'hypoth~se nous suffit: ce n'est peut~tre pas pour devenir le Juriste que Tyr a perdu sa main droite; c'est du mains parce qu'il ~tait le Juriste que, seul entre les dieux, ill'a perdue.4 This relationship between Tyr's mutilation and his function is stated more confidently in 1948. De meme, Ia perte de Ia main de Tyr est en relation fondamentale avec Ia fonction juridique de ce personnage: (This is followed by Snorri's account, W.B.) ... VoilA pourquoi Tyr est le dieu juriste.5 This passage is literally cited again in 194~, though without the last sentence. In the meantime this interpretation has assumed a position of importance. In a chapter entitled "Rehabilitation de Snorri"7, Dumezil tries to defend the reliability of Snorri against the attacks by E. Mog!cS. The myths which Snorri related in connection with his scaldic kennings were not imaginative products designed to explain these kennings, as Mogk had assumed. Without realising it, the connection which Snorri had outlined between the mutilation of Tyr and his function is proof of the antiquity of the myth preserved by Snorri. The connection is not just a Germanic one, but it is Indo-European. If this tradition is authentic, why should Snorri be unreliable on other points? D~ Jars comment admettre que ce ressort (Ia trompeuse mise en gage de Ia main droite) qui est l'essentiel, puisque, i\ l'~poque indo-euro¢enne, c'est lui seul qui justifiait le symbolisme de Ia mutilation du dieu (le dieu Juriste devait etre manchot de sa dextre comme le dieu Voyant devait l\tre borgne), ait ~t~ oubli~ des Germains, puis retrouv~. r~imagin~ au Xllle si~cle par un caprice de Snorri, - alors surtout que Snorri ne percevait certainement pas avec Ia meme clart~ que nous pouvons le faire aujourd'hui, grace i\ l'~tude comparative des religions indo-euro¢ennes, Ia solidarit~ antithetique d'Odhinn et de Tyr ni Ia compl~mentarit~ de leurs deux mutilations, de l'oeil de l'un et de Ia main droite de l'autre, et que, par consequent, il ne comprenait peut-l\tre plus bien le rapport entre Ia dextre perdue et le caract~re juriste du dieu Tyr? 9
4 5 6 7 8 9
ibidem, p.114. Dumezit, 1948b, p.450. Dumezil, 1949a, p.162. Dumezil, 1948c, p.81-109. Mogk, 1923; 1925a; 1925b. Dumezil, 1948c, p.95-96.
148
CHAPTER FIVE
Horatius Cocles - Mucius Scaevo/a Dumezil fmds a parallel to these functionally mutilated gods in Roman history. The variant is "rattachee non pas a deux < Souverains > (!'orientation republicaine de ces recits ne le permettant pas), mais a deux < Sauveurs de l'Etat > " 10• These two heroes are Horatius Cocles and Mucius Scaevola. The former defended the Roman bridge single-handed when Porsenna wanted to take the city. Horatius is called "Cocles" because of his "grimaces terribles, qui paralysent l'ennemi" 11 • This is either because he had already lost an eye in a previous fight, or because his nose was so flat and his eyebrows met, giving a one-eyed impression. Mucius Scaevola creeps into the enemy camp in an attempt to kill Porsenna. He is unsuccessful, but he does manage to kill the man sitting next to Porsenna. Then he swears that there are three hundred young Romans ready to attempt to kill Porsenna as he has done. During this oath he thrusts his hand into the fire and burns it. Whether Mucius' claim is correct or not, Porsenna believes him. It is now possible to make a comparison between Mucius and Tyr. Le n'a certes pas meme m~canisme dans Ies deux cas: Ia main d'abord risqu~e de Tyr est vraiment une caution de loyaut~, Ia main imm~diatement d~truite de Mucius est un echantil/on d'Mrorsme; mais Ie r~sultat est Ie meme..... 12
Whether motivated by fear or admiration for the bravery of his opponent, Porsenna is led by this action to begin peace negotiations with Rome. This fits in well with the Mitra aspect of the first function. Comme il convient A un repr~sentant de Ia ~rie , Mucius est vraiment un pacificateur, qui tourne I'esprit de l'ennemi Aune paix honorable, Aune amiti~ durable .... 13
Tyr can be considered as the saviour of the gods and Mucius as the saviour of Rome. This is because both have lost a hand in an attempt to get the enemy to believe in a deceitful plan which is necessary for success or in a necessary lie. Odhinn and Horatius can be compared with one another because of the way in which they can paralyse magically by their look. Although this gaze fits in with another concept as well (seep. 158f.), Dumezil stresses the magic aspect in this connection. This can be seen in 1948 as well, where the "conception bipartite de }'action souveraine s'exprimait par un double symbole: le personnage qui triomphe par Ia pratique magique n'a qu'un oeil, est borgne" 14• However, in the 10 11 12 13 14
Dumezil, 1940, p.115. ibidem,p.l17. ibidem, p.122. ibidem. Dumezil, 194&, p.95.
BIPARTITAMINORA
149
same year Dumezil emphasises the absence of function of Horatius Cocles and Mucius Scaevola. D~s 1940, j'ai soulign6 cette situation ambigue de Cocles et de Scaevola - ~ qui correspondent, dans Ia mythologie germanique, par exemple, les deux plus grands dieux: ils sauvent bien Rome, ils re~ivent bien des honneurs inou~, mais ils n'ont aucun r61e politique, ils ne sont ni consuls, ni s6nateurs, ni orateurs; dans cette r6publique, ils sont d6pays6s, ils sont - mutiles l'un des yeux, l'autre des mains. 37
34 35 36
37
Dumezil, 1974, p.21. Dumezil, 1977, p.199.
Kaushitaka Br§hmana, VI, 13.
Dumezil, 1940, p.137.
154
CHAP'IER FIVE
In 1948 Dumezil comments in connection with the interpretation of the hero Yudhishthira as the representative of the one-handed figure (see below, p. 202) that there is a contradiction between this interpretation and that of Savitr and Bhaga. The use of the theme at this point corresponds to the Roman and Germanic usage, "que n'est celle de Ia theologie vedique, oil il y a inversion des symbolismes"38• In 1956 Dumezil refers to the 1940 passage in the course of a discussion of Savitr and Bhaga, but he also remarks that he no longer accepts the interpretation given there. Bhaga is blind, which is not the same as one-eyed39• He can also refer to this passage later in countering an attack by J. Brough during a BBC broadcast on 10 April 1957. His attack on the interpretation of Savitr and Bhaga is right in principle, Dumezil admits, but it has now been overtaken by events. At first I tackled this enormous research on my own, later on I did it with a few others, Dumezil says, and mistakes are made in the course of the progress step by step. He then adds: II est de stricte justice que les critiques serieux, avant d'aborder une de nos questions, s'assurent qu'ils en connaissent le dernier ~tat: un ~change de lettres, facile entre nous, eat inform~ Brough. 40
Dumezil refers to this interpretation once more, in 1959. This time too he retracts the remark on Bhaga. "Aveugle" est tout autre chose que "borgne": done supprimer le dernier chapitre de ~meed., 1948, pp. 189-204... rtdig~ avant qu'eut ~t~ reconnue Ia structure compl~te de Ia th~ologie de Ia souverainet~.41
Mitra-Varuna.
This new structure, in which Bhaga has a role to play, will recur later on (see p. 204ff.). By "structure complete" Dumezil means the model described in 1952 in which the Adityas could be accommodated42• In 1985 Dumezil says a definitive farewell to the interpretation of both Savitr and Bhaga. "Erreur: pas dans l'Inde, oil Bhaga completement aveugle, Savitr prive de ses deux mains, relevent d'un autre symbolisme"43 • Cyclops- 100-handed giants
The Greek material also plays a subordinate role in this mythological variant of the bipartition of the first function. In 1940 and 1948 Dumezil refers to
38 39
40 41
42 43
Dumezil, 1948d, p.81, n.3. Dumezil, 1956c, p.97, n.2. Dumezil, 1957a, p.24. Dumezil, 1959d, p.9, n.22. Dumezil, 1952a, p.40-78. Dumezil, 1985, p.325, n5.
BIPARTITAMINORA
155
Apollodorus' account of the battle of Zeus versus Cronus and the Titans. Zeus is victorious over the Titans when he calls in the assistance of the sons of Ouranus who had been thrown down into Tartarus: the one-eyed Arges, Steropus and Brontes, and the 100-handed Briareus, Gyges and Cottus. The one-eyed Cyclops provided Zeus and his brothers with the weapons needed for the victory and after the Titans had been defeated and thrown into Tartarus, the 100-handed giants were appointed as guards. Dumezil concludes: Or l'histoire des Ouranides........ .fait intervenir, comme premiers enfants et victimes du terrible Ouranos, puis comme allies~ Zeus, deux groupes symetriques d'etres dont les uns n'ont qu'un oei~ dont les autres ont cent mains.44
Dumezil repeats this in the 1948 revised edition, but he does not make any further use of this variant. Taxmoruw- Ahriman
In 1974 Dumezil finds yet another variant. This one is from Iran and is confmed to the one-handed figure 45 • In the Avesta the second king of the world, Taxma Urupi, appears as the rider of Angra Mainyu. He did this for thirty years. What happened after the thirty years is contained in a late Parsi Rivayat, which Dumezil uses in Christensen's translation46• (Dumezil had already used this passage in 1929, see p. 161.) Taxmoruw ( = Taxma Urupi) is thrown off and eaten by Ahriman ( = Angra Mainyu). Taxmoruw's brother Jamshid managed to free him on the advice of Srosh ( = Sroasha). Ahriman is crazy about music and paederasty. Jamshid seduces Ahriman, but he asks Ahriman to let him remove Taxmoruw through Ahriman's anus before the latter reaches a climax. Ahriman is already extremely excited and agrees. As soon as Jamshid has freed his brother they escape. However, his hand withered and began to smell, until he cleansed it with cow urine. In the Iranian version the story is developed as a struggle between good and evil, between Ahura Mazda and Ahriman. It is therefore impossible in this version for Ahriman to be bound in primeval time, as happened with the wolf Fenrir. The theme with the combination of the hand that is bitten off and the tying up of the dangerous monster cannot be expected here. Unfortunately, there is no one-eyed figure here.
44 45 46
Dumezil, 1940, p.140. Dumezil, 1974, p.22ff. Christensen, 1917, p.184-189.
156
CHAPTER F1VE
Thus I am tempted to suggest that this Iranian story has been inherited from very ancient times, but that it lost its ancient import as a consequence of the Mazdean reform. 47
Summary: "Le Borgne"- "Le Manchot". Odhinn and Tyr were not only conceptually related, but they were related in mythology too. Odhinn is a magician, because he pledges his eye. Tyr is a lawyer, because he loses his hand. Dumezil finds a variant of this theme in Roman history. Horatius Cocles defends the bridge over the Tiber against Porsenna, assisted by the magical influence of his single eye. Mucius Scaevola is able to make a deep impression on Porsenna, and to get him to make peace, by means of a false oath, by which he loses his right hand. Although Dumezil did not fail to respond to criticism, he maintained these interpretations and kept on adding new arguments. He also looked elsewhere for parallels besides these two variants, of which the Roman one lacks a functional colouring. At first Odhinn and Tyr seemed to be exponents of a widespread Indo-European theme. Initially Dumezil listed the Celtic variants as follows: Nuada - Balar; Lludd - Yspaddaden Penkawr; and Lludd - a magician (1940), followed by Nuada -Lug Samildanach and Lludd- Llevelys (1948). Later he only retained Nuada (1974), and in 1977 Nuada was removed from the list of one-handed figures as well. An Indian variant was retracted later and a Greek variant was only mentioned in the two editions of Mitra-Varuna (1940 and 1948). Finally, Dumezil also mentions an Iranian tradition featuring a one-handed figure. Thus Dumezil adduces many vague parallels in the Indo-European traditions to support his theory that Odhinn - Tyr and Horatius Cocles - Mucius Scaevola represent a theme of Indo-European mythology. Practically all Indo-European parallels are withdrawn in a later phase of the project, but the conclusion based on the existence of a large number of variants is still retained. Concepts related to the bipartition Dumezil has also developed a number of other, less important scenarios closely connected with the MN concept and the theme of "Le Borgne" - "Le Manchot". These are the "polytechnicien" versus the "grand magicien", the "mauvais roi temporaire" and the "grimace heroi"que".
The "polytechnicien" versus the "grand magicien" This scenario is to be found in 1939, 1940 and the second edition of 1948. In 1939 Dumezil contrasts Ullr/Ollerus to Odhinn as "le Majestueux'' to "l'Inspire 47
Dumezil, 1974, p.24.
BIPARTITAMINORA
157
furieux'' (seep. 139f.), but that is not all there is to it. Ullr is also the initiator of a number of techniques. He has invented the ski and the skate. He is called the "god of the bow'', and the shield is called "ship of Ullr" in the scaldic kenning. Dumezil remarks on this material: Ne s'opposerait-il done pas A I'Inspir~ comme le Technicien, au dieu de Ia Grande Magie des runes comme le dieu des petites magies artisanes en des connaissances ajust~es ..... ?48
This scenario opposes the "patron des travailleurs" to the "patron des grands". This is also expressed in the toponyms. The aristocratic Odhinn hardly occurs in this more "popular" material, but Ullr does. Dumezil now also places the abdication of Nuada 'with the silver hand' in favour of Lug Samildanach in the same scenario. The latter is master of all the skills of a smith, carpenter, etc. In contrast to the previous theme, the one-eyed Odhinn and the one-handed Nuada are here associated with one another, and the same applies to Ullr and the one-eyed Lug Samildanach. Dumezil does not go into this scenario in more detail in his 1940 treatment of the MN concept. Nous laisserons de c6t~ Ia forme de l'antith~se: elle ouvre en effet une voie de recherche fort importante, mais qu'il ne peut ~tre question d'aborder ici, puisque Ullr semble s'opposer a Odhinn, entre autre s¢cifications, comme le patron des techniques pr~cises (il est l' du patio, du ski..) au maitre de Ia grande magie toute-puissante, comme le dieu artisan au dieu chamane. 49
Here too Lug Samildanach is referred to as the "dieu de taus les metiers".
The "mauvais roi temporaire" The main figures in this scenario are the "royaute temporaire", the "faux roi", the "roi de carnaval", etc.5°. They all involve a period of temporary misrule by a usurper in between two periods of normal government. For example, the arrogant Nahusha wields power during a period in which Indra has been deposed; the tyrant Azdahak comes to the throne after the fall of the Iranian Yima, only to be deposed in turn by Feridun; and Tarquinius Superbus rules in the period between Servius Tullius and the consuls. One or both of the two governments on either side of the carnival-like interregnum are of the military kind (Indra - interregnum - Indra; Yima - interregnum - Feridun; Servius -
48 Dumezil, 1939a, p.40.
49 Dumezil, 1940, p.%.
50 ibidem, p.88.
158
CHAPTER F1VE
interregnum - Brutus). The opposition within the MN concept cannot be defined in terms of legal-illegal or good-bad. What we are here presented with are two legitimate and necessary aspects of power and normal government. The theme of the "antithese des deux types de souverain" is strictly separated from that of the "mauvais roi temporaire" in India, Iran and Rome. In the case of Othinus- Mithotyn/Ollerus, however, they overlap. The counterpart of Othinus within the MN concept is also the temporary usurper. Dumezil only returns to this material in 1948 (in the second edition ofMitra-VtllUna). We saw above that in the opposition between Othinus and Mithotyn there was a difference with respect to the distribution of property. Mithotyn stood for an organised society with private property, Othinus for an undivided system of property. The "mauvais roi temporaire" is here the ruler of the orderly system of private property, while the "bon roi" Othinus propagates the holding of property in common. The same opposition can be found in Ireland. After Nuada had lost his right arm he was unfit to rule, and had to make way for another. This was Bress, a chief of the Fomoire, who was also related to the Tuatha De Danann. Bress' government results in an economic tyranny. He levies exorbitant taxes, makes use of forced labour and campaigns against private property. After he has been cursed by a "file", a bard who ridicules him in a poem, his power rapidly slips away. Bress and Othinus are opposed to private property and are in favour of state ownership, while Nuada and Mithotyn are against communal ownership and are in favour of private property. The roles are thus reversed among the Germans and among the Irish: while Mithotyn is the usurper among the former, Bress is the usurper among the latter51 • The ''grimace heroi"que"
In 1939 Dumezil discusses the initiation of warriors under the heading "le premier duel" 52• During an initiatory duel the warrior receives his title and power, and often his outward aspects as well as a kind of stigmata. One of these stigmata was the "grimace heroique". When Cuchulainn grimaces, one eye becomes as small as the point of a needle and the other becomes as large as a cup. Another example is Egill Skallagrimsson, who adopts a "heroique" pose before the king to demand payment for his services in a battle in which the king was victorious. This "heroique" pose consists of lowering one eyebrow to his chin and raising the other to his hair. The Indian tradition recognises physical characteristics (lakshana) in the case of the great men (mahapurusha). Dumezil discusses Cuchulainn and Egill again in 1940 and 1948. Now they feature in a section dealing with the one-eyed figures. Dumezil comments on Egill's "grimace heroique":
51 52
ibidem, p.107-110. Dumezil, 1939a, p.92-106.
BIPARTITAMINORA
159
...pour le viking Egill, il s'agit d'une grimace qui fait partie d'une gesticulation h6roi"que apparemment traditionnelle, puisqu'elle est comprise de celui il qui elle s'adresse.53 This grimace recalls the Harii, who had a terrifying appearance when they went out to do battle, according to Tacitus54 • Odhinn certainly also used "le mauvais oeil" to paralyse his opponent in war. In 1940 Dumezil includes the magic of Lug, who dances around the army on one leg and with one eye closed, in the same theme. In 1948 he is the one-eyed counterpart of Nuada, and the "grimace heroi'que" has now become a "pose magique". Summary: Concepts related to the bipartition A number of concepts played an important part in the fonnulation of the M/V concept besides the "Le Borgne"- "Le Manchot" theme. The scenario of the ''polytechnicien" versus the "grand magicien" embodies the opposition of the artisan to the aristocrat. This was supposed to strengthen the interpretation of Ul/r- Odhinn and Lug Samildanach - Nuada. In the scenario of the "mauvais roi temporaire" the ruler is temporarily deposed by a bad king. This opposition between a good and a bad king provides more insight into the interpretation of Othinus - Mithotyn/0//erus, because in the Gennanic tradition this scenario overlaps with the M/V concept. One of the external characteristics of the initiated warrior was the "grimace herofque ". This grimace recalls the gaze of the one-eyed figures from the "Le Borgne" - "Le Manchot" theme. Material was borrowed from this scenario in a number of cases. Theoretical precursors of the bipartition
Dumezil advanced a number of other theories before he developed his theory. of the bipartition of the first function. They are important for an understanding of the origins of the theory of bipartition. Le probleme des Centaurs
This 1929 publication is an investigation of the possible kinship between Sanskrit Gandharva and Greek Kentauros. Dumezil's approach is based on the European carnival. He constructed a prototype of the European carnival, drawing mainly on Slavonic material. The term "gody'' is found in a number of Slavonic languages. Its meanings include festival, Christmas or Easter festival,
53 Dumezil, 1940, p.l18. 54 Germania, 43.
160
CHAPTER FIVE
and year. Originally this was a New Year's festival lasting twelve days. Young men wandered about disguised as animals, making a lot of noise and harassing the young girls. They are mysterious beings which are the object of both awe and laughter. The game with the Horse also belongs to this period. The Horse can be constructed in different ways, and the manner in which the head, body and tail are made varies from village to village. Various people accompany the Horse as it makes its rounds from farm to farm. These creatures - the young men in disguise and the Horse - were demonised by the Church early on. In popular belief they were seen as creatures which live in the wilds and are closely connected with the world of the dead. Figures du Temps promen~es au changement d'ann~e; d~mons de toutes sortes profitant de l'occasion pour s'approcher des hommes; ames des morts hantant les vivants et voulant revivre, telles sont les trois principales cat~gories de repr~sentations mythiques qui, sans souci de logique ni meme d'tquilibre, expliquent concurrement Ia procession des m5ues A Ia fin de l'hiver et, lorsque c'est lA une diff~rente, AIa fin de l'ann~e. 5
Comparable beings, the Kallikantzaroi, can be found in modern Greek folklore. This is followed by an investigation into the older Indo-European mythology. The question is whether the Centaurs and the Gandharvas can also be described in these terms. Are they connected with carnival-like festivals at the end of the year, and with demons and the souls of the dead who visit people in the civilised world at this time? We shall review Dumezil's treatment of the Iranian, Indian, Greek and Roman material and see what conclusions he draws from it. Iran People are visited by the souls of the dead during the last ten days of the Old Year. During this period people stay indoors. It is terminated by the Ameredad-Sal, a joyous festival on the first day of the New Year. The New Year begins officially on the sixth day, the day of Khordad-Sal. The period of sixteen days forms a unit. The most important festival, the Khordad-Sal, has attracted all kinds of myths like a magnet. Dumezil distinguishes three elements in this collection of myths. There are artificial additions. It is the day on which the world was created, the first human couple made its appearance, religion was revealed to Zoroaster, and it will be the day on which the dead will be resurrected in the future. There are myths which originally belong to the Khordad-Sal. These are the myths which Dumezil had studied in his dissertation and which belonged to the "ambrosia cycle" that he postulated there 56• There are myths which really belong to the more popular
55 56
Dumezil, 1929, p.48. Dumezil, 1924a.
BIPARTITAMINORA
161
Ameredad-Sal festival. This corpus includes myths associated with Azdahak, who rules in the palace of the stork, surrounded by demons with a monstrous appearance, in a way which appeals to the imagination. His predecessor Takhmurop compels Ahriman to be his mount. Ahriman swallows his rider, who is rescued by his brother Yima. Yima also initiates the New Year festivals, the No-Ruz. He does so as he rides through the air on his ivory chariot on the ftrst day of the ftrst month (Farvaddin). Later he will be thrown from this chariot for his overweening pride. This chariot is borne by demons. The narrative in which Sam Keresaspa kills the demon Gandarep(ak) also belongs to this carnival-like setting. Gandarep(ak) is one of the sons of Yima. His brothers include a bear and a monkey. They are born during the carnival-like reign of Azdahak, from whom Yima has fled. This material recalls the tradition of European animal disguises. Later Islamic traditions do not contain a duel between Sam Keresaspa and Gandarep(ak), but the latter crops up again as Kndrv. He is a minister of Zohak ( = Azdahak), who is dethroned by Feridun exactly six months after Khordad-Sal. Kndrv organises a big festival for Feridun, with drink, music and a banquet. In some traditions Gandarep(ak) is thus a demon, while in other traditions he is a master of ceremonies. This is in accordance with the awesome and ridiculous character of the European carnival figures. The duel between Sam Keresaspa and Gandarep also contains elements drawn from the carnival. One tradition motivates the duel with a rape, and the demon is flayed afterwards, which reminds Dumezil of the animal disguises. He concludes: Mais revenons, pour conclure, a notre monstre. Ses aspects changeants, ses natures contradictoires, ses relations et associations diverses, les deux formes de son mythe.......s'expliquent des qu'on rapporte tout a cette donn6e centrale, directement attestee d'ailleurs: Gandareva etait un demon des fetes de changement d'annee. 51 India In the Brahmanas the Gandharvas are mainly connected with the Vajapeya offering and the Rashtrabhrt oblation. Dumezil regards the Vajapeya, whose symbolism is closely connected with the months and seasons, as a typical New Year festival. A number of elements in the Vajapeya immediately remind him of the European carnival that is the object of his investigations. There are horse races, the person making the offering rises to heaven (like Yima on his ivory chariot), alcoholic drinks are consumed, the wife of the person making the offering is involved, and there is a preceding period in which people are not visited by the dead, it is true, but in which daily sacrifices are made to the Gandharvas. The Gandharvas are invoked during the bridling of the horses of
51
Dumezi11929, p.91.
162
CHAPTER FIVE
the winner, who is known beforehand. Dumezil is convinced that they are "genies de Ia fete de changement du Temps"58• The Gandharvas occur in the mythology as musicians in the company of Indra, in association with the carnival-like kings Nahusha and Uparicara, and with the demigod Kubera. In the myths in which the latter three figures occur they ride round on wagons from which they are pulled down, as in the European ritual. These wagons were borne or drawn by strange creatures. The European disguises also explain the multiple and sometimes contradictory ways in which the appearance of the Gandharvas is described. The inconsistencies are essential. Not only Nahusha but also his grandfather Pururavas is closely associated with the New Year festival. Urvasi, a friend of the Gandharvas, falls in love with Purilravas. They live together for a long time until Pururavas no longer complies with her terms: she was never to see him without his royal robe, but the Gandharvas ensure that this happens. They steal Urvasi's goats, and when Pururavas prepares to pursue them without dressing first, the Gandharvas illuminate the night sky with a flash of lightning. Urvasi returns to the Gandharvas and Pururavas does not see her again until New Year's Eve. The next day the Gandharvas grant him a wish. He wishes to become one of them. He is initiated and his wish is fulfilled. The myth thus explains how it is possible to become a Gandharvas at the end of the year.
Greece The Greek Centaurs fit into this picture as well. These equine creatures live in a natural habitat, have contact with the dead, drink together and chase women. They are the classical counterparts to the disguised young men of the carnival. Heracles kills a number of Centaurs in a fight near Pholos and has to be purified through his initiation in the lesser Eleusinian mysteries. This rite took place during the transition from winter to spring. The Centaurs are thus connected with the Old and New Year. Heracles descends to the underworld after he has been purified, where he is represented with carnival-like traits, running round with Hades on his back.
Rome The festival of Lupercalia was held in Rome on 15 February. This festival belonged to a sequence of other festivals with a carnival-like character. The most important traditional ritual was that of the naked Luperci, who rode round striking the bystanders with their goatskin whips. The association with goatskin, their equestrian connection and their etymological link with the wolf (lupus) are all elements which resemble the revellers in the European carnivals. The key term is Februa-. The whips are called februa; the people who are
58
ibidem, p.122.
BIPARTITAMINORA
163
purified by being beaten with them are februati; the day of the Lupercalia is februatus; and the ceremony is februatio. Although not every tradition records Romulus and Remus as the first Luperci, at any rate they are the first Luperci to be named. The Luperci run round naked in memory of the race of Romulus and Remus, who tried to recover their stolen herd as follows: they ran away naked, leaving their meat behind on the fire. Remus was the first to return and he ate the meat. The running Luperci who had no time to dress as they chased the cattle thieves bear a close resemblance to Purftravas, the running man-Gandharva, who had no time to dress when he ran naked to catch the thieves of Urvasi's kids. India too had its ritual race in the Vajapeya, which was also known as Annapeya because of the food that the offering was expected to yield. As we saw, Remus and his men eat the meat after the race. As for Romulus, he is reared by a she-wolf and later becomes a ravisher of women. The myth of the rape of the Sabine women may have been a Lupercalian myth. Conclusion Dumezil points to parallels between European folklore and the ancient traditions of Iran, India, Greece and Rome with respect to the transition from the Old Year to the New. The group of revellers which is a feature of these festivals can be derived from Indo-European. Skr. Gandharva, Av. Gandarev are derived from *Guhondhc/oruo. The Gr. Kentauros goes back to *Guhendharuo; Lat. Februum is based on *Guhedhruuo; Phryg. Gordios is based on *Guhodher(uo); Lit. Gondu is based on *Guhondh(uo); and Slav. Gody on *Guhodhu. This gives us the Proto-Indo-European *Guhl;/o(n)dh-r-u-o, in which the (n) and the -r- are variables. These reconstructed terms refer to the companies of revellers which operated in different guises at the New Year. These companies were also connected with myths and rituals, such as those of the rape and race, disguises and palace scenes. Sexual morality is much freer during this carnival period than during the rest of the ye~9•
Ouranos - Varnna An investigation begun in Constantinople in 1928 saw the light of day in 1934. It was concerned with whether and why Ouranos and Varuna are comparable. Dumezil cites Job. Schmidt60: "Varona und Uranos darf man nur sprachlich, nicht mythologisch gleichsetzen".61 This "attitude "scientifique"" can be explained by the many disappointing attempts by the school of Max Muller to
59 ibidem, pp.251-260.
60 Schmidt, 1924. 61 Dumezil, 1934, p.38.
164
CHAPTER FIVE
fmd mythological similarities between these two gods. Dumezil hopes to arrive at a solution using "Ia theorie sociologique du pouvoir'>62• He borrows various terms from this theory without making it clear exactly what he means by this theory. First the material connected with Varuna is treated. Varuna is above "le Dieu Souverain", "Dieu-Roi": "le Roi dans ses fonctions de gouvernement, de justice et de magie (c'est tout un) et non pas dans ses fonctions militaires"63 • There are not many rituals associated with Varuna, but the royal consecration, the rajasuya, is his ceremony. In the ritual the king loses his virility; in the myth Varuna loses his virility, which is transferred to the herd of one of his relatives. This is followed by a comparison with Ouranos, Heaven, though Dumezil sees this as a secondary attribute. Ouranos was the First Father, who had a relationship with Mother Earth, and the First King, sovereign of the world. In this role it is easy to equate him with heaven, which is why heaven is named after Ouranos. It is an essential aspect of Varuna and Ouranos that each of them is a "Dieu-Roi". Ouranos acquired naturalistic characteristics later on, Varuna did not. Ouranos was the god of a culture in which the *reg- no longer played a part, while Varuna was important at the consecration of the powerful raj in India. The loss of his virility by Ouranos can therefore be allowed to culminate in a total castration, not so that of Varuna, where temporary impotence, which can be imitated in the ritual, is more appropriate. Following H. Petersson64, Dumezil derives the names of these gods from the Proto-Indo-European *uer-, "lier", to bind. Varuna is the binder god par excellence. He is often described as the god who stands ready with his ropes in his hands to bind the sinner. Every knot is Varunian. Ouranos binds his sons, the Cyclops, before hurling them into Tartarus. Cronos binds his opponents too. Ouranos uses the act of binding to protect himself against his "detroneurs naturels" 65 and thereby to maintain his own sovereign position. In the case of Varuna the binding is a symbol of the magic power of the Sovereign. The *uorueno- is a "lieur", and "le Lieur, c'est le Souverain"66• From the "theorie sociologique" referred to we know that one of the functions of the king is to guarantee fertility. Varuna is connected with this theme through the loss and recovery of his virility. There is also an opposition between solidarity, on the one hand, and rivalry between the king and his relatives, on the other. Expressions of solidarity can be found in the Indian rajasuya when the king identifies himself with his son, and when it transpires that his virility has passed into the cattle of one of his relatives, he recovers it and absolves the relative. The only element of rivalry in the legend is where it is stated that the rajasuya is initiated by Varuna at the moment when the substitute for the king's son,
62 ibidem, p.7. 63
ibidem, p.41. Petersson, 1918. 65 Dumezil, 1934, p53. 66 ibidem, p57. 64
BIPARTITA MINORA
165
Shunahshepa, is to be sacrificed. However, in the case of Ouranos and his successors dynastic rivalry is the Leitmotif. Ouranos and Varuna are thus "Dieux-Souverains" whose power is not based on military superiority but on magical resources. Not only do they possess political power, but they are also important for fertility. "Souverainete et Fecondite etant d'ailleurs (quiconque a feuillete le Golden Bough le sait) des puissances solidaires, et comme deux aspects de Ia Puissance"67• They are also involved in a conflict of dynastic solidarity and rivalry. The peaceful end to the conflict in the case of Varuna is continued in the pact that he concludes with the Earth, while in the case of Ouranos rivalry is expressed in Gaia's hostility towards him. Flamen - Brahman
In 1935 Dumezil published the result of another investigation that he had begun earlier on, this time concerning the similarity between Latin flamen and Sanskrit brahman. Here Dumezil calls for a study of the persons, the flamen and the brahman, instead of a study of the flamen/brahman couple as a neuter object of investigation. He is particularly interested in a comparison of the "fonctionnement des rouages sociaux" in which both are involved. Dumezil begins by trying to trace the original meaning of brahman. On the one hand, it occurs in the Rig Veda with the meaning of the person making the offering; on the other hand, it also occurs there with the meaning of "homme sacre", but he is not yet a member of any caste here. It is difficult to decide which meaning is the more ancient. However, his silent presence at the sacrifice is completely unlike the behaviour of the other priests. As "homme sacre" he is closely associated with the king. Mais d6jA RV., IV, 50, 8 d6finit Ia veritable symbiose par laquelle le brahmane vit du roi en meme temps que le roi prospere par le brahmane....... le rajan et le bra.hman sont deux rouages de Ia machine sociale faits pour s'emboiter, pour se comp16ter. 68
The myth of Shunahshepa, which was recounted during the rajasiiya, reveals the original nature of this symbiosis. In this myth a son is born to Haricchandra, who has been childless until now. He promised to sacrifice this son to Varuna immediately after his birth. After agreeing to postpone this sacrifice on repeated occasions, Varuna agrees to a substitute sacrifice. The brahman Shunahshepa will be sacrificed by his father. Dumezil concludes from this that the function of the brahman is the following: "victime humaine proclamee preferable aux princes et, comme telle, substitue aux princes"69• A similar
67 68 69
ibidem, p.31. Dumezil, 1935, p.17. ibidem, p.27.
166
CHAPTER F1VE
scapegoat function can be detected in Rome in the immersion of the Argei (human effigies made from straw) in the Tiber during the procession on 16-17 May in which the flaminica, the wife of the flamen dialis, also took part. During the later Lupercalia, a festival dominated by the theme of purification, it is the flamen dialis who joins the rex to hand out the februa. This link between the rex and the flamen also throws light on the fact that the place where the flaminica makes her offering is the regia. Dumezil quotes Jullian on the flamen as follows: Esclave du dieu dont il est le pretre, rien d'humain ne lui semble permis. Le service divin l'enchafne plus qu'il ne l'honore, le contraint plus qu'il ne !'oblige. Il est une victime vivante, sans cesse par~e pour le dieu auquel elle appartient. 70
The relationship between rex and flamen is the same as that between raj and brahman. Since there is no difficulty in associating the rex and the raj with one another, any linguistic objections to the equivalence between the flamen and the brahman lose their weight. Dumezil also finds the scapegoat function in the case of the Iranian baresman and the Greek pharmak6s. In 1938 he sums this research up in the following words: ...il n'y a pas lA deux faits ind~pendants, un accord certain sur le nom du roi, et un accord probable ou possible sur le nom du flamine; il n'y a qu'un fait, mais un fait complexe: on tient, dans l'Inde et A Rome, les deux noms d~signant deux organes solidaires, plus exactement les deux moiti~s in~parables d'un organe unique, l'organe de Ia Souverainete. 71
Jeunesse, etemite, aube An article by Dumezil published in 1938 bears on a discovery made by E.
Benveniste. The latter had derived Lat. aeuum, "age, temps, eternite" and iuuenis, "jeune", from the same Proto-Indo-European root: *H2ei-. This root appeared in two forms: 1. *H2ei-w-; 2. *H2y-eu-. Lat. aeuum, Skr. ayu, "age, duree", Av. ayu, "eternite", Gr. ai6on, "age, vie, duree, eternite" were derived from the former, Lat. iuuenis, Skr. yuvan-, Got. juggs, Old Irish oac were derived from the latter72• Dumezil assumes that this root must have meant: "force vitale", and the *-u(w)-en-is was thus the "etre pourvu au maximum de force vitale" 73• He also links this with a number of other data. Thus Rome displays a sharp opposition between young and old, as evidenced not only in sporadic data on the killing of old people and the existence of a special goddess of youth, Iuventas, but also by the "societes secretes". Summing up earlier work,
70 ibidem, p.43. 71 Dumezil, 1938a, p.189. 12 Benveniste, 1937. 73 Dumezil, 1938b, p.290.
BIPARTITAMINORA
167
Dumezil claims that normal Proto-Indo-European life was regulated by "organes statiques"74, viz. the rex/flamen, the raj/brahman, Proto-IndoEuropean *reg!*bhlagh-(s)men-. Once a year, at New Year, the order of public life was inverted by the "monstres-masques", ...qui, en bandes, violaient syst6matiquement taus Ies tabous (sexuels, alimentaires, etc.), 6branlaient Ia soci6t6 et Ie monde dans leurs fondements, mais du meme coup Jes purifiaient, Jes revigoraient, les f6condaient pour une nouvelle p6riode.75 Youth is the most important feature of the Gandharvas, Centaurs and Luperci. Purfuavas' son is called Ayu(s). In Rome the Luperci are not only from the equestrian order, but they are also iuuenes. Romulus and Remus are iuuenes par excellence. Quoting from this article in the same year, Dumezil opposes the aged senior Noma, the disbander of the Celeres, to Romulus as representative of the iunior type, "comme pour annoncer qu'une morale de Ia gravitas et de Ia tranquillitas va equilibrer Ia mystique de Ia celeritas et de I'impetus" 16• Romulus and Numa thus feature in this context as opposite types.
Relation between the theoretical precursors and the M/V concept Regnum- interregnum = Mitra- Varuna These early precursors of the MN concept display other oppositions besides the bipartition within the concept itself, with a division based on more or less extensive epithets like wild/calm and so on. Thus we find power, on the one hand, with the gods of the *reg, Ouranos and Varona, and on the other hand groups of young men which operate at the New Year in a carnival interregnum. Regnum is opposed to interregnum in social reality. While in Dumezil's later work the bipartition within the MN concept is the result of a close analysis of the terms used for the first function, the earlier work reveals a clear connection between this concept and phenomena that can be perceived in social reality. Thus in 1939 the regnum!interregnum opposition led Dumezil to oppose Mithotyn and Ollerus to Othinus. The Berserkir, the warriors of Odhinn, are then congruous with the Gandharvas, Luperci and other representatives of the type which he had discussed in 192977• These Berserkir are apparently influenced by the "Germanic shift", they form a combination of two separate Indian categories: the carnival-like Gandharvas
74 ibidem, p.295. 75 ibidem, p.296. 76 Dumezil, 193&, p.190. 77 Dumezil, 1939a, p.79-91.
168
CHAPTER FIVE
and the belligerent Maruts led by Indra78• In 1939 Othinus is thus in opposition to Mithotyn!Ollerus as a regnum in opposition to an interregnum of the "roi des mascarades". This opposition can also be situated within the MN concept. Since there are other (toponymical) grounds for equating Ollerus/Ullr with Tyr, the latter can thus in turn be opposed to Odhinn. At this stage, however, there is no clear-cut association between the regnum/interregnum complex and the MN concept. Dumezil takes this step in 1940. After dealing with the raj-brahman relationship and the Gandharvas in accordance with his earlier theories, Dumezil now establishes a link between the two: on the one hand is the brahmanic religion, on the other the brief interregnum of the Gandharvas. " ...il s'agit de deux groupes de representations non seulement differents, mais opposes"79• One of the manifestations of the opposition is in the field of duration. The brahmans stand for "Ia religion permanente et constamment publique dans laquelle trouve place - a I'exception d'un seul jour - toute Ia vie de Ia societe et de tous ses membres"80• The Gandharvas, on the other hand, represent a different aspect. .. .ils rel~vent, eux, d'une religion qui n'est publique et accessible que dans une apparition ~pMm~re ... , qui ml!me, sous sa forme romaine tardive, n'existe que dans cette apparition et qui, en tout cas, sous des formes plus anciennes, ne pouvait l!tre, hors du jour des Lupercales, que constamment secr~te. 81
While the brahmans represent the "ordre sacre" and the "religion statique, reglee, calme", the Gandharvas represent the "desordre non moins sacre" and the "religion dynamique, libre, violente". Furthermore, they are opposed to one another in terms of a dichotomy which was to play such an important part in the MN concept later on: as "l'autre monde" (Luperci and Gandharvas) in opposition to "ce monde-ci" of the brahmans and flamines82• The ancient Brahmanic literature does not contain a clear elaboration of the opposition between the "religion des brahmans" and the "religion des Gandharva", but the opposition between Gandharvas and brahmans can be detected in a number of prescriptions, and a similar opposition is evident in the statutes on the Luperci and the flamines. Regulations affecting contact with horses, dogs and the obligations and prohibitions connected with wearing certain items of clothing oppose Luperci and flamines diametrically to one another. Gandharvas and brahmans are strictly opposed to one another in the regulations on the consumption of alcoholic drinks, contact with horses, the obligations and prohibitions connected with wearing certain items of clothing and the display of reticent or emotional behaviour. Furthermore, it is possible to qualify the 78 79 80 81
ibidem, p.154-155. Dumezil, 1940, p.13. ibidem. ibidem. 82 ibidem.
BIPARTITAMINORA
169
opposition in a number of other ways too. The Luperci stand for speed (celeritas), they are the iuuenes and they are creative, while the flamines are ruled by gravity (gravitas) as the dignified seniores who preserve tradition. The postulated opposition between a religion lasting most of the year and a religion which only dominates public life for a brief spell each year also yields the terms for the opposition in the MN concept. As we have seen, in dealing with the tripartition initially Dumezil saw a one-to-one correspondence between social facts and ideology. The same applies to this bipartition of the first function, and here too this correspondence declines in importance later on.The opposition reappears in an article that already existed in 1940 but was not published until 1948. In this article, Dumezil comments on the *Guhedhrwo-, *Guhondherwoas follows: II semble done qu'on tient ici et !'image et le nom indo-euro¢en d'un type de sacerdoce violent, magique, mysterieux, oppose au sacerdoce regie, ritualiste, public des *bhelgh-men-, et que cette opposition est fondamentalement liee a celle des deux types de rois, de souverains: Romulus et Pun1ravas d'une part, Numa et Manu d'autre part.83
He makes a similar remark on the opposition between the Luperci and the flamines, and on that between the Gandharvas and the brahmans, in his inaugural lecture, delivered on 1 December 1949. They form: "des couples de corps sacerdotaux a caracteres opposes comme le sont les deux dieux souverains eux-memes"84• This is the last occurrence of such a correspondence. When Romulus is mentioned as one of the Luperci, and Varuna as the "roi des Gandharva", this is illustrative of the MN concept and follows from it, but it is not a reason to postulate the concept itself. Brahman - raj
=
Mitra - Varuna
There is not just an opposition between the brahmans and the Gandharvas. The static, normal period itself is also subdivided. The raj and the brahman form a closely associated pair. They appeared earlier in this chapter as the "deux rouages de Ia machine sociale" and as the "deux organes solidaires... d'un organe unique, l'organe de Ia Souverainete". This king-priest duo is opposed to the Gandharvas!Luperci as the Mitra aspect is opposed to the Varuna aspect in the MN concept. Here we come up against a problem: that of the double function of Varuna and Odhinn in these pairs. In 1935 Dumezil investigated the brahmanic aspect of the raj-brahman pair, while a year earlier he had investigated the raj aspect in his study of Ouranos and Varuna. In 1939 Odhinn is qualified as a "Varuna" and a "roi-magicien". However, Varuna and Odhinn 83
84
Dumezil, 1948b, p.447-448. Dumezil, 1951a, p.229.
170
CHAPTER FIVE
not only feature in the raj-brahman pair, in which they are the divine representatives of the raj, but they also play a part in the Gandharvasbrahmans opposition. Varuna is thus both the god from the rajasuya and the "roi des Gandharva", and Odhinn is both the Varunian "roi magicien" and the leader of the Berserkir. The results of this complication can also be found in the MN concept. This is sometimes the expression of the distinction between a normal, quiet period and an abnormal, carnival-like interregnum, and sometimes the expression of the king-priest duo. In 1940, 1941, 1943, 1948, 1949 and 1951 Dumezil introduces reflections on the bipartition of the first function with an examination of the relation between rex and flamen. Dumezil formulates the connection in 1940 and retains it unchanged in 1948. After an account of the Hegelian structure of the Indian social classes, i.e. the brahmans against the kshatriyas, these two as the "ubhe virye" against the vaishyas, and these three as Aryans against the non-Aryan shUdras, Dumezil also detects a similar opposition within the first function: the raj-brahman couple. He remarks on this archaic pair: ...mais, meme dans l'ttat historique que nous pouvons observer, en plein imptrialisme brahmanique, les tltments et Ia formule de Ia synth~se restent perceptibles si nous considtrons non les brahmanes eux-memes, mais les dieux qui les doublent, les dieux qui gouvernent de haut Ia grande affaire des brahmanes, le sacrifice, - et qui se trouvent etre en meme temps les dieux souverains, Ia projection cosmique de Ia Souverainett terrestre: Mitra et Varuna.ss In 1941 Dumezil discusses the duplication of the first function in India in the section "Raj et brahman, rex et flamen dialis". Dans l'lnde, Ia premi~re fonction sociale est double dans sa nature, Ia toute-puissance ttant a Ia fois magique et juridique, et elle a pour patron un couple divin: celui du magicien Varnna et du juriste Mitra. 86 He also refers to a similar relationship in 1957 when he speaks of the "dualite" V aruna-Mitra. ...mais il est tvident que Ia partie de l'idtologie qui contient a Ia fois, en un diptyque conscient, une moitit magico-religieuse et une moitit juridico-religieuse, patronntes par le dieu a maya et par le dieu-Contrat, inttresse a Ia fois le culte et le gouvernement, les pretres et tout ce que nous appellerions les agents de l'autoritt, roi, ministres,juges, etc.87
8S 86
87
Dumezil, 1940, p.43-44. Dumezit, 1941c, p.69. Dumezil, 1957a, p.14.
BIPARTITAMINORA
171
Shortly before this, however, in 1956, a change took place which has a bearing on our problem. From then on the relationship between rex and flamen dialis is an independent problem. Dumezil presented this new view at a conference (a summary of Dumezil's paper was published in 1956BII) and published a longer version in 195~. Dumezil's claim was that rex and flamen dialis are a double organ, but the functional synthetic character of the rex now occupies a prominent position. It is particularly evident in the cults in the Regia. The rex is the synthesis of the three functions; all the same, he is sometimes closely associated with the first function, and he is also sometimes superior or external to the social division. In 1957 Dumezil repeated part of the 1956 paper and went on to say that the situation in India was different 90• There the raj came from the second class, that of the kshatriyas. He is therefore closely connected with this class in some rituals, although as leader of the society he stands above all classes. Dumezil referred to this intricate problem in 1958 in a section entitled "le probleme du roi"91 • No consistent picture emerges on the social functions from the Vedic material on the raj, the Latin material on the rex, or the Celtic material on the ri'g. At any rate, the old idea of the rex and his ritual surrogate, the flamen dialis, is abandoned in the longer version ofthe paper. J'en ai jadis conclu quelques fantaisies sur I'origine du jlamonium, avant d'avoir pris conscience de Ia valeur fonctionnelle de Ia triade des dieux des flamines majeurs, et dans un sens que je croyais, que je voulais frazerien; il faut abandonner ces illusions....92
The problem of the king and of his relation to the three functions is left untouched until1983. Dumezil cites from an earlier study on tripartition among the Germans: the 1958 investigation of the Rigsthula93• A feature of this poem had been neglected, Dumezil now remarks. On the one hand, this poem contains the name Konungr, a typically German term, while on the other hand the name of the ancestor and tutor of the king, Rigr, is a Celtic word meaning king. Dumezil's explanation of this use of the two terms is as follows: the poet was familiar with both the Germanic and the Celtic view of the relationship between the king and his functions, and he wanted to express this distinction. The Celtic king was seen as a synthesis of the three functions and was not included in the social hierarchy. The Germanic king, on the other hand, fulfils both aspects of the first function as a magico-religious leader. Dumezil presents this in a diagram94 :
88 89
90 91
92 93
94
Dumezil, 1956d. Dumezil, 1959c. Dumezil, 1957a, p.16. Dumezil, 1958a, p.32-33. Dumezil, 1959c, p.416. Dumezil, 1958d. Dumezil, 1982, p.225.
172
CHAPTER FIVE
Figure 15: The Germanic and the Celtic king
rm (I,II,III)
Irlande
Angles et Saxons
I. Druides II. noblesse militaire, flaith III.eleveurs-agriculteurs, bo-airig
I ( + II + III) roi II. noblesse militaire III.paysans libres
The konungr is the representative of the first function in the Germanic system. Rigr, the bearer of the Celtic name, functions as a Celtic king, who is the synthesis of the three classes. He is the ancestor of all social groups. At the level where Rigr should have a Germanic counterpart to the druids we now fmd the magico-religious king with a Germanic name. The Celtic king is external to the social structure, the Germanic king belongs within it. Dumezil also sees in this the explanation of a conflict in the church in the Middle Ages. In the Celtic construction the bishops could replace the druids, but there is no room in the Germanic construction for a separate priestly class except in place of the king. Dumezil claims that this gives an extra dimension to the later conflict between the Pope and Henry VIII. A common feature of the two extremes of the Indo-European cultures, the Italo-Celtic and the Indo-Iranian, is that they both had an organised priestly class and a term for the king *reg-, while these features were lacking in the intervening cultures. But this has taken us a long way from the original interpretation of the rex-flamen relation and its connection with the MN concept. Summary: Theoretical precursors of the bipartition Dumezil had already developed a few models closely connected with the M/V concept before he thematised the M/V concept itself. In 1929 he came to the conclusion that carnival-like festivals were already held in Proto-Indo-European culture. These festivals marked an umuly interregnum, dominated by Gandharvas, Centaurs, etc. The rest of the year was under the control of the brahman and the flamen. In 1934 Dumezil outlined a picture of Varona which was later to fit in well with his bipartite model of the first function. In 1935 he concluded that power ("Ia Souverainete") has two aspects or two sides to it, which are embodied, for instance, in the rex and the flamen. The opposition between brahmans and Gandharvas, which Dumezil had discovered in 1929, is translated in tenns of young and old in 1938. Numa and Romulus are good examples of this. The relations between the M/V concept and the other theories can be summed up as follows. The M/V concept originated in the ritualistic Frazerian interpretation of the raj-brahman couple. Dumezil then goes on to fill this opposition in with tenns originating in a further elaboration of his earlier interpretation of Centaurs and
BIP ARTITA MINORA
173
Gandharvas. Until 1951 Dumezil considered it necessary to discuss these two theories before proceeding to describe the M/V concept. Subsequently the two models lose their function. The carnival-like inte"egnum of the Gandharvas is no longer mentioned, and the interpretation of the raj-brahman pair is formally abandoned in 1959. The question of the co"ectness of these models is no longer relevant. As heuristic theories, they have had their day. The exclusiveness of the bipartition
As we have seen, the opposition within the MN concept can be described in a number of ways. It might be argued that the distinction becomes so general that it can no longer be used for a genetic reconstruction, which makes heavy demands on the degree of specificity. Dumezil discusses this point in 194095 • He compares the Chinese yin/yang opposition to bring out the specificity of the MN concept. The salient feature of the yin/yang opposition is that it is all-inclusive: it is a general principle of classification which can illuminate the contrasting sectors of society, culture, the arts, etc. Mitra and Varuna, on the other hand, are exclusively the contrasting elements within the domain of Sovereignty. They do not apply to the "combattants non-souverains", nor to the "pasteurs-agriculteurs". At the pressing of the soma sacrifice, we fmd not only Mitra and Varuna but also Indra and Vayu and the twin Ashvins. The latter are "solidaires, equivalentes, au point d'etre indiscernible", and the relation between Indra and Vayu is merely one of the possible relations between Indra and other deities. However, Dumezil does find a non-Indo-European form of bipartition in the Christian dogma of the son as intercessor and the father as the one who punishes. This Near Eastern view of divinity is probably taken from Iran, an Indo-European culture. This exclusiveness of the bipartition of the first function is abandoned later on. Now the second function displays bipartition too, first and foremost because of the Indra-Vayu opposition, which has now increased in importance in Dumezil's eyes. His interpretation of the Mahabharata (see pp. 198ff.) provided him with two representatives of the second function: the speedy, wrestling, rough Bhlma and the courtly Arjuna. Bhlma was the son of Vayu, whom Dumezil interpreted at that time as a "dieu premier". Gods of this type, which will not be further discussed here, did not fit into the tripartite division. However, in 1956 the Indra-Vayu opposition is seen as typical of the second function. Dumezil notes that Indra is combined with various other gods. These combinations are for the most part less structured than those of the first function, but "La seule exeption pourrait etre Ia liaison que signale le compose Indravayfi"96• This composite form is the embodiment of an opposition which we encounter in Bhima and Arjuna as: "deux aspects,
95 96
Dumezil, 1940, pp.143ff. Dumezil, 1956b, p.9.
174
CHAPTER FIVE
brutal et chevaleresque, de la force guerriere...." 97• Dumezil compares this opposition with that between Heracles and Achilles. Viennent ensuite deux guerriers, mais bien diff~rents: Bhima, le fils de Vayu, , est un Hercule brutal et pas tr~s intelligent, volontiers solitaire, arm~ souvent d'une simple massue, mais surtout de sa force colossale; Arjuna fils d'Indra, est le guerrier-chevalier, chef d'arm~e. expert ~ !'arc et ~ toutes les armes classiques.98 In the Germanic material, as we have seen, there is a "glissement guerrier". Odhinn is therefore not only "dieu souverain" but also "dieu guerrier". A number of heroes are associated with him, such as Sigurdhr and Helgi. The uncouth Starkadhr is a favourite of Thorr. The heroes and their gods are opposed to one another like Indra and Vayu. In 1958 Dumezil qualifies the "heros du type Vayu" as a "sorte de bete humain", in contrast to the "heros du type Indra", who is described as "reussi et civilise"99• The list of heroes exhibiting this dichotomy is now expanded by the addition of the Iranian Sam Keresaspa, the rough fighter with the club, and his counterpart, such as Thraetona. The *Indra- *Vayu opposition is elaborated for Odhinn and ThOrr as follows. After discussing Odhinn, Dumezil notes: Nous voyons maintenant qu'il s'agit principalement (sans que le d~marcation soit rigoureuse: c'est ThOrr, comme lndra, qui reste le Dieu tonnant, le dieu de Ia bagarre atmosph~rique) de Ia partie que, chez les lndo-Iraniens, patronnait *Indra, alors que Ia partie que patronnait *Vayu est plutOt celle de ThOrr, le brutal cogneur, l'avonturier des ex¢ditions solitaires contre les g~ants. 100 The opposition between Indra and Vayu is strongly reminiscent of that between Mitra and Varuna. This did not escape Dumezil, though he does not comment on this division of the second function, but on another similar one. His analysis of this function yields the following features: "II y a deux dieux, aux noms assonants, de type voisin, et pourtant rebelles a !'association et meme a I' opposition systematique, Ind(a)ra et Rud(a)ra" 101 • This Rudra .. .introduit dans Ia deuxi~me fonction un ~l~ment inqui~tant, presque dtmoniaque, comme fait, dans Ia premi~re, le magicien et surprenant Varuna, ~ rott du juste et rassurant Mitra. 102
97 98 99
ibidem, p.17. ibidem, p.70.
100 101 102
ibidem.
Dumezil, 1958a, p.n. Dumezil, 1953c, p.S.
ibidem, p5-6.
BIPARTITAMINORA
175
In a note in the same article he remarks that, despite the differences between them, both Indra and Vayu are warriors103• In 1953 he also refers to the bipartition of the third function. This had already been introduced in 1948 as a formal bipartition. The Ashvin twins are represented in the Mahabharata by the third-function heroes Nakula and Sahadeva (see pp. 199ff.). The former is concerned with tending horses, the latter with tending cattle104• In 1953, however, Dumezil adds more material to this formal opposition 105 • The German data contain the divergent fates of Freyr and Njordhr: Njordhr receives a burial worthy of Odhinn, unlike Freyr. This opposition recurs among more of these Dioscurian pairs. In the case of the Greek Castor and Polydeuces, one is the son of a mortal and the other is the son of Zeus. Castor ends his life miserably, while his brother, who is almost in despair, is invited by Zeus to take up residence with the gods. Romulus and Remus display a similar distinction. While one of them dies dishonourably, the other appears as Quirinus after his death. The same opposition can be detected in the case of the Nasatyas: "l'un est flls de "Sumakha" (inconnu), I'autre est ftls du Ciel" 106• They differ from one another as "Ciel et la Terre" 107 and as "le jour et la nuit". The Iranian Haurvatat and Ameretat can also be distinguished in this way. HaurvaHit ( ), patron(ne) des eaux, des liquides et des liqueuers, et Ameretat ( , du conflit.51 L'~po¢e
After discussing a number of variants from the Bdihmanas in 1945, Dumezil now concludes: II y a done une tradition indienne multiforme et diversement exploit~e mais vivace et constante dans son principe, qui affirme Ia distinction et I'in~galit~ initiales d'Indra et des deva d'une part, des A~in ou Nasatya d'autre part. Et. l'interpr~tation du fait est imm~diate: elle ressort des objections d'lndra a !'admission des A~in. II s'agit d'une v~ritable in~galit~ sociale, Ies A~n repr~ntant, sous Ia brillante aristocratie divine, une vrai pl~be, habile, Iaborieuse, serviable, mais sans lustre.52 In 1947, as we saw, Dumezil shifted the emphasis from the result of the struggle to its course. The subdivision into episodes with functionally distinct methods of warfare also crops up in this conflict. The assistance which Chyavana provides for the Nasatyas is preceded by an episode in which the latter try to seduce
48 49
ibidem.
Dumezil, 1968, p.289. so ShB, IV,1,5. 51 Dumezil, 1941c, p.176. 52 Dumezil, 1945, p.45.
192
CHAPTER SIX
Sukanya, the young wife of old Chyavana, by bribery. They offer her jewels and luxurious materials, but in vain. Dumezil comments on this passage: Plus ferme que Tarpeia, plus saine aussi que dont Gullveig ... , Ia jeune femme indienne rc5siste, et Ia tentative de corruption c5choue, comme c5chouera compensatoirement, ~ l'c5pisode suivant, Ia foudre brandie par Indra pour intimider les Nasatya et leur protecteur; et, de ces deux ~bees, celui des reprc5sentants de Ia troisi~me fonction et celui du repr~ntant des fonctions su~rieures, sort finalement le pacte par lequel les Na&atya soot dc5finitivement re(_(us dans Ia soci~t~ divine ...53 In 1948 Dumezil elaborates the correspondences between the Germanic Kvasir and the Indian Mada. Kvasir was created from the spittle of the Aesir and the V anir which was collected in a single bowl at the conclusion of the peace. Kvasir roams throughout the world as a wise man until he is murdered by two dwarfs. The mixture of his blood with flour is the poetic mead which is later stolen by Odhinn. According to Mogk54, this myth of Kvasir was invented by Snorri to explain a kenning which he could no longer understand. Dumezil considers Mogk's thesis as ingenious, "Mais, a cette ingeniosite, le progres des etudes comparatives permet d'opposer des faits que ne connaissait ni ne prevoyait Mogk" 55 • Kvasir was created at the conclusion ofthe conflict between the Aesir and the Vanir. In the comparable Indian conflict, Chyavana produces the monster Mada, "ivresse", through the power of his asceticism and hereby enforces a peaceful solution. The monster is divided into the four intoxicating parts. There are many differences between Kvasir and Mada: Kvasir is created physically after the peace has been concluded, while Mada is created mystically to bring about the peace; Kvasir is quantitatively divided by the dwarfs, while Mada is qualitatively divided by the gods; Kvasir is only intolerable in the eyes of the dwarfs, while Mada really is unbearable; Kvasir is good-natured, while Mada is a monster that menaces the world of the gods. All the same, however, Dumezil remarks: II n'en existe pas mains un scMma commun: c'est au moment oil se constitue
d~finitivement Ia soci~te divine par l'adjonction des representants de Ia f~conditc5 et de Ia pros~rit~ ~ ceux de Ia souverainet~ et de Ia force, c'est done au moment oil les repr~sentants de ces deux groupes antagonistes font leur paix, qu'est suscitc5 artificiellement un personnage incarnant Ia force de l'ivresse et nomme d'apr~s elle. Comme cette force s'av~re trop grande au regard des conditions de notre monde pour le bien ou pour le mal - le personnage ainsi fabriquc5 est ensuite tuc5 et fractionnc5 en trois ou quatre parties dont ~n~ficient ou patissent les hommes. Ce ~Mma est original. On ne le rencontre, ~ travers le monde, que dans ces deux cas.
53 Dumezil, 1947a, p.288. 54 Mogk, 1923.
55 56
Dumezil, 1948c, p.lOl. ibidem, p.l04.
TRIPARTITA MINORA
193
The fact that intoxication appears at precisely this moment in the figure of Kvasir or Mada is understandable, since: ...en particulier, l'ivresse ~tant, d'une part, l'un des ressorts fondamentaux de Ia vie du pretre-sorcier et du guerrier-fauve de cette civilisation, et, d'autre part, ~tant procur~e par des plantes qu'il fallait cultiver et cuisi.ner, on comprend que Ia de l'ivresse et tout ce qui s'ensuit soit situ~e au moment de l'histoire mythique ou Ia soci~t~ se constitue par Ia r~conciliation et !'association des pretres et des guerriers d'une part, des agriculteurs et des d~positaires de toutes les puissances f~condantes et nourrici~res de l'autre.57 In 1953 Dumezil compares Kvasir with Mithotyn. They are both omniscient figures who wander until they meet their end. And in comparing both of them with Mada, he remarks in a footnote: Le correspondance indien de Kvasir, Mada, 94. The scheme can thus be expanded as follows: Figure 23: Roman "dieux souverains mineurs"9S
Dieu souverain principal
Rome Protection des juvenes et de vitalite ( duree) romaine
DiusF. Jupiter O.M. Jupiter
+ Juventas
Protection des proprietes terriennes et de l'espace romain +Terminus
Dumezil also uses this model in 1959 in order to interpret the generation of Pandu. His brothers, Vidura and Dhrtarashtra, can be seen as "souverains mineurs". Bhishma asks Vidura's advice about marriage. Dumezil comments: ...et ille consulte en lui faisant valoir que telle ou telle princesse est convenable pour , pour : ce sont Ies soucis memes d'Aryaman.96
Dumezil has the following to say on Dhrtarashtra: La
premi~re intervention de Dhrtarastra a lieu immMiatement apr~s. quand son fr~re Pandu revient de ses triomphales campagnes, charg~ de butins et de dons; il . 104 The placing of the Vishnu - Shiva opposition in an apocalyptic setting, however, is not Vedic, but dates from more recent times. In that case, the Krshna Ashvatthaman opposition and the conflict concerning Parikshit are later additions. Mais avant cette retouche, qu'y avait-il? Est-it sOr que les plus vieux Arya de l'Inde soient entrts dans leur nouveau domaine sans y apporter une eschatologie, non point Ia thtorie cyclique des Ages, mais Ia conception d'une crise cosmique unique, prtvue pour l'avenir, mettant fin a un monde mtdiocre et dtbouchant sur un monde excellent? 105 Dumezil makes a brief comparison between the events in the Mahabharata and those in the Scandinavian eschatology in order to provide an answer to this question. Although he had earlier treated the opposition between the two opponents, Arjuna and Ashvatthaman, as being of importance for the epic, Dumezil now concludes this comparison between the Indian and the Germanic material with the remark that this theme is a secondary one. A l'antantissement total du monde chez les Scandinaves rtpond, dans l'lnde, l'antantissement presque total de Ia dynastie par Ia destruction de ses espoirs, les jeunes. Pour le reste it semble que ce soit le regne idyllique de Yudhisthira, aidt de l'aveugle et du conciliateur, qui, dans l'Inde, rtponde a Ia renaissance du monde sous Ia royautt de Baldr rtconcilit avec HOdhr, le reste de l'tpopte, l'tpisode de l'embryon avec ses anttctdents et ses constquences, l'opposition de Krishna et d'Ashvatthaman et a travers eux de Vishnu et de Shiva, ttant peut..etre une adjonction posttrieure au du mains le remaniement shivaile d'une affabulation qu'il serait vain de prttendre reconstituer. 106
104 ibidem, p.218. 105 ibidem, p.220-221. 106 ibidem, p.230.
210
CHAPTER SIX
Finally, Dumezil mentions another perspective opened up by Wikander 107• The latter referred to the battle of Bravellir, described by Saxo Grammaticus, as an eschatological combat. Dumezil differs from Wikander on this point: Je ne crois pas en effet que Ia bataille de Bnivellir doive s'interprtter, comme le Mababbarata, par un mythe eschatologique historicist: l'eschatologie scandinave est bien connue et tout y est difftrent, fl commencer par le principal, le r~le et le destin d'Odhinn. 108 Although there is a resemblance between Bravellir and Kurukshetra, the correspondences are not located within the "intrigue centrale" of the Mahabharata109• There is a relationship between Bravellir and Kurukshetra on two levels. On the one hand, they can be compared with one another, a comparison which illuminates certain details in both accounts. On the other hand, the Mahabharata is an eschatological battle projected into history, but Dumezil is reluctant to view Bravellir in the same light. The same ambiguity can be seen in the lessons which Dumezil draws from the interpretation of the Mahabharata for the interpretation of the traditions connected with the early Roman republic.
The "suite romaine" After translating Wikander's article and adding his own comments, Dumezil followed this up in 1948 with a "suite romaine" containing a "perspective d'etude plus inattendue" 110, in which he dealt with the traditions associated with the emergent Roman republic. As we have already seen, Dumezil had situated two figures from this period - Horatius Codes and Mucius Scaevola - within Indo-European mythology. A third person, Cloelia, is often mentioned besides Horatius Codes and Mucius Scaevola, and there are others too who are assigned an important role by Dumezil. They lead him to wonder: "L'exegese qui a valu pour deux fragments de ce recit ne permettrait-elle pas de comprendre et les autres fragments pris una un, et l'agencement de tous?" 111 Or more explicitly: ...l'tpopte des premiers sauveurs de Ia Rtpublique romaine n'est-elle pas, autrement mais au meme titre que l'tpopte des Pandava, un expost dramatique des rapports, actions et mtrites d'un groupe de htros (plus une Mrorne)
107 108 109 110 111
Wikander, 1960. Dumezil, 1968, p.257. ibidem, p.257; Dumezil, 1971, p.132. Dumezil, 1948a, p.85. ibidem, p.86.
1RIPARTITAMINORA
211
traditionnels, incarnant chacun l'une des trois grandes fonctions ou bien une modalitt d'une des trois fonctions, suivant le schema canonique prtctdemment dtfini? 112 The opponents in this conflict are the king Tarquinius, who had been expelled from Rome, together with his non-Roman allies, the Etruscans, versus the rebellious Roman aristocracy. The direct cause of the revolt was Lucretia's
suicide after she had been raped by Tarquinius. The military aristocracy
assumes power and since then the words rex and regnum are viewed with suspicion. This has affected the traditions relating to the fust function. Cocles and Scaevola have no political power, which is vested in Valerius Publicola, the "fondateur de Ia liberte et des sages institutions qui l'assurent" 113• It is his duty to interest the populace in the new regime. He determines its origin both temporally (throughout the period he is consul, either on his own or as one of a pair of consuls) and in terms of action (he shapes the republic). "Cela reconnu, ce personnage nouveau et specifiquement romain etant mis a part, !'ordonnance du recit, l'articulation de l'equipe des sauveurs apparru"t clairement" 114 • These saviours are: -Horatius Codes and Mucius Scaevola. Horatius defends the entrance with his magic grimace. We saw that Bhima had a function as "guerrier initial" by virtue of his descent from the "dieu premier". Horatius has the same function, and .. .il accomplit cette performance A Ia porte de Rome, exactement a l'entrte dupont (Vayu was posted at the Cinvat bridge and Heimdallr at the BifrOst bridge, W.B.) qu'on coupe derritre lui, et au pied de Ia colline du dieu Janus. 115 While Yudhishthira was threatened with the loss of his hands, (see above, p. 202), Mucius Scaevola actually loses his hand. -Brutus in the military field. After his expulsion from Rome, Tarquinius asks for help from the inhabitants ofVeii and Tarquinia. He receives it and attacks Rome. Arruns, the son of Tarquinius, and Brutus kill one another in a duel. Nothing is known about the successors of Brutus, Lucretius and Horatius, except their names, but these are the names of two gentes with military characteristics. -Spurius Larcius and Titus Herminius are vital for sustaining the besieged city in two versions. In one of them, they are able to get the herds to graze outside the city. In the other, they bring ships bearing food supplies into the city. They do so in concerted action ("jumeles" 116).
112 113 114
ibidem, p.86-87. ibidem, p.88. ibidem, p.88-89. 115 ibidem, p.98-99. 116 ibidem, p.97.
212
CHAPTER SIX
The negotiations between Porsenna, with whom Tarquinius had sought refuge on the second occasion, and the Romans get under way after the action by Mucius Scaevola. Cloelia is one of a group of hostages. She sets an example by swimming across the Tiber with the group. Publicola sends her back, but her courageous deed results in a definitive peace settlement. Porsenna then frees her together with a group of young men of her choice. She chooses to take this group with her because it is in the greatest danger. Le sens de cet ~pisode est clair: it y a assaut de
loyaut~ et de moralit~ entre les chefs romains et le chef ennemi (qui, a cette occasion, se ~pare du Tarquin, cause de toute Ia guerre); et c'est une femme, Ia plus distingu~ de celles que les romains ont dO livrer, qui, par son courage, obtient du roi, avec sa propre Iibert~, celle des 6tages dont Ia Rudeur est le plus menaree, et aussi le d~cide a Ia paix d~finitive, qui sauve Rome. 7
Thus we come across the second function, using the furor, in the duel between Brutus and Arruns; the magical aspect of the first function with the typical mutilation of the eye; the third function characterised by the provision of food and the presence of "twins"; the juridical aspect of the first function with the characteristic mutilation of the hand; and finally the female action, which is associated with all three functions at this point (see pp. 226ff.). Dumezil's conclusion is: "Et peut-on raisonnablement voir, dans cette quintuple rencontre bien ordonnee, une coincidence, un jeu du malin genie?" 118• Brutus also has features which recall Ouranos and Varuna in this complex. In Dumezil's earlier work, Ouranos and Varuna were associated with the rivalry between father and son (p.164). Dumezil comments on the day on which Brutus sentences his sons after a failed coup: "ce jour-la, le dux Brutus, en tant que premier magistrat et heritier des faisceaux du rex, a bien revetu le personnage du < souverain terrible> " 119• The following diagram illustrates tripartition in the Roman republic: Figure 24: Tripartition in the Roman republic Varuna Mitra vayu
Pandu Yudhishthira Bhlma
Indra Ashvins
Arjuna Nakula and Sahadeva Draupadi'
Godess
111 ibidem, p.94-95.
118 119
ibidem, p.98. ibidem, p.99.
Horatius Codes (Brutus) Mucius Scaevola Horatius Codes (as "dieu premier") Brutus,Lucretius,Horatius (the consul) Brutus Spurius Larcius and Titus Herminius Cloelia
TRIPARTITA MINORA
213
The war between Rome and the alliance of Tarquinius and Porsenna had already led Dumezil to make comparisons (see p. 190). In 1940 and 1941 Dumezil compared this struggle with that between the Tuatha De Danann and the Fomoire.The Romans who have driven out the tyrant Tarquinius ftght against an Etruscan army which is supporting the tyrant's restoration. Codes is one of the leading defenders. The Tuatha De Danann, who have driven out the tyrant, the Fomoire Bress, ftght against the Fomoire who want to restore Bress. Lug, a one-eyed dancing ftgure, is one of the main defenders. In both cases the liberation from the tyrant plays a major part. "II est done possible que les deux legendes, l'irlandaise comme Ia romaine, soient d'anciens mythes de regifugium" 120• "Systemes de representation" 121 are linked to history, so that Dumezil can say of the war between the Romans and the Etruscans: Que nos critiques ne nous attribuent pas Ia these ridicule que les conflits , , soient : Janus" 124• Dumezil once again, in 1973, discussed the "equipe des grands Romains" who resist Porsenna. Two of them, Horatius Codes and Mucius Scaevola, had already been discussed in connection with the "borgne" - "manchot" opposition. As we saw above (p.148f.), they are both detached from the first function as far as their effective wielding of power is concerned, and this cannot be the result of colouring due to the war. Peut~tre ce couple trouverait-il une place plus prtcise si l'on considtrait dans son ensemble ce que j'appelais tout A l'heure , Ossipago ." BSL, XXXVIII (1937): p.103-112. - "Traditions indo-iraniennes sur les classes sociales." lA, CCXXX (1938): p529-549. - "La doctrine medicate des lndo-Europeens." RHR, CXXX (1945): p5-12. Bergaigne, A La religion v~dique d'apres /es hymnes du Rig-VMa, Paris, 1878-1883. Bloch, R Origines de Rome, Paris, 1946. Bolle, K. "Myths and Other Religious Texts." F.Whaling (1984): p.297-364. Bonnet, J. ed. Georges Dumezi~ Cahiers pour un temps, Paris, 1981. Brough, J. ''The tripartite ideology of the Indo-Europeans: an experiment in method." BSOAS, 22 (1959): p.68-86. Cardona, G., ed. Indo-European and Indo-Europeans, Philadelphia, 1970. Childe, V. Gordon The Aryans, London, 1926. _ Prehistoric migrations in Europe, Oslo, 1950. Christensen, A Les types du premier homme et du premier roi dans l'histoire Mgendaire des Iraniens, I, Stockholm, Leiden, 1917. - "Quelques notices sur les plus anciennes periodes du zoroastrisme, 2. Sur l'origine du pantheon gllthique." AO, IV (1926), Leiden. De Vries, J. "The problem of Loki." FFC, 110 (1933), Helsinki. _ Altgermanische Religionsgeschichte, I, Berlin, 1935. _ Altgermanische Religionsgeschichte, II, Berlin, 1937. _ Altgermanische Religionsgeschichte, I, Berlin, 2e Auf!., 1956. _ Altgermanische Religionsgeschichte, II, Berlin, 2e Aufl., 1957. Duchesne-Guillemin, J. Zoroastre, Paris, 1948. _ The Western response to Zoroaster, Oxford, 1958. _ La religion de /'Iran ancien, Paris, 1962. Dumezil, G. Le Festin d'immortalitt!, etude de mythologie compar~e indo-europeenne, Paris, 1924a. _ Le crime des Lemniennes, rites et Jegendes du monde ~geen, Paris, 1924b. - "Les bylines de Michajlo Potyk et les legendes indo-europenne de l'ambroisie." RES, V (1925a): p.205-237. - "Soukhmanti Odykhmantievitch, le paladin aux coquelicots." M~langes Paul Boyer, Paris (1925b): p.280-288. - "Les fleurs Haurot-Maurot, et les anges Haurvatllt en Ameretllt." REA, VI (1926): p.43-69. _ Le probleme des Centaurs, ~tude de mythologie compar~e indo-europeenne, Paris, 1929. _ L~gendes sur les Nartes, suivis de cinq notes mythologiques, Paris, 1930a. - "La prehistoire indo-iranienne des castes." lA, CCXVI (1930b): p.109-130. - Ouranos-Varuna, ~tude de mythologie compar~e indo-europeenne, Paris, 1934. _ Flamen-Brahman, Paris, 1935. - ''Temps et mythes.", RPh, V (1936): p.235-251. _ "Le crime de Sisyphe." TCF (1937): p.156-162. - "Laprehist