R i t u a l a n d M a n t r a s :

Rules Without Meaning

R i t u a l a n d M a n t r a s :

R u l e s W i t h o u t M e a n i n g

FRITS STAAL

MOTILAL BANARSIDASS PUBLISHERS

PRIVATE LIMITED • DELHI
 First Indian Edition: Delhi, 1996

Published by arrangement with Peter Lang Publishing, Inc.

© Peter Lang Publishing, Inc., New York 1990, 1993
All Rights Reserved. Reprint or reproduction, even partially,
in all forms such as microfilm, xerography, microfiche,
microcard, offset strictly prohibited.

First published as Rules Without Meaning:

Ritual, Mantras and the Human Sciences

ISBN: 81-208-1411-8 (Cloth)

ISBN: 81-208-1412-6 (Paper)

Also available at:

MOTILAL BANARSIDASS
41 U.A. Bungalow Road, Jawahar Nagar, Delhi 110 007
120 Royapettah High Road, Mylapore, Madras 600 004
16 St. Mark's Road, Bangalore 560 001
8 Camac Street, Calcutta 700 017
Ashok Rajpath, Patna 800 004
Chowk, Varanasi 221 001

FOR SALE IN INDIAN SUBCONTINENT ONLY

PRINTED IN INDIA
BY JAINENDRA PRAKASH JAIN AT SHRI JAINENDRA PRESS,
A-45 NARAINA, PHASE I, NEW DELHI 110 028
AND PUBLISHED BY NARENDRA PRAKASH JAIN FOR
MOTILAL BANARSIDASS PUBLISHERS PRIVATE LIMITED,
BUNGALOW ROAD, DELHI 110 007
For Wichai
 Contents

Preface xiii
A note on the pronunciation of Sanskrit xxi

Part I. Methods, Meanings and Rules

Introductory Note 1
1 Meanings and Rules in the Human Sciences 3
2 The Empiricist Caricature of Science 7
3 The Positivist Critique of Meaning 17
3A. Ludwig Wittgenstein 21
3B. Martin Heidegger and Hans-Georg Gadamer . . 25
4 Linguistic Background 33
5 The Origins of Linguistics in India 37
6 Contemporary Linguistics and Rules 47
6A. Noam Chomsky .. 52
6B. Chomsky versus Wittgenstein 55

Part II. Ritual

Introductory Note 61
7 Vedic Ritual 65
8 The Agnicayana Ritual. 71
9 Basic Rites 79
9A. The Isti 79
9B. Soma Sequences 81
10 The Structure of Ritual 85
x Contents

11 Embedding and Transformation . 91

11A. The Kindling Verses 91
1 I B . From Agnistoma to Agnicayana 92
11C. The Final Sastra for the As vin Twins 94
11D. Conclusions 94
12 Ritual and G r a m m a r 101
13 Interpretations of Ritual 115
13A. Indian Interpretations 117
13B. Western Interpretations 122
13C. The Meaninglessness of Ritual 131
14 Anthropology without Asia 141
14A. Raymond Firth 143
14B. Victor Turner. 152
15 Syntax, Semantics and Performatives 157
16 Music and Ritual 165
16A. Seven Musical Structures 165
16B. Seven Ritual Structures 178
16C. Conclusions 182

Part III. Mantras

Introductory Note .' 191
17 Six Vedic Mantras 199
18 Ritual Context 201
19 The Syntax of S t o b h a s . 209
20 Vedic and Tantric Mantras r.... 223
21 Performatives, Pragmatics and P e r f o r m a n c e . . . . . . . . . 237
21 A. Performatives Again 237
21B. Pragmatics 242
21C. Performance and Competence 245
22 Mantras and Language 253
23 Mantras and Bird Songs 279
24 Mantras by Chance 295
 Contents xi

25 Anthropology without Mantras 313

25A. Object Semantics: Stanley Tambiah . . 314
25B. Psycho-Semantics: Gananath O b e y e s e k e r e . . . 320
25C. Thick Semantics: Clifford Geertz 326

Part IV. The H u m a n Sciences

26 The Science of Ritual 349
27 Oral Traditions. ! , 369
28 Religions 387
28A. Seeking Religion in Asia 387
28B. Seeking out the Buddha 406
28C. Mythology 416
28D. Philosophy of Religion 418
29 The Myth of the T w o Cultures 421
30 Conclusions 433
30A. The Origin of Language 433
30B. Michel Foucault 441
30C. Concluding Generalities 447
E X C U R S U S : "Writing in the S k y " — H e r m e n e u t i c s and
Astrology: A Neoplatonic Parallel 455
Bibliography '. 459
Index of N a m e s . 481

FIGURES
1. The Ritual Enclosure 71

TABLES
1. The Sâmidhenï Verses 92
2. Sequence of Rites on the Third and Fourth Days 93
 Preface

Philosophers tell us that no sharp boundary can be drawn

between innate knowledge and knowledge based upon experi-
ence. This is not surprising for knowledge that is innate in
individuals may have been acquired by the species over a period
of time. This description applies not only to animal instinct, but
also to knowledge of the rules of ritual and mantras that are
studied in this book. Often innate and unconscious, they lead a
life of their own, independent of religion, society and language.
But how can ritual and mantras be governed by rules that lie
beyond the human sciences and the study of religion?
It took a long time before I recognized the nature of this
question and its many puzzling overtones. For more than a
decade,,! had been engaged in the study of the 1975 performance
of the Vedic Agnicayana ritual by Nambudiri brahmins in South
India. The result was published in 1983 in two illustrated
volumes entitled AGNI—The Vedic Ritual of the Fire Altar. In
the course of this work, it dawned upon me that these data were
extraordinary, and that no one knew how to account for them.
Relevant specialists—Indologists, anthropologists, psycholo-
gists, or scholars of religion—offered no conceptual tools that I
could use. If I wanted to go beyond mere description and offer
explanations, I had to strike out on my own. My efforts are not
completed but there is enough for others to continue, with or
without me.
Although the core of this book is the discussion of ritual and
mantras in parts II and III, I stumbled upon related issues that
could not remain unmentioned. It now looks to me as if we have
been skirting the tips of an iceberg which is almost entirely under
xiv Preface

water. But can we be certain it is an iceberg we have been

skirting? Are we above or below the surface, and surface of
whatl When I don't know my way about, am I climbing, diving,
or riding waves that elude identification? Or am I on a summit,
with distant views hidden behind thick and near-permanent
clouds? Part IV shows that the landscape or seascape remains
mysterious. The compasses and maps displayed in Part I may
not be reliable. What seems certain is that we have to put
together what is not generally put together.
Ritual and mantras have been strangely neglected in the
theoretical sciences. Socrates questioned everything, from
Greek grammatical usage to the existence of Gods. He paid
attention to dreams and oracles, but not to ritual. When Plato
tells us that Socrates sacrifices a cock or a hen, at home or on an
altar of the state, he performs this rite automatically, without
recognizing the need for an explanation, much as we would
switch on our favorite channel.
" M a n t r a s " have also been ignored; outside the counter-
culture, the term is not even English. Although there are seventy
million of them according to a Sanskrit text, their forms and uses
are totally unexplained. The eighteenth century Danish mission-
ary Bartholomaeus Ziegenbalg discovered a few in Malabar on
the south west coast of India. Most Indian mantras originated in
the Vedas and have since become virtually indestructible. They
are not yet fashionable like the silk road, used by imperial China
to export, in addition to silk, porcelain, lacquers, furs, cinna-
mon, vermilion and tea. But what was the emperor of China
content to receive in return for these riches? Apart from
marihuana and Ganges water (still smuggled to Mongolia),
primarily one commodity: mantras. Unchanging in ever chang-
ing contexts of language, religion and society, generally kept
secret and guarded jealously, mantras travelled from India not
only throughout Southeast Asia and Indonesia, but crossed the
Himalayas from Kashmir into Central Asia and China (first
century A . D . ) , went on to Korea (fourth century) and Japan
(sixth century), and ascended the Tibetan plateau (eighth cen-
tury). They have now reached California where they fetch high.
 Preface xv

prices. All transactions take place behind a smoke-screen of

labels such as " H i n d u , " " B u d d h i s t , " " T a n t r i c , " " T a o i s t , "
and, belatedly, " s h a m a n , " labels originating from the outside.
Ritual and mantras are still basic features of the civilizations of
Asia. The only American response is glossolalia.
The Frontispiece of this book (photograph by Adelaide de
Menil) illustrates one of its main theses: that ritual is transmitted
not only without meaning, but often without language. Cheru-
mukku Vaidikan Vasudevan Akkitiripad (a title he acquired by
performing the Agnicayana more (han thirty years ago) shows
his son Vallabhan how to perform the Nihnava rite. This
mysterious rite is performed twice daily, in the morning and
afternoon, during several consecutive days. The ritual patron
and several of his priests place their hands on a bundle of grass,
which lies to the west of the altar. The position of the hands is
different in the two surviving traditions. According to the
Kausïtakins, it should be as follows:

Morning: left hand palm down; right hand

on left hand with palm up;
Afternoon: left hand palm up; right hand
on left hand with palm down.

The Taittiriya tradition calls for different positions:

Morning: left hand palm up; right hand

on left hand with palm down;
Afternoon: right hand palm up; left hand
on right hand with palm down.

The Plate illustrates the first of these positions: the father is

about to place his right hand on the left. The rite is mentioned
below (pages 74, 85, and 93), described in my book A G N I (I:
358-59, 366-67) and discussed by Pamela MacFarland in "Ritual
Language Dismantled" (to be published in Le rituel. Actes du
colloque du centenaire de L'Ecole Pratique des Hautes Etudes,
Section des Sciences Religieuses).
xvi Preface

Rules without Meaning is not only based on A G N I but is also

related to three of my earlier books and four that are more recent
or about to appear. Ritual and mantras are independent of
religion in ways similar to mysticism, the subject of Exploring
Mysticism: A Methodological Essay (University of California,
1975). Ritual and mantras were studied in ancient India as
described in The Science of Ritual (Bhandarkar Oriental Re-
search Institute, 1982). These studies were connected with
Indian linguistics which is the subject of my Reader on the
Sanskrit Grammarians (MIT, 1972). Parts I and IV include
material touched upon in Over zin en onzin in filosofie, religie en
wetenschap (Meulenhoff, Amsterdam, 1986). Chapters 4
through 6 are supplemented by Universals: Essays in Indian
Logic and Linguistics (University of Chicago Press, 1988a)
which Chapter 5 overlaps slightly. Some portions of the book
correspond to Jouer avec le few. Pratique et théorie du rituel
védique {Publications de Vinstitut de civilisation indienne, Paris,
1989). Chapter 28C should be supplemented with Kailas. Center
of Asia (University of Chicago Press, 1988b).
The materials of Parts II and III have been approached from
various angles in a series of special studies (for details, see the
Bibliography). While Chapters 7 through 11 derive from A G N I ,
an early version of Chapter 12 appeared under the title "Ritual
S y n t a x " in the Ingalls Festschrift of \9S0 (Sanskrit and Indian
Studies, edited by M. Nagatomi, B. K. Matilal, J. M. Masson
and E. C. Dimock, Jr., and published by D. Reidel Publishing
Company, Dordrecht). Parts of Chapter 13 appeared in " T h e
Meaninglessness of Ritual" (N'umen. International Review for
the History of Religions 1979, edited by M. H e e r m a van V o s s ,
E. J. Sharpe and R. J. Z. Werblowsky, and published by E. J.
Brill, Leiden) and parts of Chapter 14 in " R i t e s That Make N o
S e n s e " (The Communication of Ideas: 10th International Con-
gress of Anthropological and Ethnological Sciences, edited by
J. S. Y a d a v a and V. Gautam in 1980, and published by Concepts
Publishing C o m p a n y , N e w Delhi). Chapters 15 and 16 include
material from " T h e Search for Meaning: Mathematics, Music
and R i t u a l " (American Journal of Semiotics 1984, edited by
 Preface xvii

Thomas Sebeok). A related version was published by Elemire

Zolla in Conoscenza Religiosa (1983) under the title: " R i t m o nel
r i t o . " Portions of Chapters 18, 20, 21A and 22 overlap " V e d i c
M a n t r a s " , edited by H. P. Alper for Understanding Mantras
and to be published in 1988 by the State University of N e w York
Press, Albany. Chapter 19 appeared under the title " M o o n
Chants, Space Fillers and Flow of Milk" in the Sreekrishna
Sarma Felicitation Volume, Surabhi, printed in 1983 at Kalak-
shetra Publications Press, Tiruvanmiyur, Madras. Chapter 23
corresponds to an article of that title in the Journal of the
American Oriental Society (1985), edited by Ernest Bender.
Chapters 24 and 25 share parts with " T h e Sound of Religion"
also published in Numen (1986) and forthcoming in a different
form in the volume of essays in honour of Milton Singer, edited
by Michael M. Ames and Murray Leaf and to be published by
the Centre for South Asian Studies at the University of Toronto.
Chapter 26 is extracted from my monograph of 1982 of that
title, already mentioned. Chapter 27 includes fragments from
' T h e Fidelity of Oral Tradition and the Origins of S c i e n c e "
{Mededelingen der Koninklijke Nederlandse Akademie van We-
tenschappen, Afd. Letterkunde, Nieuwe Reeks, Deel 49, N o . 8,
published by the North-Holland Publishing Company, Amster-
dam/Oxford/New York 1986). Chapter 28 incorporates material
from "Substitutions de paradigmes et religions d ' A s i e " {Cahiers
d'Extrême Asie 1985, edited by Hubert Durt for the Ecole
Française d'Extrême-Orient, Section de Kyoto). Chapter 30A
corresponds to parts of a contribution entitled "Ritual, Mantras
and the Origin of L a n g u a g e " to the Professor R. N . Dandekar
Felicitation Volume Amrtadhärä, edited by S. D. Joshi and
published by Ajanta Publications, Delhi (1984).
I am grateful to the editors and publishers of these publica-
tions for permission to use them in the preparation of this book.
Fragments have also been presented in various stages of com-
pletion or incompletion to audiences at Amsterdam, Berkeley,
Berlin, Haverford, Kyoto, Leiden, Madras, Nanterre, N e w
Delhi, Oxford, Paris, Santa Barbara, Sydney, Utrecht, and other
places. At the University of California, Berkeley, my courses on
xviii Preface

the philosophy of religion and on the philosophies of India have

reflected phases of the development of my thinking.
I am grateful to friends and colleagues who have commented
on earlier drafts and without whom errors, mistakes and mis-
judgments would be more numerous and glaring. Paul Attinello,
Hubert Durt, Robert Goldman, Mark Juergensmeyer and Pa-
mela MacFarland have commented on the published or forth-
coming articles mentioned which acknowledge their assistance.
I am especially grateful to Morris Halle who read an earlier
version of Chapters 4 through 6 and explained to what extent I
was out of touch with recent developments in linguistics—an
ignorance, I am afraid, that is not yet cured. My principal debt
is to Harold Arnold who read the pre-final version in its entirety,
corrected mistakes and gave me the benefit of his judicious
comments on matters of detail and substance.
Contemporary technology has replaced the support of secre-
taries and the pleasure of thanking them. But new developments
bring new pitfalls. When I was preparing the bibliography, with
temperatures soaring in the upper eighties, my printer broke
down. The work was completed owing to prompt assistance
rendered by Rand Morimoto and Scott Tachiki of Computer
Options, to both of whom I express my thanks.
My final gratitude goes to Donald Wiebe who invited me to
write this book for the Toronto Studies in Religion of which he
is the General Editor. A similar book addressing another audi-
ence could have been written in a different vein. The data are
sufficiently rich for the accent to have been placed on philoso-
phy, language, music, ethology, sociobiology, semiotics, cogni-
tive psychology, the sociology of knowledge, the future, or
something else more directly related to the present needs of
humanity. This might have resulted in a different book, more
readable, relevant and welcome. N o doubt, a book like this
should be written before long. My book has followed its own
course but the direction it has taken suggests that I am the
person least suited to have been its author. I am not a student of
religion, anthropologist, or specialist in any of the fields of study
I have been foolhardy enough to comment on. I take comfort
 Preface xix

from Arthur Waley's translation—a Bloomsbury version—of the

twentieth chapter of L a o T z u ' s Tao Te Ching:

I droop and drift, as though I belonged nowhere.

The world is full of people that shine;
I alone am dark.

All men and women huddle together in religious communication.

I alone am outdated, without context, abstracting formulas from
thick description, pressing ancient wisdom only to be left with
bare b o n e s .

All men have enough and to spare.

I alone seem to have lost everything.
All men can be put to some use.
I alone am untractable and boorish.

September 15, 1987 FRITS STAAL

 A note on the pronunciation of Sanskrit

T H E Sanskrit vowels, pronounced as in Italian or Japanese, are

short (a, i, u, r, I) or long, viz. twice as long as the short ones (ä,
ï, ü, f, e, o, ai, au). The consonant r is pronounced as in Italian,
while the vowels r and / contain a vocalic element similar to the
short •/ in the third syllable of English ability. Among the
c o n s o n a n t s , there are a scries of dentals (/, th, d, dh, n, s) and a
corresponding series of retroflexes or cerebrals ((, th, d, dh, n,
s). The former are produced when the tip of the tongue touches
the teeth (as in French), the latter when it is bent backward and
touches the palate (the English pronunciation of/, d, n, s is more
or less in between). In the case of Sanskrit s and s, the tongue
points in the same directions, but breath passes over it. The
palatals are pronounced when the tongue touches the front of the
palate: c as in English chairJ as in jar, n as in Spanish, and s as
in sheet. The aspirates (kh, gh, chjh, th, dht th, dh, ph, hh) are
pronounced with a clearly audible release of breath following the
consonant: th like in ant-hill, dh like in bald-head, etc. The m
expresses nasalization of the preceding vowel, and the h sounds
like an h followed by a short echo of the preceding vowel (agnih
as agnih1). The other consonants are pronounced as in English.

Unlike in English, there is in Sanskrit a différence between the

stem of a noun (e.g. dhyana-, dhïti-, karman-) and the Nomina-
tive case, which is used when the noun functions syntactically as
the subject of the sentence (e.g. dhyänam, dhîiih, karma). I
generally refer to nouns by using their stem form, but some of
the sources I quote use the Nominative case form.
 P a r t i

Methods, Meanings and Rules

Introductory Note

This book proposes a series of related hypotheses that help to

understand and explain ritual and mantras. Part I introduces the
approach I have adopted and its philosophical and linguistic
presuppositions. This approach is scientific in t h a t it is based
upon the assumption that ritual and mantras can be studied like
other objects and are not beyond the pale of an investigation that
is empirical and rational and therefore akin to science. Adopting
a scientific approach does not imply that ritual and mantras can
be fruitfully or interestingly studied with the help of, for exam-
ple, quantitative, statistical or behavioristic methods. I assume,
simply, that the data can be isolated and set forth clearly after
which rational hypotheses can be formulated to account for
them. The hypotheses I present are concerned with a variety of
other topics, including religion, mythology, language, music,
and ritualization among animals. Part IV places the results in a
wider perspective and discusses the status of a human science
under which they fall.
Ritual and mantras have been studied by textual scholars,
anthropologists and scholars of religion. All contribute essential
data but none have explained why people engage in ritual
activities and chant, recite, or meditate on mantras. Scholars of
religion have reacted to unsuccessful positivistic and other
reductionist suggestions by advocating self-proclaimed non-
scientific methods, like those inspired by philosophical move-
ments such as phenomenology or hermeneutics. These methods
describe and interpret, but they do not seek to explain. Among
the reasons given for the need of such special methods is the
2 Methods, Meanings and Rules

circumstance that man is different from everything else and that

man studying himself as a subject is different from man studying
anything else as an object. We shall take a closer look at the
reasoning that lies behind these assumptions in Part IV.
Promising results, mostly in the area of human language, have
been reached in the human sciences without resorting to the use
of such unscientific methods. N o definite conclusions have been
reached in the areas of ritual and mantras, but the reason is that
these have not been studied within such a perspective. I believe
that a promising point of departure for the study of our topics lies
in the human sciences. This belief is not an unshakeable or
metaphysical conviction; it is a matter of method, that is, of trial
and error. N o r does it imply that we must follow some existing
approach within the human sciences. If we place ritual and
mantras within the wider domain of human activities and com-
petences, and relate them to other features and characteristics of
the human animal, we have gained some insight and perspective
and are perhaps on our way to a more adequate understanding.
 Meanings and Rules in the H u m a n

Sciences

In the analysis of ritual and mantras, two concepts play an

important part: meaning and rules. To understand these con-
cepts is not easy since they have been ceaselessly discussed and
continue to be controversial in modern philosophy and linguis-
tics. The six chapters of Part I attempt to sketch this two-fold
background in an abbreviated and simplified manner. Although
this will necessitate some detailed discussion, my only purpose
is to explain well-established results that are relevant to the
investigation into ritual and mantras. F o u r topics are of partic-
ular relevance and their rough outline will provide some idea of
the direction in which the argument will move us.
First, we shall learn that a characterization of the scientific
method, first given by philosophers—specifically, empiricists
and early positivists—and so widely accepted outside their
circles that it is now a common assumption, is a caricature that
represents only a few of the methods that scientists actually use.
This observation is not confined to meanings and rules, but is
pertinent to their study.
Second, we shall see why the home of meaning is language.
That is, " m e a n i n g " is a concept that applies primarily to certain
features of language, and only derivatively, or metaphorically,
to other things. Whether we should say that, within the domain
of language, meaning pertains to " u t t e r a n c e s , " " s t a t e m e n t s " ,
" p r o p o s i t i o n s , " " s e n t e n c e s , " or something else, is a topic that
has been much discussed and is worth discussing; but it is not
4 Methodsy Meanings and Rules

our problem. We shall see that logicians have related meaning to

truth and that philosophers have tried to go beyond this.
Third, we shall learn that " m e a n i n g , " although primarily
linked with expressions of language, need not automatically be a
property of all members of that class: for some expressions of
language are meaningless.
The first three topics, taken together, may be described as a
thumbnail sketch of what is living and what is dead in the much
maligned philosophy of positivism.
Fourth, we shall see that the notion of " r u l e , " which is
primarily used in logic and linguistics, is basic for the under-
standing of many human activities, not only in those restricted
areas where " t h e rules are the g a m e ; " that rules are primarily
descriptive and only derivatively prescriptive; and finally, that
recent philosophic scepticism with regard to their use and
understanding need not be taken too seriously.
Since " m e a n i n g " is primarily discussed in philosophy and
" r u l e s " in linguistics, I shall start with philosophy and deal with
linguistics afterwards. I apologize to the practitioners of both
disciplines for they will probably find much that is not to their
liking. And much in this account must remain sketchy. My aim,
however, is not to provide introductions to philosophy or
linguistics, but to give the background needed to prevent mis-
understandings of the explorations that follow. The hypotheses
I shall formulate are, like all scientific hypotheses, amenable to
improvement and modification. They will need to be checked
with other data than that I have considered, and be subjected to
further conceptual analysis and rational argumentation. My
reference to linguistic concepts has led some scholars to char-
acterize my methods as "linguistic." This is misleading for I
shall show that neither ritual nor mantras are languages: hence,
linguistic methods in the strict sense are unlikely to be helpful.
My approach shares with linguistics the recognition that the
object of study is concerned with rules and, though profoundly
human, is amenable to scientific study.
Finally, a remark about terminology. The English uses of
" s c i e n c e " and "scientific" are different from the corresponding
 Meanings and Rules in the Human Sciences 5

uses in German, Dutch, or French. In English, these terms are

chiefly applied to t h e ' ' ' e x a c t " sciences, that is, the mathematical
and natural sciences. In German, Wissenschaft expresses an
4
' e x a c t " and yet more flexible concept and the same holds for
the Dutch wetenschap and the French science. N o n e of these
European terms exclude the humanities. It is ironic that many
American scholars of religion have adopted continental philos-
ophies such as phenomenology and hermeneutics that are criti-
cal of science whereas many European scholars of religion
continue to refer to their discipline as a science. T o use the term
" h u m a n s c i e n c e " in English may suggest a controversial nov-
elty: an application of scientific methods to the study of the
human animal. In other languages, the expression " h u m a n
s c i e n c e " raises no eyebrows.
 T h e Empiricist Caricature of Science

Empiricism may be defined in the words of Thomas Aquinas,

who accepted this view and attributed it to Aristotle, as the
doctrine that nothing is in the intellect which was not previously
in the senses. It is contrasted with rationalism, which may be
defined in various ways but always stresses that some knowl-
edge is i n n a t e . - F o r Plato, rationalism was connected with
certainty: it meant that sense knowledge (because of the change-
ability of the world of the senses) lacks the certainty that rational
knowledge (for example, mathematical knowledge) possesses.
For Kant, who tried to combine empiricism and rationalism, all
knowledge starts with knowledge of the senses but it does not
start from that knowledge.
If the empiricist doctrine is correct and all knowledge starts
from the senses, the question arises as to how more general
knowledge is acquired. The obvious answer seems to be:
through generalization. David Hume seems to have subscribed
to this view: according to him, every general hypothesis is a
generalization from observed instances. Such a generalization is
often referred to as induction. According to John Stuart Mill,
even mathematical knowledge, which most philosophers regard
as deductive, is based upon induction for it consists of very
highly confirmed generalizations from experience. Early positiv-
ists, such as Auguste Comte, espoused similar views.
The question whether induction is justified has plagued many
philosophers. I am inclined simply to follow Plato's lead and be
satisfied that empirical knowledge is never or rarely certain. The
exceptions generally contain conceptual terms, regard particu-
lars, or both: for example, "Paris is the capital o f ' F r a n c e . "
8 Methods, Meanings and Rules

Nelson Goodman (1955, Chs. 3 and 4) has argued that induction

can never lead to certain knowledge by introducing a predicate
grue which means "blue before time / and green after time f."
N o w if all topazes found before time / were found to be blue,
were they blue or grue? And what about the question: is the next
emerald to be examined after time t, grue or green? We cannot
give absolutely certain answers to these questions. But we can
say this: grue is an artificial and far-fetched predicate (Goodman
calls it: " l e s s well-entrenched"), and blue and green are not.
Therefore, it is more likely that the topazes were blue and that
the emerald will be green. These answers are probably correct
and I find them satisfactory provided we remember that quite a
few improbable statements have turned out to be true.
The view that general knowledge arises through generalization
from observed instances is untenable. Why did such a view arise
at all? To answer this question we have to go back to the
beginnings of Western philosophy. The philosophical theories of
Plato and Aristotle were based upon a double reflection on the
place of man in the universe and the data of contemporary
science. The latter comprised in particular mathematics and
physics as they were being developed by Greek scientists and by
the philosophers themselves. Thus Plato's theory of ideas or
forms was related to mathematical notions on the one hand and
conceptual analysis on the other. During the Western middle
ages, further logical and conceptual precision was gained, but
respect for the data of science declined as is shown by the fact
that scientific knowledge beyond what was known to the ancient
Greeks was rarely taken into account. 1

1
This is not an overstatement. It may be illustrated by the relatively simple
example of geography. Greek scientific cartography reached its greatest height
with Ptolemy (second century c.E.). After Ptolemy, map making was replaced
by religious cosmography that had no basis in fact. Only after 1300 did
geography begin to develop again. Joseph Needham compared this Western
development with the history of geography in China: at first, the Chinese had
nothing of the quality of Ptolemy (or even Herodotus or Strabö) at times
contemporary with them; but during the gap between the third and thirteenth
centuries, when European learning stood still, the Chinese were progressing
 The Empiricist Caricature of Science 9

Modern philosophy tried to make a new beginning and re-

establish its links with science. But the sciences, especially the
natural sciences, went their own way and became rapidly
independent of philosophy. Descartes and Leibniz were among
the last Western philosophers who were creative scientists.
After them, adequate knowledge of the mathematical and phys-
ical sciences became virtually impossible for any non-specialist
to attain. Philosophers remained conversant with the humanities
and the history of philosophy for another century or t w o .
Empiricists and positivists were right in emphasizing that
philosophy could not make any progress unless it took into
account what knowledge the sciences provided. But they went
wrong in two respects: they emphasized only the natural sci-
ences, whose successes had been the most spectacular, and
attributed to these the "scientific method of generalization" that
was a product of their imagination. This attribution was due to
the very gap between philosophy and sciences, the existence of
which they decried. Their caricature of the scientific method was
widely adopted. Many critics of science—phenomenologists, for
example—based their criticism upon it.
The philosopher and scientist Alfred N . Whitehead pointed
out that many scientific discoveries result from imaginative
speculation that does not begin, but ends with observation. H e
illustrated this with the discovery of the planet Pluto " i n order to
avoid the suspicion of biased selection," since Pluto was the
most widely publicised recent discovery at the time of his writing
(1930: Whitehead 1933, reprint 1942:150). I shall instead discuss
the discovery of Neptune which is even more apposite.
Hegel had stated a priori that there were seven planets shortly
before the eighth, Neptune, was discovered. Legend has it that
he reacted to the discovery when it was reported to him by
saying: Umso schlimmer für die Tatsachen ( " s o much the worse
for the facts"). The discovery of Neptune was connected with
the discovery of irregularities in the orbit of the planet Uranus

steadily and toward the end of that period were far ahead of the Europeans
(Needham 1959,111:512-532; cf. also Staal 1988b).
10 Methods, Meanings and Rules

that led Bessel and other astronomers to postulate the existence

of a hypothetical center of gravity by which they could be
explained. Two astronomers calculated its probable position:
J. C. Adams in Cambridge and U. J. J. Le Verrier in Paris.
Challis examined the relevant photographic plates and saw a
speck of light in the area of the heavens in which Adams had
predicted the planet would be found, but did not recognize the
planet. J. G. Galle in Berlin detected it much later (in 1846) at a
short distance from where Le Verrier had predicted it would be
found. Thus Neptune was discovered.
It is clear that the discovery of Neptune was neither in line
with Hegel's pronouncements, nor in accordance with the
empiricist and early positivist account of how science works.
For if "generalization from e x p e r i e n c e " were all that astrono-
mers aspired to, they should have been satisfied with the
description of the orbit of Uranus and would not postulate
centers of gravity elsewhere. True, Uranus' orbit looked irreg-
ular from the point of view of a specific scientific expectation,
and was a deviation in those terms. But the formula describing it
was adequate and elementary. As Whitehead remarked (1952:
153), " t h e r e is a motive of unrest which urges scientists beyond
mere satisfaction with the simple description, beyond even the
general description. It is the desire to obtain the explanatory
description which may justify the speculative extension of
L a w s , beyond actual, particular instances of o b s e r v a t i o n . "
Recently, philosophers of science such as Thomas Kuhn and
Paul Feyerabend have argued that scientific discoveries do not
take place in any regular fashion. Whitehead already knew that;
he stated clearly what Kuhn reiterated and is now widely
accepted:

The advance of any reasonably developed science is twofold. There is

the advance of detailed knowledge within the method prescribed by the
reigning working hypotheses; and there is the rectification of the working
hypothesis dictated by the inadequacies of the current orthodoxy
(Whitehead 1942:258).
 The Empiricist Caricature of Science 11

This view of science is supported by the history of science not

only in the West but also in India (see Staal, 1988a). The
empiricist-positivist account of scientific method is one-sided,
limited, and generally incorrect.
Another widespread error is that science contradicts the
" m e t a p h y s i c a l " distinction between appearance and reality that
Bertrand Russell referred to as " o n e of the distinctions that
cause most trouble in philosophy." He was thinking, probably,
of Hegelians such as F . H. Bradley whose " A p p e a r a n c e and
Reality" of 1893 was influential in philosophical circles in Britain
when he was young. The distinction between appearance and
reality, however, is basic to scientific analysis. The most inter-
esting results in science are hypotheses about the existence of
entities such as elementary particles or fields that are not
directly observed but are assumed to be " r e a l " in order to
account for " a p p e a r a n c e s " that are observed such as tables,
movements and other events that take place in the macroscopic
universe.
Squeamishness about unobservables, then, comes not from
scientists but from philosophers. Since natural scientists went
about their work without paying attention to philosophy, mis-
taken ideas about scientific method had little effect on them. But
these same ideas had a devastating effect on the humanities and
the social sciences that were groping for a methodology and
therefore paid attention to philosophy. As a result, empiricist
and positivist phobias and strictures have prevented progress in
these disciplines. We shall examine in this regard Emile Durk-
heim's account of religious phenomena which does not include
mysticism because it was regarded as based upon unobserv-
ables. The situation in linguistics at the beginning of the century,
and in some parts of psychology even at present, is similarly
restrictive. Bloomfield, for example, confined himself to utter-
ances that are observed and defined language as " t h e totality of
utterances that can be made in a speech c o m m u n i t y , " a mis-
taken view abandoned by the Indian grammarians more than two
thousand years ago (Chapter 5). Behaviorism is still favored by
many social scientists who are under the impression that it is
12 Methods, Meanings and Rules

"scientific." Chomsky observed that ä profitable debate on the

subject has become difficult: " T h e behaviorist position is not an
arguable matter. It is simply an expression of lack of interest in
theory and explanation" (1965: 193 note 1).
The following accounts of ritual and mantras are not restricted
by these empiricist and positivist prohibitions; their aim is the
opposite, namely, to pave the way for adequate and interesting
theories and explanations. That such explanatory machinery
may be unobservable is a fact about theory formation. Ritual and
mantras can only be accounted for when unobservables are
taken into account. Even a description in terms of structures
goes beyond the linear concatenation that is observed (Chapter
12). Sometimes the data themselves are unobservable: some
mantras are not recited or chanted, but gone through mentally.
Priests put down bricks in a specific order which is " i n their
h e a d . " Some rites and recitations can only be adequately
described by taking into account what has been omitted (Chap-
ters 11C, 27). This is not behaviorism even in the extended sense
used by Quine and others which includes "dispositions to
behavior."
It is instructive to see what physicists do as distinct from what
philosophers say they do. I shall discuss examples taken not
from contemporary physics, which may be uncharacteristically
imaginative, but from the entrenched areas of classical physics.
Georg J o o s ' "Theoretical P h y s i c s " of 1934 introduces electro-
magnetism in the following terms:

Two différent methods are employed in the theoretical treatment of

electrical and magnetic phenomena. The first method starts with macro-
scopically measurable quantities and describes mathematically the rela-
tionships between such quantities found by experiment. An enormous
range of phenomena may be brought within the scope of this method in
the form of a system of differential equations. The integration of these
equations for particular cases then permits us to answer, by rigorous
calculation, a large number of questions, including many of technical
importance. This method employs only experimental laws and their
mathematical consequences; it is therefore entirely free from hypothesis
and cannot come into conflict with experience. Since it makes no
assumptions whatsoever concerning the structure of electrical charges or
 The Empiricist Caricature of Science 13

of matter, we call this the continuum theory, or the field theory.

Nevertheless, it was soon recognized, that a large number of very
striking phenomena, e.g. those of electrolysis, are entirely left out when
this point of view is taken. In order to explain these phenomena, we must
make assumptions concerning the structure of matter and of electrical
charges. We cannot verify these assumptions by direct observation, but
only by testing their consequences. This second method will be called
the atomistic method. While the continuum theory will be treated first,
we shall nevertheless not attempt to exclude all reference to atomistic
concepts; on the contrary, we shall occasionally avail ourselves of a
side-glance at the atomistic picture of the phenomena for greater
vividness (Joos 1947 reprint:249).

After 200 pages filled with electro-magnetic equations and their

discussion, Joos introduces the theory of heat:

The Theory of Heat, like the Theory of Electricity, may be developed

from two different points of view. The phenomenological method em-
ploys only concepts like temperature, quantity of heat, &c, which are
taken from the macroscopic world of observations and which can be
measured directly. The laws thus obtained have the advantage of being
free from hypothetical assumptions. On the other hand, in the Theory of
Heat, e.g. in connexion with the Law of Entropy, we feel the need of a
deeper 'explanation.' Such an explanation, deeper because more vivid,
is furnished by the atomistic, statistical view. Here again, as in the
electrical case, we find that although we develop the theory separately
from the two sides, it will often prove advantageous, while we follow one
line of approach, not to shut our eyes completely to the other (Joos 1947:
457).

Joos was a physicist and not a philosopher. H e is clear when

he deals with equations but careless with regard to words and
conceptual issues. For example, he calls " t h e o r y of h e a t " two
different things, of which only the second refers to the theory of
heat. Similarly unhelpful is his belief that the best justification
for theory and hypothesis is " v i v i d n e s s . " H e places "expla-
n a t i o n " within quotation marks that appear to be apologetic. But
it is significant that Joos refers to the non-theoretical approach
that he describes first, as " p h e n o m e n o l o g i c a l . " This shows how
the areas of physics that conform to the requirements of empir-
icism and positivism which are also those of phenomenology,
14 Methods, Meanings and Rules

constitute only a first step and are followed by theoretical and

explanatory accounts that do not conform to these strictures.
The empiricist, positivist, behaviorist, and phenomenological
critiques, by confining themselves to the antechambre, have
missed the most interesting areas of the interior.
This assemblage of thinkers has belatedly been joined by
"ordinary language" philosophers. They, too, reject the " m e t a -
physical" distinction between appearance and reality, and try to
avoid speculation that goes beyond the phenomena. All these
philosophers advocate that one should limit oneself to ordinari-
ness, surfaces and appearances. Phenomenologists confine
themselves to the world of everyday experience (Husserl's
Lebenswelt, in French: le monde vécu); ordinary language
philosophers to ordinary language; both criticize science be-
cause it makes use of abstraction, formalization and theory
construction. Wittgenstein, for example, after rejecting his own
logico-positivistic views in favor of a new ordinary language
philosophy, wrote: " A n d we may not advance any kind of
theory. There must not be anything hypothetical in our consid-
erations. We must do away with all explanation, and description
alone must take its p l a c e " (Wittgenstein 1958:47). Historians of
philosophy have not failed to note that Wittgenstein agrees with
the phenomenologists in their "espousal of description divorced
from explanation" (Munson 1962). I have elsewhere discussed
this remarkable convergence in contemporary philosophy that is
largely due to its alienation from science (Staal 1975: Chapter 4).
In 1961, a meeting at Royaumont brought together represen-
tatives from continental phenomenology and Anglo-American
analytical philosophy. The proceedings were published in La
philosophie analytique (Beck, ed., 1962). At that time it was not
yet fashionable to try to reconcile and harmonize trends from
different philosophical camps (cf. Chapter 30B). Gilbert Ryle
remarked that " H u s s e r l always wrote as if he had never met a
scientist and never heard a witty r e m a r k . " Van Breda, the editor
of the Husserliana, reacted strongly: he did not doubt that
Husserl lacked a sense of h u m o r — " I am paid to be aware of
i t " — b u t he insisted that Georg Cantor and David Hubert were
 The Empiricist Caricature of Science 15

among HusseiTs most intimate friends and that Max Planck was
ringing his doorbell practically every other day. Father van
Breda's veracity is beyond dispute, but it puts HusserFs imper-
viousness to the scientific ideas that surrounded him in an even
harsher light.
Among later positivists, the so-called logical positivists re-
mained empiricists but abandoned the extreme insistence on
observables. A. J. Ayer, in his Language, Truth and Logic of
1936, often regarded as a typically empiricist and positivist
manifesto, agreed " w i t h the rationalists that the process by
which scientific theories come into being is often deductive
rather than inductive" (page 137). Logical positivism, which
paid, increasing attention to scientific theory and explanation,
was brought to Chicago by Rudolf Carnap and influenced
American philosophers such as Quine. Its point of departure had
been that "scientific" statements must be liable to verification or
falsification. Logical positivists first proposed that the meaning
of an empirical statement was " t h e method of its verification or
falsification." In the case of the " a n a l y t i c a l " statements of logic,
truth depends on meaning alone, which implies that they tell us
nothing about the world. However, direct "verifiability" was
found to be non-existent or impossible in many interesting cases.
Karl Popper insisted on "falsifiability", but it soon became clear
that a counter-example calls for an explanation, but need not
invalidate a theory. Quine holds that it is impossible to clearly
demarcate between analytical and empirical statements, that
individual statements are not decidable and that our theory as a
whole, or all our theories together, "impinge on experience only
along the e d g e s " (Quine 1953:42; often elucidated, e.g., Quine
1981:39^0). On such a view there is no clear distinction between
truths that depend on meaning and matters of fact, or between
natural science, logic and speculative metaphysics. This is a far
cry from the original empiricist and early positivist caricature of
the so-called scientific method, and probably closer to the truth.
 T h e Positivist Critique of M e a n i n g

If centrality admitted of degrees, the problem of meaning

might be regarded as the most central problem of contemporary
philosophy. It came into prominence when positivism reacted
against the more pompous claims and expressions of metaphys-
ical speculation that preceded it. The ancient Greek philoso-
phers provided not only the main themes of Western thought but
also a logical terminology by means of which these could be
expressed and discussed. For the Greeks, clarity of thought and
of language had always been important. Socrates' interest in
definitions illustrated these concerns which were further devel-
oped by Plato's conceptual analysis and strengthened by
Aristotle's logic. These kinds of analysis and especially formal
logic were further developed during the Western middle ages, as
we have already noted.
The modern philosophers who tried to make a new beginning
and to reestablish the links with science, continued to follow the
Socratic tradition and pay attention to precision and logic. But
formal logic began to decline. With Leibniz as a lone exception,
it went downhill and reached a nadir around 1800. By that time
philosophers had already lost contact with science; with logic
also gone, they began to cultivate ideas of the Romantic period,
noted for its interest in art and religion and its dislike of science
and logic. Romantic ideas were beneficial to some of the human
sciences, especially the slowly burgeoning science of language
(Chapter 6). But they also initiated a reign of obscurantism from
which we have not yet recovered and in the wake of which
contemporary religious romanticism thrives. The greatest
among the obscure philosophers, Hegel, could claim that the
18 Methods, Meanings and Rules

development of reality was the unfolding of an Absolute Idea

without adding precision, clarification or argument; for although
its claims were expressed in voluminous publications, these
were virtually unintelligible. In Greek philosophy, obscurity had
existed, but it remained exceptional. Heraclitus, whom Hegel
regarded as an important predecessor, had been singled out for
being CTKOTEIVÓS or " o b s c u r e . " It took twenty-three centuries,
however, for obscurity to become a fashion. Lack of concern for
clarity or logic did not prevent Hegel from being influential. On
the contrary, it was the main cause o f t h a t influence.
The posivitist critique of meaning can be illustrated with the
help of almost any sentence written by Hegel. The following
passage in his philosophy of religion is more remarkable than
many because it incorporates an attempt at "clarification":

The Jewish God does not attain to the consummation of spirit. To attain
the consummation of spirit1 means precisely that subjectivity should
offer itself up as infinite: this absolute antithesis is the outermost
extremity of spiritual appearance and negative, infinite return; (that is, it
is) subjectivity, and precisely this subjectivity; (it is) an individual for the
intuiting consciousness (Hegel 1827,111:48; English translation by R.F.
Brown, P.C. Hodgson, and J.M. Stewart with the assistance of H.S.
Harris, 1985,111:113).

Positivists asked: what meaning do such expressions convey?

Sentences cannot be understood in isolation, but that does not
imply that no criteria need to be observed for meaning to be
conveyed. In the above passage, for example, what is meant by
" m e a n s precisely"? What is "offer itself up as infinite"? There
is declared to be an "absolute antithesis," but of what? H o w can
it "oifer itself up as infinite" and also be "infinite r e t u r n " ? That
this subjectivity is this subjectivity does not add information to
its already having been called subjectivity. Even the team of
translators (to whom the additions within parentheses are due)
have not cleared up these difficulties.
According to the positivists, the kind of criteria needed in
order for sentences to convey meaning are similar to the explicit
or implicit principles of language that prevent expressions such
 The Positivist Critique of Meaning 19

as ''mistakes felt bemuse although" or " ( a + ) p = . " Such

expressions convey no meaning because they are not well-
formed. Hegel's sentences were also lacking in well-formedness,
but of a different, semantic kind.
Syntactic well-formedness is a clear concept in the case of the
artificial language of mathematics, but it also applies to ordinary
language. The expression " ( a + ) p = " is not well-formed
because expressions that contain " + " or " = " are only defined
when those symbols occur between two expressions, as in " a +
b " or " a = b . " The expression " a + " is a fragment that is not
well-formed and therefore does not convey meaning. Similar
principles or rules account for the fact that " m i s t a k e s felt
bemuse although" is ungrammatical. Any imaginative person
can think of an interpretation for such,an expression and poets
do not fail to produce them; but unlike an English sentence, this
sequence of four words does not possess an interpretation that
follows from the structure of English and that any speaker of
English will unhesitatingly accept. This lack of meaning is due to
obvious circumstances, for example, that most English sen-
tences contain a main verb. Either " f e l t " or " b e m u s e " would be
suitable candidates, but their juxtaposition, without intervening
connective, makes no sense. Similarly, " a l t h o u g h " is only used
to relate one clause to another, but in the example under
discussion there is only one clause.
The expression "spinach wept wet h i s t o r y " is different from
" m i s t a k e s felt bemuse although" in that it does not seem to be
syntactically wrong: it has one main verb, a subject, and an
object preceded by a qualifier. All is in accordance with the rules
of syntax and yet, it does not make any sense. Here the lack of
well-formedness is not syntactic but semantic. It is easy to
specify what went wrong in this particular example. The causes
lie in t h e structure of the English vocabulary which abounds in
facts such as the following: " s p i n a c h " is not an agent but the
subject of " w e p t " must be an agent; " h i s t o r y " is not a physical
object but " w e t " must qualify such an object.
Critiques of meaningless language are not only found in
Western philosophy. The Buddha similarly objected to the
20 Methodsy Meanings and Rules

question " w h e r e does the saint go after d e a t h ? " which he

regarded as " n o t fitting" and therefore " t o be set a s i d e "
(thapanîya) in the same way as " i n which direction has the fire
gone after it has gone o u t ? " (Majjhima-nikäya 1.483-488). The
notion of "non-fitting" or "semantically a n o m a l o u s " plays an
important role in early Buddhist philosophy (see e.g. Jayatilleke
1963). These problems were also formulated in more systematic
and positive terms. In Indian theories of language, " s e m a n t i c
compatibility" (yogyatä) plays a similar role and is distinguished
from "syntactic compatibility" (of different kinds: e.g.
äkähksä). Examples given by Indian theorists of expressions
that lack semantic compatibility include angina sincati, " h e
wets it with fire"; those that lack syntactical compatibility
include gaur asvah pumsio hastl " c o w horse man e l e p h a n t " (see
e.g. Kunjunni Raja 1963:157,164-166; Staal 1969 = 1976).
Rudolf Carnap objected to Martin Heidegger's Nichts nichtet,
"nothing n o t h i n g s " on kindred grounds. But even if such
expressions should be " s e t a s i d e , " it does not follow that the
problem they may hint at ceases to exist. For with regard even
to such expressions, we are in a position to ask meaningful
questions: not merely whether they are well-formed, and how
they convey meaning if they do, but also, if they d o n ' t , whether
they suggest a meaning that can be expressed in a more
intelligible manner by other expressions. Conversely, clarity is
not enough. There must be something of interest to convey as
well. Clarity of expression is a necessary, not a sufficient
condition for philosophy and science. Plato, Aristotle, De-
scartes, Leibniz, Spinoza or H u m e did not only offer ideas and
theories for discussion but offered them in such clear terms that
we are in a position to discuss them. The Greek concern for
definitions corresponds to the Indian notion of sûtra, developed
in order to express meaning clearly, unambiguously and
"without embellishment" (Chapters 5 and 26).
The contrast between philosophy on the one hand, and
religion or poetry on the other, lies partly in this concern with
meaning. The early stages of religion are characterized by the
fact that " m e a n i n g " is assigned to a variety of entities, e.g.,
 The Positivist Critique of Meaning 21

rivers, mountains, trees or the sky. Poetry expresses similar

sentiments metaphorically. Ordinary usage preserves such be-
liefs, semi-seriously, for it is common and normal to talk about
" t h e m e a n i n g " of a sunset, life, the universe o r man. " M a n
Makes S e n s e , " the title of a reader in anthropology (Hammel
and Simmons 1970), is understandably popular. Following the
positivist critique of obscurity, however, most contemporary
philosophers accept that meaning is not the kind of commodity
that can be assigned to everything; it is primarily, a property of
linguistic expressions. This explains the "linguistic" phase in
contemporary philosophy and the recent importance of linguis-
tics. This development (which also occurred in India) explains
the gap between philosophy and science on the one hand, and
religion on the other; for meaning, minimized by the former t w o ,
is maximized by the latter.
If " m e a n i n g " belongs to expressions of language, it is natural
to assume that these expressions refer to reality directly and, as
it w e r e , " d e p i c t " it. Frege had improved the medieval insight
that some terms mean without referring. His famous examples
are " m o r n i n g s t a r " and "evening s t a r , " expressions with a
difference sense that refer to the same planet. The most precise
and powerful theory of reference and truth was developed by
Alfred Tarski.

3A. Ludwig Wittgenstein

In his first book, the Tractatus Logico-Philosophicus (1922),

Wittgenstein stated that the expressions of language are elemen-
tary or composite. Elementary expressions describe atomic
facts, are true when those facts exist, false if they d o n ' t , and
consist of series of names that directly refer to things. Composite
expressions are truth functions of elementary expressions, con-
structed from the latter with the help of logical connectives such
as not, and, and or. The truth or falsity of composite expressions
is completely determined by the truth functions of the elemen-
tary expressions which they comprise. A composite expression
which is true independent of the elementary expressions of
22 Methods, Meanings and Rules

which it consists (e.g., "/? or n o t - p , " where p stands for any

expression), is called a tautology. According to Wittgenstein,
logic consists of such tautologies, which " d o not say anything
about the world and thus have no factual c o n t e n t . " Logic,
therefore, is without reference and, in a sense, meaningless
(Naess 1965:16). Since many philosophical expressions cannot
be constructed from elementary expressions, their truth value is
undetermined and they are therefore meaningless as well. This
holds a fortiori for statements that claim to refer to the ineffable.
For that of which we cannot speak we must remain silent.
Criticism of this account came from different directions.
J. L. Austin objected to the thesis that all statements of language
must be true or false. Tarski had not made that assumption but
Wittgenstein's treatment suggested it. Austin pointed to a fact
well known to philologists and students of language, that many
utterances of language do not conform to this thesis: questions,
c o m m a n d s , expressions of hope and, perhaps, belief, and utter-
ances by which something is not only said but also done. Austin
included among the latter " c e r e m o n i a l " statements such as
salutations or baptizing a ship: when it is called " Q u e e n Eliza-
b e t h , " it is not given a description which may be characterized
as true or false, but a name. Other conventions must hold. The
person baptizing must be qualified to do so, unlike a person who
says " 2 x 2 = 4 . " " S u p p o s e , " says Austin,

that you are just about to name the ship, you have been appointed to
name it, and you are just about to bang the bottle against the stem; but
at that very moment some low type comes up, snatches the bottle out of
your hand, breaks it on the stem, shouts out 'I name this ship Genera-
lissimo Staling and then for good measure kicks away the chocks. Well,
we agree of course on several things. We agree that the ship certainly
isn't now named the Generalissimo Stalin, and we agree that it's an
infernal shame . . . (Austin 1961:226-227).

Austin suggested that some of the expressions that are used

under such circumstances are not " t r u e " or " f a l s e " , but "felic-
i t o u s " or "unfelicitous." The utterances of language which, in
saying, do something, he dubbed "performative u t t e r a n c e s . "
 The Positivist Critique of Meaning 23

Following Smart (1965), Quine has been critical of Austin's

notion because Tarski's account works equally well for perfor-
matives:
li
I bid you good morning" is true of us on a given occasion if and only
if, on that occasion, I bid you good morning. A performative is a notable
sort of utterance, I grant; it makes itself true; but then it is true. There are
good reasons for contrasting and comparing performatives and state-
ments of fact, but an animus against the truth/false fetish (a reference to
Austin 1962:150) is not one of them (Quine 1981:90).

The fiercest critic of Wittgenstein's first book was Wittgen-

stein himself. In his second book, published thirty years later
and entitled Philosophical Investigations (1953), Wittgenstein
showed that expressions of language rarely consist of names that
refer to things. Words have a variety of functions and it is a
cause of philosophical confusion to assume that they all function
like names. Sentences, therefore, do not depict reality in any
direct or straightforward fashion. Expressions that are similar in
syntactic or grammatical structure are often different in seman-
tic or logical structure. Wittgenstein attributed to Bertrand
Russell the insight that the apparent logical form of proposition
need not be its real one. Although such insights were new to
philosophers and were heralded as revolutionary, they have
never been rare in classical linguistics. The eighteenth century
grammarian Du Marsais, for example, showed that " I have an
i d e a " and " I have a b o o k , ' ' despite their identical syntactic
structure, have different semantic structures, and should not
only be interpreted differently, but also in a different manner.
When Wittgenstein made his famous declaration that the mean-
ing of a word often lies in its use he echoed what the Latin
grammarians, and many schoolmasters following them, had
always known: verba valent usu, " w o r d s derive their meaning
from their u s e . " The Sanskrit grammarians linked this insight
with their criticism of commentators who believed, as many
people still do, that the meaning of a word is based upon its
etymology; this is incorrect because yogäd rüdhir bally ah,
" u s a g e is stronger than etymology." The Sanskrit grammarians
24 Methods, Meanings and Rules

also distinguished clearly between logical and grammatical'form

(see Kiparsky and Staal 1969; now republished in Staal 1988a:
184-281).
These developments explain why philosophy became entan-
gled in questions that resort under a more specialized scientific
discipline, namely, linguistics. Western linguistics did not have
ready answers to many questions about usage or about the
distinction between logical and grammatical structure since it
was itself relatively undeveloped and had suffered, as we shall
see in Chapter 6, from unnecessary strictures and prohibitions.
The question how meaning is expressed in a particular language
is not a philosophical question( although it possesses philosophic
features and implications), but an empirical question about the
" s e m a n t i c s " o f t h a t language or about a "universal s e m a n t i c s "
that applies to all languages.
The difference between logical and grammatical categories
does not coincide with that between semantic and syntactic
structures. Syntactic structures often express differences that
traditional grammarians would have called " l o g i c a l . " This is
illustrated by the following pair of sentences, made famous by
N o a m Chomsky:

John is eager to please (1)

and

John is easy to please (2)

These two expressions differ with respect to one word; to be

precise: the second syllable of a word. This much is apparent
when we restrict ourselves to " s u r f a c e " features. However,
once we analyze their structure, we discover a great deal more.
In both sentences, " J o h n " is the grammatical subject. But (1)
and (2) express meaning in a different manner and have different
syntactic structures. In (1), " J o h n " is the logical subject of the
sentence, which recurs in related expression such as " J o h n ' s
eagerness to p l e a s e . " In (2), " J o h n " is the logical object, which
recurs in " t o please John is e a s y " , "it is easy to please J o h n , "
 The Positivist Critique of Meaning 25

etc. If we reverse these structures, we obtain ungrammatical

results which I shall mark, in linguistic fashion, with the asterisk

* "John's easiness to please"

* "to please John is eager"
* "it is eager to please John."

Though these expressions have been obtained by analogous

processes, it is not clear what meaning, if any, may be assigned
to them. Something similar holds for expressions such as
"fearful c r e a m " and "spinach wept wet h i s t o r y , " which are
syntactically well-formed but semantically unacceptable.
The positivist critique of meaning led to the insight that the
meaninglessness of certain metaphysical statements constitutes
a special case of the meaninglessness that linguists try to explain
in natural languages. The assistance of linguistics is valuable
because obscurity is not necessarily a mark of lack of thought or
insight. Descartes' first exposition of analytical geometry was
extremely obscure and is still very difficult to read even at
present when every schoolboy is familiar with the subject. The
critique of meaning, therefore, reduces to three parts: obscurity
remains undesirable because it obfuscates the distinction be-
tween what makes sense and was does not; since sometimes
sense turns out to be concealed, everything that is obscure
should not immediately be " s e t a s i d e " ; linguistic and semantic
analysis is necessary to find out which is which.

3B. Martin Heidegger and Hans-Georg Gadamer

I shall illustrate these observations with a brief discussion of

two obscure philosophers; in the first instance, but not in the
second, sense can be detected. The first is Martin Heidegger.
When he wrote down the phrase das Nichts richtet, "nothing
n o t h i n g s , " to which Carnap, Ayer and others objected, Heideg-
ger was not unaware of the strangeness of the expression. H e
had already introduced other similar ones into his philosophy. At
26 Methods, Meanings and Rules

the end of the Introduction to Sein und Zeit of 1927, he wrote

that for performing the task he set himself, not merely words
were lacking but above all, " g r a m m a r : " we need only compare
certain passages in Plato's Parmenides or in Aristotle's Meta-
physics with a narrative from Thucydides, to realize to what
extraordinary formulations the Greeks were subjected by their
philosophers. " A n d since the forces are essentially weaker, and
the area of being that should be unlocked is ontologically much
more difficult than what the Greeks had before them, the
circumstantiality of concepts and the harshness of expression
will accordingly i n c r e a s e " (Heidegger 1953:39).
We should be charitable in accordance with a principle of
Quine (uncharitably discussed in Staal 1988a: Introduction) and
assume that Heidegger tried to convey meaning. Later in Sein
und Zeit, Heidegger actually makes a proposal that reflects a
critical analysis of linguistic usage and is not different in char-
acter from the critical observations linguistic philosophers had
offered. He argued that it is misleading to ask, with regard to our
own being, " W h a t " (is it)? Since our being is always our own
and as such unique, we must ask " W h o ? " This implies that the
Aristotelian categories, which arose from an analysis of ques-
tions of the form " w h a t ? " , cannot be used. Heidegger therefore
introduces a new kind of characterizations of being, which he
calls Existenzialien, "existentials." In his introductory chapters
he tried to clarify the distinction between " e x i s t e n t i a l s , " which
deal with our own being, and " c a t e g o r i e s , " , which deal with
other being. I shall not follow this exercise further, but it is not
surprising to find that new expressions, including verbal neolo-
gisms such as the nichten of " d a s Nichts n i c h t e t , " emerged from
it.
Whether it is true that our own being is fundamentally
different from other being because it is "always our o w n , " is a
topic to which we shall return: for it is related to the doctrine that
the humanities (or Geisteswissenschaften) are fundamentally
different from the natural sciences {Naturwissenschaften) and
require a different method of study (Chapter 29). As for the new
" g r a m m a r " that Heidegger needs, it is not far-fetched or pro-
 The Positivist Critique of Meaning 27

hibited to abandon ordinary language and introduce new lan-

guages that use a different grammar. The best examples of such
artificial languages are the languages of mathematics. There are
various misapprehensions here, however. It is not true, for
example, that there is only one such language, and that it deals
with " q u a n t i t y " and not with " q u a l i t y . " On the contrary,
mathematical languages come in great variety and if we have
specific needs it is often possible to construct an artificial
language that meets these needs. Einstein, for example, used the
language of " t e n s o r analysis" to express relationships he
needed in his general theory of relativity, and which he hap-
pened to have studied.
The use of artificial languages is mostly confined to the natural
sciences and is largely responsible for their success. The rise of
modern science in the sixteenth and seventeenth centuries is due
precisely to this application of mathematics to the study of
nature (for an interesting discussion of why this did not happen
in China, where both mathematics and the natural sciences were
highly developed, see Needham 1959:150-168). The application
of artificial languages, is not confined to the natural sciences. It
has been useful in economics and especially in logic and linguis-
tics. It could be used in logic because it was in some respects
based upon Aristotle's categories: the expression P(x), "jt is
P , " for example, reflects Aristotle's distinction between subject
and predicate, as does the notion of a function expressed by
" f ( x ) . " In mathematics and physics, this one-place function is
generalized to many-place.functions such as "f(x, , . . . , x n ) . "
It is likely that an artificial mathematical language could be
constructed to express whatever is clear in Heidegger's
" e x i s t e n t i a l s . " To abstract a notation from verbal notions, use
could be made of the A-operator introduced by the logician
Alonzo Church. This operator enables us to abstract a feature
from a compound expression, and introduce and employ it
elsewhere. Such operators may be used to construct a language
that expresses meaning intelligibly in a systematic manner,
unlike an isolated expression such as " d a s Nichts n i c h t e t . "
Carnap's criticism of that expression was therefore justified:
28 Methods, Meanings and Rules

for such an expression is helpful and productive only if it is

well-formed and functions within the context of a well-defined
language. To create such a language is not only common in
mathematics or physics. Among philosophers, Aristotle, Leib-
niz, and even Spinoza, not to mention Peano, Frege, Russell and
Whitehead, tried to construct such artificial languages. In Hei-
degger, the expression Carnap criticized may not be meaningless
but remains an isolated curiosity, liable to spread confusion
rather than illumination. Heidegger's entire development bears
this out: for he never attempted to provide "general a c c o u n t s "
that could be studied within the framework of a general lan-
guage. On the contrary, he refused to accept any truth that is
independent of " h u m a n being" {Dasein): " t h e laws of N e w t o n ,
the principle of noncontradiction, any truth whatsoever is only
true as long as Dasein is . . . before N e w t o n ' s laws were
discovered, they were not t r u e ; " to which he added, curiously
(for Heidegger has never evinced sympathy for intuitionism,
which exludes the principle of the excluded middle), " t h a t does
not however imply that they were false" (Heidegger 1953: 226).
My second example illustrates an obscure philosopher w h o ,
however charitably we analyze his expressions, does not seem
to yield sense. This takes us closer to contemporary problems
for it is concerned with " h e r m e n e u t i c s , " a topic that will be
discussed later (Chapters 29 and 30 A - B ) . The exposition of
hermeneutics in G a d a m e r ' s Wahrheit und Methode ("Truth and
M e t h o d " ) of 1960 exemplifies meaninglessness better than He-
gel, Heidegger, or any other contemporary philosophy with
which I am familiar.
G a d a m e r ' s "philosophical h e r m e n e u t i c s " combines Heideg-
ger with HusserFs phenomenology and Dilthey's distinction
between Natur- and Geisteswissenschaften to provide a new
philosophical foundation for the Geisteswissenschaften by
means of a theory of interpretation. Gadamer makes a surprising
move to begin with: he accepts only one kind of method—that of
the natural sciences. He characterizes this method in the same
terms as empiricists and positivists have done before him: it is
the caricature of scientific method in its pristine form. Gadamer
 The Positivist Critique of Meaning 29

rejects it, good for him; and since it was the only method he
recognized; he rejects all method.
This move sounds radical but Gadamer is not the radical type
of philosopher we see in Heidegger. Heidegger was radical in
style and substance. The difference in style between G a d a m e r
and Heidegger is illustrated by their manner of referring. In
Heidegger references are scarce, and limited to Plato, Aristotle,
Aquinas ( " T h o m a s von A . " ) , Nietzsche, Husserl, and a few
others including increasingly in his later work, the classical
German poet Hölderlin (1770-1843) who wrote, for example:

Nah ist
und schwer zu fassen der Gott,
C'Close, and difficult to grasp, is the god.")

These lines (the first of Patmos) introduce a being similar to

Dasein: in Sein und Zeit, almost beyond reach, but in Heideg-
ger's later work always already " t h e r e " {da). This notion
resembles the Vedäntic concept of brahman, an entity beyond
everything that is also identified with my self. In Heidegger, we
meet not only with references, but also with m o m e n t o u s put-
d o w n s , for example, of Descartes, Dilthey or Kierkegaard:
" t h e y have not advanced beyond X " is a typical Heideggerian
form of critique in which elementary insights that fit the case are
substituted for " X " . Whatever the force of these criticisms, to
be quoted by Heidegger is tantamount to immortality. A w e s o m e
and difficult to grasp are the differences of opinion between
philosophers thus depicted. When Heidegger differs with
Aristotle, his prose takes on the purple glow of a combat
between giants, a collision between mountain massifs as in
Japanese Sumo.
T o be quoted by G a d a m e r is nothing special. H e refers to the
haute burgeoisie of European, especially German, culture.
Great names rank with contributors to Festschrifte, and since
texts are the primary objects of " h e r m e n e u t i c s , " this is not
surprising. Hermeneutics had been introduced into textual and
historical interpretation by Dilthey (Chapter 29). It plays an
30 Methods, Meanings and Rules

important role in Heidegger, but shares its name with Aristotle's

small treatise De Interpretationen in Greek: irepi ép/xTji/eios.
Here Aristotle was concerned with topics such as subject and
predicate, affirmation and negation, modalities, etc. That the
negation of the sentence is the negation of its predicate, is first
stated there. This inspired the logical notation Russell and
Whitehead introduced in the Principia Mathematica that logi-
cians have adopted: for " ~ ( P ( x ) ) " is the same as " ( ~ P ) ( x ) " so
that it is sufficient to write " ~ P ( x ) . "
G a d a m e r ' s hermeneutics attributes to logic and natural sci-
ence "unassailable and anonymous a u t h o r i t y . " As it happens,
science, like the number of planets, is not unassailable; it is
subject to revision. It is also not anonymous: we speak of
" E u c l i d ' s Fifth P o s t u l a t e , " " F e r m a t ' s Last T h e o r e m , "
" P l a n c k ' s c o n s t a n t , " and " E i n s t e i n ' s Theory of Relativity." It
has no authority at any rate among scientists, who know that it
is provisional and never the last word. Having set up his straw
man, Gadamer contrasts him with hermeneutics. The latter does
not act piecemeal, but " a i m s at totality," just as learning
Serbo-Croatian is more than "going through a dictionary of
words."
We shall return to the chief notion of hermeneutical analysis:
Verstehen, for which " u n d e r s t a n d i n g " is too general a transla-
tion and " e m p a t h y , " too narrow. According to Gadamer, Ver-
stehen Calvinism is " t o be like a Calvinist." H e adds that this
concept does not exist in the natural sciences. It is true that we
are not birds, trees or stars, and most of us are not Calvinists;
but we are animals, bodies, objects, substances, and consist of
cells, molecules and atoms. Gadamer also stresses that Verste-
hen includes, but science excludes " i n t u i t i o n , " " e m o t i o n , " or
"feeling," which, as a matter of fact, play an important role in
the discovery or invention of scientific theories although they are
not part of the theories themselves. But Gadamer wishes to
make the difference between the sciences and humanities as
large as possible and in so doing does not hesitate to place the
humanities in a light that must seem disturbing to their best
practitioners. For on his account, Verstehen as applied to a text
 The Positivist Critique of Meaning 31

is arbitrary: " N o r m a t i v e concepts such as the author's intention

or the original reader's understanding represent in fact nothing
but empty slots, which may be filled with Verstehen as the
occasion a r i s e s " (Gadamer 1965:373).
If Verstehen can put meaning in all openings, like putty, we
may interpret Gadamer according to our wish. He seems to
intend this, for "language is speculative . . . in as far as the finite
possibilities of the word are correlated to the intended sense as
a trend toward infinity" (page 444). This means, as is seen from
the context, that an expression of language does not mean what
it says, but points to an infinity of things unsaid. G a d a m e r
restricts this poly-interpretability again in what must be pro-
claimed the star sentence of his work, " i n propositions, the
meaning horizon of what is actually to be said is concealed with
systematic e x a c t i t u d e " {ibid.). This means that a sentence
always conveys the opposite of what is says.
Such a scenario brings to mind Aristotle's characterization of
the principle of non-contradiction as a principle with which one
cannot disagree without accepting it (Metaphysica F). Either
one disagrees with what Gadamer says, in which case one must
agree with what he means; or one agrees with what he says by
disagreeing with its meaning. One must in all cases agree and
disagree, and G a d a m e r ' s originality lies in this combination. H e
has adopted from the positivist-empiricist tradition its most
monumental error—the caricature of the scientific method—and
failed to heed its most valuable contribution—the critique of
meaninglessness. And this philosophy aims at instructing us
about interpretation!
 Linguistic Background

The history of linguistics does not begin in Greece but in India.

The proper way to introduce the subject would therefore be to
start there and mention the West only when Indian ideas were
introduced into European speculations about language. This
would conflict with prejudices of long standing, e.g., that science
is an exclusively Western development, and be misleading
because it would not explain why Indian ideas were so long and
consistently misunderstood in the West. I shall therefore begin
and proceed in the conventional manner with Western antiquity,
turn to India on reaching the stage of Western development
when Indian ideas were beginning to be introduced, and return
to the West after surveying Indian linguistics. This sketch
contributes to the present investigation because we shall later
study the Indian science of ritual, which is closely related to
Indian linguistics but has no counterpart among Western sci-
ences (Chapter 26).
In Western Antiquity, the Alexandrian school of Dionysius
Thrax (around 100 B . C . E . ) , extended by Apollonius Dyscolus,
adopted a conception of grammar in which words were regarded
as the smallest functional units of language and were considered
indivisible like atoms. The only smaller units which the Greek
grammarians recognized were letters. They referred to these by
means of the Greek term crroi^eiW which is the same as the
word for " e l e m e n t " Euclid used in the title of his work on
geometry, ' T h e E l e m e n t s . " His commentator, the Neoplatonist
Proclus, a metaphysician and mathematician, commented on
this similarity:
34 Methodsf Meanings and Rules

Just as an utterance in language consists of first principles that are the

most simple and indivisible, and each word Àéfiç and each speech Aóyos
is built up from those, similarly there are certain theorems that precede
all of geometry, constitute principles for the theorems that follow, extend
over all, and provide proofs for many special cases; they are called
elements (ed. Friedlein, 1873:72).

That the ancient Greek science of grammar did not develop is

partly due to the fact that letters are not in any significant sense
the elements from which words are constructed. What they do is
provide information about spelling. The confusion between
grammar and spelling has pervaded European ideas about lan-
guage for many centuries and is still occasionally to be found.
The same holds for the related misconception that grammar is a
prescriptive discipline which determines correct usage.
The confusion between grammar and spelling existed on a
different level in China, where writing characters has always
been a more complex and significant matter than in languages
that use an alphabetic system. In the Far East in general,
calligraphy occupies a central place in art and civilization. It is a
"Gesamtkunstwerk" that would have appealed to Richard Wag-
ner if he had been able to appreciate how it integrates painting,
sound, meaning and harmony. The Chinese conception of lan-
guage is permeated by the art of writing as is illustrated by the
efforts of Chinese Buddhist monks to understand Sanskrit. They
thought that this knowledge consisted in learning how to write
and how to pronounce the written language. In Chinese, both
tasks are immense, and when one has completed them, one has
in the process mastered a language with which one was already
familiar to begin with. In Sanskrit, on the other hand, both tasks
are simple. Having mastered the Sanskrit alphabet and its
pronunciation, the Chinese believed that they knew the lan-
guage; and when they repeatedly noticed that they did not, they
were at a loss to explain why. A curious circumstance supported
their assumption: many Sanskrit texts they looked at abounded
in mantras that happened to conform to their idea of language.
For knowing how to derive the recitation of a mantra from the
way it is written is almost all there is to know about it.
 Linguistic Background 35

In E u r o p e , the concept of grammar of the Alexandrian school

" e n d u r e d with little real change for more than ten c e n t u r i e s "
(Robins 1976:18). The Greek grammarians were replaced by
Latin grammarians such as Priscian and Donatus, but their
outlook and methodology remained the same. A new science of
grammar prospered between the twelfth and fifteenth centuries,
and inspired theories and philosophies of grammar for a few
more centuries. Important innovations were due to the Mo-
distae, who constructed grammars in the late thirteenth and
early fourteenth centuries in line with logic and semantics or
" t h e mode of signifying" {modus significandi; see, e.g., Bursill-
Hall 1971). These efforts depended on philosophy, especially
Aristotelian categories and logic which had been introduced
earlier into medieval philosophy. The notion of " r u l e " (régula)
was introduced from logic along with these other notions, but its
use remained largely confined to logical rules of deduction such
as occur in Aristotelian syllogistics. With the help of this
conceptual apparatus, the " g r a m m a r i a n s " triumphed over the
humanistic schools of the " a u t h o r s " who had been more textu-
ally oriented. But a more scientific and empiricist approach was
introduced into Western linguistics only after the discovery of
Sanskrit.
N e w data about language, and a concomitant turn to empiri-
cism, were needed to transform Western ideas about language
into a science. The new data arrived, but the resulting empiri-
cism followed H u m e and adopted the caricature of scientific
method we have reviewed above. If Western scholars of lan-
guage had adopted the rationalism of the Indian grammarians,
which was embodied in their notion of rule, linguistics could
have been related to logic and mathematics in a fruitful manner.
Instead, Western scholars abandoned logic and concentrated on
texts. This limitation, combined with positivism, was fruitful in
historical and comparative philology and phonology, but pre-
vented progress in pure linguistics for another two centuries.
At the end of the eighteenth century, Sir William Jones (1746—
1794) discovered that Sanskrit, the classical language of India,
was related to the languages of Europe. This discovery, and the
36 Methods, Meanings and Rules

promulgation of the facts of that relationship, marked the

beginning of scientific philology. Facts about Sanskrit and
fragments of Sanskrit grammars in Latin had existed earlier, as
did scattered information about Indian linguistics. Filippo Sas-
setti (1540-1588), for example, was impressed by the Indian
discovery that different sounds are produced by the various
movements of the mouth and the tongue, a fact unknown in the
West. Sassetti attributed the large number of sounds in Indian
languages to the widespread custom of chewing betel leaves and
areca nuts (Staal 1972:30). But such hypotheses were of no
significance compared to the birth of "comparative g r a m m a r "
which was the work of Franz Bopp.
The first European who knew Sanskrit well was Sir Charles
Wilkins (17497-1836), whose " G r a m m a r of the Sanskrita Lan-
g u a g e " was published in 1808. In that grammar, he used the
Indian method of analyzing words into smaller units. Bopp
learned this technique from Wilkins' grammar and used it in his
" S y s t e m s of conjugation in Sanskrit compared with those in the
Greek, Latin, Persian and Germanic languages" of 1808. With
the help of his Indian techniques, Bopp was able to compare
related languages for the first time in a systematic and scientific
manner (Thieme 1982/1983:3).
In order to evaluate the Indian contribution to linguistics and
the lively controversies that followed its adoption by Western
linguists and philosophers, we shall now turn to linguistics in
India and return in Chapter 6 to the sequel of linguistic studies in
Europe that led to contemporary linguistics.
 T h e Origins of Linguistics in India

The origin of linguistics in India is connected with the oral

transmission of the Vedas. It is customary in Western anthro-
pology and related disciplines to distinguish between ' l i t e r a t e "
and " n o n l i t e r a t e " societies, and to regard the former as more
advanced and civilized. But in India, there was no system of
writing during the Vedic period, and scientific linguistics would
not have developed if writing had been known (see below, pages
371-372). In order to safeguard the accurate transmission of the
V e d a s , the continuous (samhita) form of recitation, called
Samhitäpätha, which corresponds to the flow of speech in
spoken language, was dissolved into its constituent elements and
the concomitant modifications were studied. The sandhi junc-
tions or euphonic combinations between these elements were
analysed. At first, attention went primarily to words (called
pada) although some stems and endings were discovered and
similarly " s e p a r a t e d " . The "word-for-word" recitation, set up
alongside the Samhitäpätha, was called Padapätha.
The Samhitäpätha and Padapätha were related to each other
by means of rules. That such rules need to be numerous and
complex can be imagined from the following example:

SAMHITÄPÄTHA: órv aprä âmartyâ nivâto devy udvâtah / /

PADAPÄTHA: / â / uni / apfäh / âmartyâ / ni-vâtah /
devl / ut-vâtah / /
6
'the immortal goddess has pervaded the wide space,
the depths and the heights."

I shall, by way of illustration, formulate three rules that are

needed to express the relationship between these two modes of
recitation. I shall not formulate these rules in the same manner
38 Methods, Meanings and Rules

in which this was done in India, for this would involve more
Sanskritic detail than is needed or wanted in the present context.
Ï shall use instead the expression of a rule by means of an arrow
adopted in modern linguistics. This can be shown to be equiva-
lent in all relevant respects to the expressions used by the
ancient Indian authors (Staal 1965a). A rule which is formulated
with the help of such an arrow, for example:
A->B,
may be read as: "replace A by B , " or "substitute B for A . " I
shall use another symbol, the " + " sign expressing immediate
succession or concatenation, without defining it, but its use will
be obvious. It should be borne in mind, lastly, that my rendering
of these rules makes use of the customary transliteration of
Sanskrit into the Roman alphabet which is in some respects
different from the exclusively oral methods of expression used
by the ancient Indians (see Chapter 27).
The first rule may now be formulated as:
a + w-> o. (1)
This rule expresses that the w V of " o r v " in the Samhitäpätha
is substituted for the combination of " a " and the first ' V of
"uru" in the Padapätha.
The second rule explains the ' V of "orv", but it cannot be
simply formulated as:
u —» v,
for this would entail that all occurrences of' V are replaced by
' V ' . This rule applies only when its context is restricted, as in:
u —» v when a follows. (2)
The third rule similarly explains "devy":
i —» y when u follows (3)
Rules (2) and (3) look similar; they express, in fact, a corre-
spondence that holds in similar circumstances. It may be ex-
pressed with the help of a generalization from (2), (3), and other
rules in something like the following form:
 The Origins of Linguistics in India 39

vowel —» corresponding semi-vowel when a different

vowel follows. (4)

Attached to each of the Vedic schools was a Prätisäkhya

which aimed at a complete description with the help of rules of
the derivation of the Samhitäpätha from the Padapätha of that
school. These works studied a few, well-defined grammatical
topics such as sandhi and accent, always with reference to a
specific corpus, and discussed the notion of rule that developed
over the centuries. Rules such as (1), (2) and (3) are common in
this literature. But rules of a more general form such as (4) were
only beginning to be formulated.
The earliest Padapätha, that of the Sâmaveda or Veda of
Songs, was composed around 1,000 B . C . E . At roughly the same
time, the sentences and verses of the Vedas, especially the
Rigveda, were set to music and transformed into mantras for use
in ritual. For this purpose, the sounds of the original were not
only taken apart and analysed, but also transformed into other
sounds. These modifications were also formulated by rules.
Some of these conformed to sound correspondences needed or
postulated by the grammarians. Other sound transformations
had nothing to do with language structure but were necessitated
by the requirements of melody. We therefore find linguistically
meaningful rules such as those that transform / into e or agni,
" f i r e " , into its Vocative case ór mode of address: agne, as well
as linguistically meaningless rules such as those that transform
agni into melodic forms like o-gna-i.
Linguistics was adumbrated in the Prätisäkhyas but originated
in the tradition of grammatical analysis of which the oldest
extant work is the Sanskrit grammar of Pânini (6th or 5th century
B . C . E . ) , generally referred to as the Astädhyäyl or "Eight
C h a p t e r s . " In this work, which consists of almost 4000 brief
rules called sütra, the interest has shifted from the fixed corpus
of Vedic recitations to the living domain of spoken Sanskrit. The
grammar was not prescriptive but descriptive and based upon
empirical data: it did not legislate what does or does not belong
to Sanskrit, but took the forms of language from usage. Pänini's
40 Methods, Meanings and Rules

commentator Patanjali (ca. 150 B . C . E . ) explained this: when you

need pots, you go to a potter; but when you need words, you
d o n ' t go to a grammarian; you go to the people who speak the
language.
When the grammarian is a native speaker (the only case
considered in ancient India), the language is always already
available to him. If he wants to form the plural of a noun, he does
not have to look for it in a recitation or a text; he knows it
because he knows his native language. His task is to postulate
and formulate rules that account for all the forms he knows and
encounters, and for their various uses. A difference between the
Prätisäkhyas and the grammars of Pânini and the other Indian
linguists, therefore, is that the former confined themselves to
finite collections of utterances (uccärana) such as were trans-
mitted in the Vedas, whereas the latter studied the infinitely
many sentences (väkya) that make up any living language.
The insight that language is infinite was formulated by Pa-
tanjali in a colorful passage that deserves to be quoted in full:

Now if grammatical expressions are taught, must this be done by the

recitation of each particular word for the understanding of grammatical
expressions—for example, should we recite the grammatical forms gauh,
"cow," asva h, "horse," p umsah, "man," hasti, "elephant," sakunih,
"kite," mrgah, "deer," brähmanah, "brahmin?" No, says the author
[that is, Patanjali], this recitation of each particular word is not a means
for the understanding of grammatical expressions. For we have a
tradition which describes how Brhaspati addressed Indra during a
thousand divine years going over the grammatical expression by enun-
ciating each particular word, and still did not attain the end. With
Brhaspati as the professor, Indra as the student, and a thousand divine
years as the period of study, the end could not be attained; so what of us
at the present day, who when we live in full live at most a hundred years?
. . . The recitation óf each particular word, therefore, is not a means for
the understanding of grammatical expressions. How, then, must gram-
matical expressions be understood? Some work containing general and
particular rules has to be composed (Mahübhäsya, ed. Kielhorn, 5.23—
6.3).

This passage refers implicitly to the practice of fcfcenumer

a t i o n " (samkhyâ) which was adopted in the Prätisäkhya litera
 The Origins of Linguistics in India 41

ture and which is characteristic of the least scientific and most

popular of the later philosophic systems, accordingly referred to
as " S â m k h y a . " But since the forms of language are infinite, they
cannot be described by enumeration, but must be characterized
in a more systematic and scientific manner, that is, by "general
and particular r u l e s " {sämänyavisesaval laksanam). In this last
expression, Patanjali used the term laksana which is a common
synonym for sütra or " r u l e . "
The infinity of language is a feature of language that, in the
West, had been known to logicians, but to which the classical or
medieval grammarians did not pay attention. A simple example
is concatenation by means of a connective such as " a n d " . If
" A " and " B " are grammatical expressions, and certain other
conditions are also met, " A and B " is also a grammatical
expression. Thus we can form "fathers and m o t h e r s , " "fathers
and mothers and sisters and b r o t h e r s , " etc. In English there is a
rule, applicable to the written form of the language, that replaces
all but the first occurrences of " a n d " by a comma, and other
languages have similar devices that may effect changes in
expressions of the form " A and B and C and. . . . " But
expressions of this latter form have to be postulated in the first
place if we wish to explain any others that express concatenation
or conjunction.
In Sanskrit, indefinitely large expressions are formed in vari-
ous ways but the most striking is nominal composition. The
collection of animals referred to by Patanjali, that in English
could be referred to by the conjunction: " c o w s , horses, men,
elephants, kites, deer and b r a h m i n s " is in Sanskrit referred to by
a single nominal compound: gaurasvapurusahastisakunimrga-
brähmanäh, in which the single elements are concatenated in
accordance with rules. Pânini's grammar describes such rules
and many others that account for the infinity of language.
It is sometimes objected that such an explanation is inappro-
priate because language is not " r e a l l y " infinite: sentences or
other linguistic expressions are not literally infinitely long since
the human life span is finite. The point of such an analysis is,
however, that it generates a structure for which it is impossible
42 Methods, Meanings and Rules

to fix an upper limit or boundary. For any sentence 5 , " H e said

that 5 " is also a sentence, and this process may be re-iterated
indefinitely. The Netherlands constitution is formulated without
making use of periods, as an enumeration of hundreds of legal
principles separated by semicolons, following the opening
phrase: " W E , Beatrix, Queen of the Netherlands, e t c . , have
ordained . . . " followed by several hundred pages of text,
formulated in the form of a single sentence. N o one has objected
to the Netherlands constitution because of this special feature; it
is accepted as a meaningful, albeit lengthy, sentence. That in
spoken language an upper bound seems to be necessary and may
be actually determined is a different matter. It is not a matter of
linguistic competence but due to limitations of memory and
other restrictions that are of a psychological nature and belong to
what is today called performance. We shall return to this feature
(Chapter 21C).
In logic and mathematics, the generation of infinitely many
forms by a single, finite rule, has long been known. The kind of
rule that effects such an apparent miracle is called recursive.
Such a rule produces infinitely many forms because it can be
applied to its own output. A simple example is:
A - ^ AB.
If this rule is applied to the " A " that occurs on the right, we
obtain:
(AB)B,
and if we re-iterate the process, we obtain, omitting the paren-
theses:
A B B B , A B B B B , A B B B B B , A B B B B B B , etc.
Pânini did not explicitly name or discuss such recursive rules
in his grammar but he made use of them in a variety of contexts.
His grammar is a device that generates an infinite class of correct
expressions of Sanskrit; it is, therefore, what is presently called,
following N o a m Chomsky, a generative grammar.
Pânini's grammar analyzed words systematically into smaller
 The Origins of Linguistics in India 43

units such as roots, stems, suffixes, prefixes and infixes. Some of

the latter were set up for grammatical purposes and attached to
forms to indicate that certain grammatical operations had to be
applied. Such markers are nowadays called "meta-linguistic"
and Pänini recognized their special and separate status. The
connexions between nouns, verbs and other forms were dealt
with in his grammar by systems of relations, the most important
of which were "grammatical relations" called käraka that oper-
ate on different levels. In this context, the distinction between
what we referred to as logical and grammatical relations (page
24), was discovered and analysed.
Adopting a modern terminology, the origin of linguistics in
India may be characterized in approximate terms by saying that
the phonetic and phonological investigations of the Prätisäkyas,
which were piecemeal and ad hoc, were extended and the results
supplemented with similar investigations pertaining to morphol-
ogy, syntax and semantics, all combined in a single generative
grammar of spoken Sanskrit. The main innovation was a meth-
odology that developed the familiar concept of rule into some-
thing more abstract and led to the study of categories of rules,
rule systems and rule order. This was possible because Pânini's
rules were formalized in a manner that corresponds to mathe-
matical formalization in the West. Metarules governed the use of
"metalinguistic" markers and explained and facilitated the ap-
plication and operation of the rules and the order in which they
had to be applied. Some of these formulated what appear to be
obvious matters, like some of Euclid's axioms and theorems.
For example, a metarule stated that the meta-linguistic markers
have to be removed from the finally derived forms. An example
from English may illustrate this. If we describe the formation of
the past tense by introducing a "suffix -ed" the past tense of
work is not:

*worksuffixed

(where the asterisk " * " denotes ungrammatically), but:

worked.
44 Methods, Meanings and Rules

A general metarule could generate correct results that would

otherwise have to be obtained by particular rules of the form:
*worksuffixed —» worked.
Such a metarule would include the linguistically more inter-
esting, but methodologically similar rule:
*dosuffixed-> did.
Since the rules were ordered, an important metarule safe-
guarded the consistency of the grammar by specifying that " i n
case of contradiction between two rules, the latter (in the order
of rules adopted by the grammar) prevails." Other logical terms
and distinctions necessitated by the grammar were also intro-
duced, e.g., the distinction between use and mention and
between object-language and meta-language.
In order to give an idea of the use and function of some of
these rules and rule-systems I shall discuss a simplified and
abbreviated derivation of the Sanskrit sentence:
devadattah pacaty odanam, (5)
" D e v a d a t t a is cooking r i c e "
(see Kiparsky 1982:2 and following; Staal 1982:8-9). The deri-
vation may appear to be arbitrary and ad hoc, but that is because
it cannot demonstrate the most powerful feature of the system:
namely, that methods of analysis that apply to this specific
sentence also apply to infinitely many others. Active sentences
such as (5) are not more basic than their passive counterparts
4
'rice is being cooked by D e v a d a t t a , " or related nominalizations
such as: " D e v a d a t t a , a cooker of r i c e , " " r i c e , cooked by
D e v a d a t t a , " " t h e cooking of rice by D e v a d a t t a , " " D e v a d a t t a ' s
cooking of r i c e , " etc. All of these were regarded as alternative
realizations of the same underlying structure. In each case, a
specific path of rules leads to a specific result. I shall follow one
such path.
In order to derive any sentence we have to take account of the
leading principle of the käraka system expressed by the metarule
2.3.1: anabhihite, "if not (already) e x p r e s s e d . " This means that
 The Origins of Linguistics in India 45

every kâraka relation must be expressed by the morphology, but

no käraka may be expressed more than once.
The derivation begins by selecting items from the lexicon and
deciding on a semantic relation between them. T o the nominal
items, gender and number are assigned, and to the verbal roots,
a time reference. We select devadatta- as " i n d e p e n d e n t a g e n t , "
odana- as " a g e n t ' s principal g o a l , " and for the verbal root pac-,
reference to "ongoing t i m e . " If we had chosen differently, we
could derive one of the related expressions mentioned before,
e.g., the passive.
The rules will now specify which abstract, "meta-linguistic"
or " g r a m m a t i c a l " markers have to be attached to the lexical
items. These enable us to attach subsequently the " r e a l "
nominal and verbal endings, but also entail the application of
other rules that are necessary for the correct forms to be
obtained. For the active form, an "infix" SaP and the ending tiP
are attached to the verbal root. (In expressing these markers I
use capitals, which could not have been done in the grammar of
Pânini which was orally transmitted and composed: see end of
Chapter 27). After the grammatical elements " 5 " and " / > " have
done their j o b , they are deleted (in accordance with a metarule
already referred to) and the form pacati is obtained. The " g o a l "
function is expressed by assigning the accusative ending -am to
the nominal item that bears that function. The " a g e n t " function
is expressed by assigning the instrumental ending -ena to the
nominal bearing that function.
N o w the leading principle that nothing is expressed more than
once comes into play. In the present derivation, the verb endings
have been chosen to express the agent. The goal is therefore not
yet expressed, and so it receives the accusative ending -am. The
agent, however, is already expressed and cannot now be as-
signed an instrumental ending. The ungrammatical form:

*devadattena pacaty odanam,

* " b y Devadatta is cooking r i c e "
is thereby correctly bjocked. Instead, a rule steps in which
assigns the nominative case when only the notion of nominal
46 Methods, Meanings and Rules

stem, gender and number remain to be expressed. Applying it to

devadatta-, the correct form devadattah results.
Indian linguistics did not maintain the high level that Pânini
had introduced. We still find it in Patanjali, three centuries later,
for he did not hesitate to criticize Pänini when the occasion
arose. But he also began to twist the use of Pânini's rules to
account for new forms that Pânini had not known because they
had come into being only later in the course of the natural
development of Sanskrit. Originally, Pânini 1 s grammar was
based on the spoken language of his compatriots. In due course,
it became prescriptive and came to be looked upon as an arbiter
of correct Sanskrit, thus arresting its natural development. By
that time the Indian sciences had entered a period of scholasti-
cism which corresponds to some extent to that of the Western
Middle Ages.
The notion of rule was permanently introduced into the
traditional sciences including philosophy. Its uses and varieties
increased and their discussion continued over the centuries. In
the thirteenth century C E . , a sütra was defined as "consisting of
few syllables, not leading to doubt, containing the essence of a
topic, fully expicit, without embellishment and faultless" (cf.
Renou 1956:53—59 and 1963). Such a "definition" may provide
beginners with a first inkling of the subject, but when it was
formulated, sciences dealing with rule-governed activities had
existed for almost two thousand years.
 6

Contemporary Linguistics and ReSes

Chomsky argued in his Cartesian Linguistics of 1966, that

Descartes introduced into European speculation on language the
view that human language does not merely serve a communica-
tive function, but is an instrument for the free expression of
thought and for expressing appropriate responses to new situa-
tions. C h o m s k y ' s thesis has been the subject of a series of
important reviews (Harman 1968, Zimmer 1968, Lakoff 1969,
Salmon 1969), some of them critical (especially Aarsleff 1970),
but its substantial correctness is not now in question. According
to Chomsky, the Cartesian emphasis on the creative aspect of
language use found its most forceful expression in Wilhelm von
H u m b o l d t ' s attempt to develop a comprehensive theory of
general linguistics. Von Humboldt emphasized that the infinity
of language is expressed by finite means, and in so doing
formulated the theoretical foundation for recursive rules without
formulating such rules himself. He played, however, an impor-
tant role in the reception of Indian ideas that led to the slow birth
of linguistics in the West.
The origin of the Western science of linguistics developed
from Bopp's adoption of the methods of the Indian grammarians
mentioned at the end of Chapter 4. Bopp's work immediately led
to a controversy which was, in essence, a discussion about
methodology and empiricism. After publishing his comparative
grammar, Bopp wrote a Sanskrit grammar, which he dedicated
to Wilhelm von Humboldt, and a " C o m p a r a t i v e Glossary of
S a n s k r i t . " August Wilhelm von Schlegel wrote in a letter to von
Humboldt of 1829: t 4 Mr. Bopp certainly has grammatical sense;
if only he had studied the Indian grammarians more diligently, if
48 Methods, Meanings and Rules

he did not always try to be original when it is inappropriate, he

would have been able to do something really g o o d . "
The real target of the criticism was Bopp's view that the
Indian grammarians had created many forms, " t h a t are not real
roots, but are the stems of verbs derived from n o u n s . " By " n o t
r e a l , " Bopp meant that these forms are not forms of the Sanskrit
language but had been postulated by the grammarians. The
majority of scholars dissented. Niels Westergaard, for example,
explained in 1841:

There are many roots we do not know from nouns derived from them,
and of which we cannot establish the use from classical passages and
books. For this reason there are people who claim that such roots have
never existed in the language at all, but have merely been invented—I
wouldn't know why—by the grammarians. But such an assertion is
astonishing, given the fact that so little is known of Indian literature (in
Staal 1973:54).

The methodological discussion between Bopp and Schlegel

about the issues of empiricism and postulated elements was
overshadowed by the dispute about the value of the study of the
Indian grammarians. We have seen that Schlegel defended the
need for such study which Bopp had questioned, at least to some
extent. The issue about empiricism was better understood by
von Humboldt who wrote to Schlegel, also in 1829:

In one respect, basic to Bopp's view, I differ from him. He regards the
roots as grammatical abstracts. But I consider them to be ancient words
' which disappeared when language Decame increasingly inflected. Î admit
that the Indian grammarians, who use them with few exceptions merely
as scientific tools, freed them from other sounds or subjected them to
other sound changes in order to transform them into pure sources from
which the forms of language originate. I am therefore always searching
for the appearance of these roots in Asian languages other than Sanskrit,
and have discovered something, but not a great deal (in Staal 1973:62).

This passage shows that von Humboldt accepted unattested

forms postulated by the Sanskrit grammarians not if they were
4
' a b s t r a c t e d , " "scientific t o o l s , " but only if they were " r e a l "
sources, that is, forms of Sanskrit, or another language, that
 Contemporary Linguistics and Rules 49

generate other expressions of the language. It is likely, there-

fore, that his ideas about the creative use of language were
inspired not only by "Cartesian linguistics" and by the general
emphasis on creativity that characterizes the Romantic period,
but were also influenced by the Sanskrit grammarians' creation
of artificial, metalinguistic elements which von Humboldt inter-
preted as "ideal f o r m s . "
During the second half of the nineteenth century the study of
Sanskrit began to flourish, first in India and later in Europe.
Ramkrishna Gopal Bhandarkar studied the historical develop-
ment of Sanskrit and determined which stage of its development
corresponded to the state of the language described by Pänini
and other grammarians. Kielhorn, Liebich and others contrib-
uted further to this kind of research. But there were also
sceptics, although not among specialists of the Indian grammat-
ical tradition. Theodor Benfey, for example, claimed in the
Introduction to his Vedic grammar of 1874 that the Indians had
left us on the one hand (in the Vedas), " t h e most beautiful
language without a grammar based upon i t , " and on the other
hand (in Pänini), " t h e most beautiful grammar without the
language upon which it is b a s e d . " The main proponent of the
view that there existed a "grammarians' Sanskrit" which was " a
thing of grammatical rule merely, having never had any real
existence as a language" was the first important American
Sanskritist, William Dwight Whitney (1827-1894).
Whitney emphatically rejected the " r e a l i t y " of the roots given
by the Sanskrit grammarians. In his own Sanskrit Grammar (first
published in 1879, and still regularly reprinted), he never ac-
cepted a root unless he had found it in a text. Needless to say,
Whitney's grammar did not follow Pänini's grammar, which his
Preface attacked for its "highly artful and difficult form of about
four thousand algebraic-formula-like rules in the statement and
arrangement of which brevity alone is had in view, at the cost of
distinctness and unambiguousness" (page xi).
The most important contributions by Pänini and the Indian
grammatical tradition were not the analysis of words into smaller
units, the postulation of abstract elements, or the view that
50 Methods, Meanings and Rules

language is more than the finite portion or corpus of data that we

come across, important as these insights are. Their greatest
contribution was the invention of the notion of rule that made all
these discoveries possible and accounted for other new insights
as weih The production and interpretation of infinitely many
forms, for example, can only be based upon a finite system of
rules. Early in the nineteenth century, von Humboldt had
understood the importance of "finite m e a n s " by which infinite
forms are generated; but he referred to rules only occasionally
and did not produce anything like a generative grammar. It took
another century-and-a-half before attempts were made to con-
struct such grammars, or fragments of grammars, with the help
of rules.
H o w could Whitney, an excellent Sanskrit scholar who was
also the leading American linguist of the second half of the
nineteenth century, have missed the central contribution of the
Indian grammarians? The reason is that in all his grammatical
work, including the Sanskrit Grammar of 1879, Whitney was not
inspired by Pänini but by the empiricism of the Prätisäkhya
literature that had been discarded by Pänini and his successors.
It is therefore misleading to claim that Whitney "did not
discredit and slight the old Hindu grammarians because of any
lack of acquaintance with them . . . He published not only the
Atharva-Veda-Prâtisâkhya but also a similar edition of the
Taittiriya-Prätisäkhya . . . " (Seymour in Sebeok 1966,1:416).
Whitney's studies of the Prätisäkhya literature s u p p o r t s . a n d
accounts for his narrow empiricism and explains his adherence
to their restrictive scientific method which happened to coincide
with the caricature of science we studied in Chapter 2.
That the approach of the Prätisäkhya literature led to Whit-
n e y ' s outlook in linguistics is supported by his own declarations.
Whitney checked the completeness of the statements of the
Prätisäkhyas with the Vedic corpus they purported to describe
and found, to his satisfaction, complete agreement. For exam-
ple, commenting on the passage of the Taittiriya Prätisäkhya
which described the conversion from dental n in the Padapätha
 Contemporary Linguistics and Rules 51

to retroflect n in the Samhitäpätha, he notes: " I have not

discovered in the Samhitä any case of a lingual nasal arising in
the conversion of the pada text into samhitä which is not duly
provided for in this c h a p t e r " (Whitney 1871:180; cf. also 281).
Such observations do not show that Whitney understood the
Sanskrit grammarians; on the contrary, they confirm that he did
not.
Whitney was very influential, especially in the United States,
and the empiricist outlook of twentieth century American lin-
guistics is not intelligible without that influence. In E u r o p e ,
knowledge of the Indian grammarians remained relatively strong
and the Romantic perspective that was beneficial to linguistics
continued for some time. Kielhorn's pupil Bruno Liebich, for
example, devoted a chapter of his monograph on Pänini to
Whitney's views and gave a spirited defense of Pänini's method.
He pointed out that a grammarian, who studies his mother
tongue, need not confine himself to forms encountered in texts,
for he is in a position to reproduce all possible linguistic forms
whenever he needs them by reflection—just as a scientist may at
any time obtain through experimentation what it would take him
a long time to find in nature.
Modern linguistics originated, finally, when de Saussure re-
acted against the diachronic considerations that characterized
historical and comparative philology and phonology, and advo-
cated a synchronistic approach that looked at language as a
system. This led to the rediscovery of structures that Pänini had
discovered two and a half thousand years earlier. De Saussure
was not impressed by the Sanskrit grammarians about whom he
knew little, because he associated them with the language which
was the bulwark from which diachronic philology had devel-
oped. But Pänini's method had been equally synchronistic as de
Saussure's albeit for a different reason: he looked upon Sanskrit
as eternal and unchanging. This illustrates in passing how
erroneous beliefs may lead to solid scientific achievements.
52 Methods, Meanings and Rules

6A. Noam Chomsky

Building on the vast amount of information that had been

collected by historical and comparative phonologists throughout
the nineteenth century, Trubetzkoy and Jakobson developed
synchronistic phonology which led to J a k o b s o n ' s discovery of
distinctive features, first published in 1938 but " n o t generally
discussed in the professional literature until the 1950s in spite of
the fact that Jakobson continued to write about them all through
the intervening p e r i o d " (Halle 1983:85-86). A little later, in the
late 1950s, N o a m Chomsky began to study grammatical rules
systematically and in a principled manner. Chomsky was influ-
enced by Zellig Harris, who attempted formalization although he
did not use rules, and was familiar with the uses of rules in logic
and mathematics. He discovered later that the notion of a rule as
a postulated, hypothetical entity, is closely related to the ratio-
nalist conception of human language which had also been the
Indian view and the opposite of Whitney's empiricist and
Prätisäkhya-inspired thesis that "language in the concrete sense
is the sum of words and phrases by which any man expresses his
t h o u g h t " (Whitney 1874:372 in Chomsky 1964:22).
Chomsky noted in the first book for which he became widely
known ("Syntactic S t r u c t u r e s " of 1957), that the more careful
proposals for the development of linguistic theory, due to Bloch,
Harris, Hockett, Wells and others, attempted to " c o n s t r u c t a
grammar of the language directly from the raw d a t a . " These
attempts were perfect embodiments of H u m e ' s empiricist view
in epistemology that every general hypothesis is a generalization
from observed instances (above, page 7). Chomsky added: " I
think that it is very questionable that this goal is attainable in any
interesting way, and I suspect that any attempt to meet it will
lead into a maze of more and more elaborate and complex
analytical procedures that will fail to provide answers for many
important questions about the nature of linguistic s t r u c t u r e "
(Chomsky 1957:52-53). Chomsky continued to'develop genera-
tive and transformational grammar and in so doing demonstrated
 Contemporary Linguistics and Rules 53

the inadequacies inherent in the empiricist account of scientific

method.
In the late fifties, Chomsky, presumably, did not know the
work of Pänini or the Indian grammarians, and the importance
they attached to rules. In 1961, he published an article " O n the
notion 'rule of grammar' ." in a volume of symposia in applied
mathematics entitled Structure of Language and Its Mathemat-
ical Aspects. Here he defined and discussed several concepts he
had already used in his linguistic work, such as context-free
rules, context-sensitive (or context-restrictive) rules, non-recur-
sive, left-recursive, right-recursive, nested dependency,
branches, self-embedding nodes, and degrees of self-embedding.
Of this battery of concepts, context-sensitive or context-restric-
tive rules were used by Pänini and defined and formulated in
essentially the same manner as was done by Chomsky (Staal
1965a; republished, 1988a: 171-180). The rules (2), (3) and (4)
(pages 38-9) are context-sensitive rules of this type. Through
systematic uses of this array of methods and techniques
Chomsky transformed the traditional European speculations
about language and the insights of philology, a respectable
branch of scholarship, into linguistics and brought this new
science to the forefront of contemporary scientific attention.
Chomsky postulated abstract structures and used rules to
derive from these first syntactic structures, and subsequently
two different levels of representation, one semantic and one
phonetic. Among the rules he used the so-called transforma-
tional rules became especially known. These rules are somewhat
similar to context-sensitive rules, but the context they depend on
is larger and more complex. In context-sensitive or context-
restrictive rules, the only context that is taken into account is
what immediately precedes or follows. This applies, for exam-
ple, to the "initial v o w e l " immediately following the element
that undergoes substitution in rules (2), (3) and (4).
In the sixties and after, transformational rules were used to
account for a large variety of facts in syntax and phonology.
They have now been replaced by other mechanisms such as case
marking, thematic role assignment, binding of variables, etc.
54 Methodsy Meanings and Rules

Whatever their precise form, such mechanisms which involve

rules of varying complexity, must account for facts that depend
on distant contexts such as pronominalization. In the four
English sentences:

The fact that I hate John does not worry him

The fact that I hate him does not worry John
John is not worried that I hate him
He is not worried that I hate John,

The pronoun " h i m " in the first three refers to " J o h n , " but the
pronoun " h e " in the fourth does not. Such simple facts known
to every user of the language exhibit complex relationships that
are now explained by the rules of "binding t h e o r y " (Chomsky
1986:164-184). I shall continue to refer to transformations since
they can be represented by simple tree-shaped diagrams or
" t r e e s , " which can also be used to represent ritual structures
(Chapters 10-12). To what extent the newer mechanisms apply
to ritual structures can be decided when more is known about
both.
Some critics argued that transformational rules are simple and
ordinary and express relationships of a familiar type that is only
to be expected. The philosopher Hilary Putnam claimed that
they are " a b b r e v i a t i o n s , " and the linguist Charles Hockett that
they are " a n a l o g i e s " (page 438). These criticisms are well taken
provided we postulate that these " a b b r e v i a t i o n s " or " a n a l -
o g i e s " have all the extraordinary properties that Chomsky had
used to define transformations. For there is nothing obvious or
natural about transformational rules or any of the newer rules
that have taken their place.
Chomsky has emphasized, quite to the contrary, that there is
no a priori necessity for human languages to be organized in this
complex way. The specific rules and structural properties that
characterize phonology and syntax are nontrivial, far from
obvious, and restricted by abstract principals of a highly restric-
tive, complex and intricate kind. Insight in this complexity
constitutes one of the most remarkable general results of con-
temporary linguistics. It demonstrates that the human mind is
 Contemporary Linguistics and Rules 55

not the tabula rasa that might be expected if the brain were an
empty bowl.
With its myriad rules, the Astädhyäyf supports the result that
human language is not a simple or trivial device. Pänini's
grammar is a generative grammar, as we have seen. It is not a
transformational grammar but it introduced levels of analysis
and distinguished between " d e e p " or underlying structures and
" s u r f a c e " structures derived from them with the help of rules
(Kiparsky and Staal 1969). However, Pänini's grammar and the
generative grammars or fragments of grammars that have been
written in the wake of C h o m s k y ' s discoveries—foremost among
them, Chomsky and Halle's Sound Structure of English of
1968—are so similar in spirit, method, organization and results,
that a conclusion follows which goes beyond the achievements
of linguistics or any other particular science: namely, that
linguistics and therefore science is not a Western invention but
a universal discovery of humanity.

6B. Chomsky versus Wittgenstein

Rules have also been discussed in contemporary Western

philosophy and I shall end this outline with a brief reference to
the resulting controversies. Wittgenstein was sceptical about the
possibility and the precise meaning of "following a r u l e . " His
account of these problems is meandering and confusing, but it
starts out clearly. He begins (cf. Philosophical Investigations
1953:185) with simple cases, e.g., a pupil is taught to write down
a sequence of numbers:

2 , 4 , 6 , 8, 10, 12 . . . .

Wittgenstein asks: how far does the student have to go in

order for us to say that he has "understood the r u l e " ? The above
numbers may seem to suffice, but only at first sight. Let us
suppose we have doubts. We will give him exercises, say, up to
1000. N o w we ask him to continue and he writes:

1004, 1008, 1012, 1016, . . . .

56 Methods, Meanings and Rules

" W e say to him: ' L o o k what y o u ' v e done?'—-He doesn't

understand. We say: 'You were meant to add two: look how you
began the s e r i e s ? " But he apparently assumed that the rule
meant: " A d d 2 up to 1000, 4 up to 2000, 6 up to 3000, and so o n . "
I shall not follow the lengthy discussions which Wittgenstein
devoted to such cases and which have led to an extensive
discussion in contemporary philosophy. It would be tempting to
say that their importance is restricted by the circumstance that
Wittgenstein remained, until the end of his life, " a boy doing
s u m s " (the expression is adapted from Anscombe 1959:171).
Recently, Wittgenstein's puzzles have been analyzed by Saul
Kripke (1982), and this analysis has in turn been discussed by
Chomsky (1986). I have no wish to even try to improve upon
these admirable interpreters, and just as Chomsky assumed that
Kripke's Wittgenstein was the genuine article and referred to
him as " W i t t g e n s t e i n , " I shall extend the assumption and refer
to C h o m s k y ' s Kripke's Wittgenstein as " W i t t g e n s t e i n . " In
C h o m s k y ' s words, Wittgenstein's "skeptical paradox about
r u l e s " is briefly the following:

Given a rule R, there is no fact about my past experience (including my

conscious mental states) that justifies my belief that the next application
of R does or does not conform to my intentions. There is, Wittgenstein
argues, no fact about me that tells me whether I am following R o r R ' ,
which coincides with R in past cases but not future ones. Each
application of a rule is 'a leap in the dark.' My application of a rule is 'an
unjustified stab in the dark. I apply the rule blindly' (Chomsky 1986:225).

What is formulated as if it were C h o m s k y ' s chief critique

sounds uncharacteristically subdued: " T h e r e may b e " (with
reference to following a rule) " a question as to how we do it, but
there seems little doubt that we do it. Furthermore, none of this
seems to have much if any utility in our lives" (Chomsky 1986:
229). That we follow rules is clear, and Wittgenstein seems only
to have argued that we cannot give a clear conceptual account of
this. " U t i l i t y , " however, is not an important consideration in
the evaluation of scientific theories. C h o m s k y ' s substantial
critique is contained in his comment on a simple example:
 Contemporary Linguistics and Rules 57

At a certain stage of language growth, children characteristically over-

generalize: They say sleeped instead of slept, brang (on the analogy of
sang) instead of brought, and so forth. We have no difficulty in
attributing to them rules for formation of past tense, rules that we
recognize to be different from our own (Chomsky 1986:227).

This passage shows several things. First of all, the children

who use these forms may not know what rules they are follow-
ing. They may not be aware of them in the same way in which we
are aware of them. It is not certain, moreover, that they form
brang on the analogy of sang. But it is the best hypothesis we
can come up with and in that respect not different from any
scientific hypothesis, only simpler than most. Like all others, it
may have to be rejected as soon as we find a better theory.
N o w as to our own awareness of what rules we follow
ourselves. We are not ordinarily aware of the fact that when we
use slept we are deviating from a general rule which describes
that the past tense in English is formed by affixing the ending -ed.
Perhaps we were aware of this at first, when we had just realized
that slept is irregular in this manner. Later we may have
forgotten about it although if anyone were to ask us about the
rule we are following this is probably what we would say.
However, we may be wrong in the same sense with respect to
the rule we are following as with respect to the rule the children
were following. Our guess has the character of a scientific
hypothesis, and if a better one is forthcoming, we would accept
it. For example, a much more general rule may account for the
formation of the past tense of verbs such as sleep in English.
Wittgenstein demonstrated that we cannot be absolutely cer-
tain about a rule that we are following in our more or less
unconscious usage of language. However, we can never be
certain about unobservables, about the future or the past. The
sun is likely to rise to-morrow but then, it might not. Perhaps it
did not rise around 1000 B . C . E . on the day that Joshua fought the
battle of Jericho when the sun and moon stood still. The
uncertainty about rules is not different from the problem of
induction and from Nelson G o o d m a n ' s puzzles already referred
to (above, page 8). Measures of probability and certainties of
58 Methods, Meanings and Rules

less than 100% are marks of observables and unobservables and,

since we cannot be certain, we are entitled to be sceptical. We
do the best we can; that is, we look for the most likely
hypothesis. As soon as a likelier one than the one we now
possess has been found, the one we possess is no longer the most
likely and must be rejected. This insight is as old as Plato's
Timaeus.
Many interesting things may be said about rules and d o n e with
them as Pänini and Chomsky, among others, have demon-
strated. Wittgenstein's scepticism is concerned with our aware-
ness or knowledge of rules; it tells us nothing special about rules.
It is anyway not clear what Wittgenstein means by a clear
account. In the areas we shall study in Parts II and III, I am
involved with rules, but the clearest account I can come up with
is often not very clear, which does not imply that we should
simply abandon those areas.
Rules are important and interesting because they express
regularities. When we come across regularities we postulate
rules to account for them. There are similarities and differences
in this respect between different areas of experience: in physics,
biology and the human sciences. Rules are in some respect like
the principles and laws we postulate in the physical sciences to
account for observed regularities. One difference is that we may
be conscious of applying or following rules. But often we are
not, and in those cases we are in a position to guess or postulate.
Since there has been some confusion about " t h e meaning of
r u l e s " I shall end this chapter with some comments on rules of
semantics. These assign meanings to linguistic expressions and
although not too much is known about them, they seem in some
respects similar in type to the rules of phonology and syntax. We
may ask what is the meaning of such rules. But that question is
not different from such questions as what the rules say, what
their functions and domains of application are, what they
effectively do, etc.
H o w do semantic rules assign meaning? There is no general
answer to that question. Different expressions of a language such
as English get meanings assigned to them in different ways. A
 Contemporary Linguistics and Rules 59

few words refer to things, e.g., the expression "this c h a i r "

generally refers to a particular thing (but not in the very sentence
I am now typing on the key-board). " A c h a i r , " " a n y c h a i r , "
" s o m e c h a i r " are different; they seem to be some kind of
stand-ins or variables. " T h e c h a i r " may even refer to a Platonic
idea. " C h a i r s " can do many of these things.
By contrast, " a n d " never refers to a thing. It is defined in
context, or recursively: that is, " A and B " is defined in terms of
" A " and " B . " Hence we have to postulate a metalinguistic and.
But even if this analysis seems to apply to the " a n d " in " J o h n
and Mary went for a w a l k " it does not apply, at least not in the
same manner, to the " a n d " in " J o h n and Mary m a r r i e d . " The
case of " J o h n and Bill married" is different again, in most states.
" S l e p t " derives its meaning from " s l e e p " but it refers to a
past event. " I n " often expresses situation or locality, and some
things Pänini said about the locative in Sanskrit apply to " i n . "
But not in such expressions as "in spite o f ' where the three
constituent words contribute to the meaning in a manner that is
different from most other words.
A system of rules that assigns meanings to the expressions of
a language is therefore a rich and complex system, and need not
be less complicated than a branch of astrophysics. The main
difference seems to be that, for the time being, the branches of
astrophysics have received more attention and nothing exists so
far that even comes close to a complete semantics for English.
For Sanskrit, we have at least Pänini who is surprisingly, though
not entirely, complete (cf. Thieme 1982/83:6-9). For English
semantics, we have a discussion of many interesting cases in
such works as Quine 1960. For English phonology, we have a
mine of information in Chomsky and Halle 1968 but even this
book is primarily interested in facts that shed light on general
linguistic theory and does not discuss irregularities and excep-
tions.
There are many kinds of rules, and innumerable rules that are
postulated to account for the ways meanings are conveyed. But
we have learned that not everything in the universe must of
necessity have a meaning. We don't ask zoologists about the
60 Methods, Meanings and Rules

meaning of elephants. To refer to the meaning of rules is not

prohibited but can easily be misleading and confusing, and it is
clearer and more helpful to our understanding to say that such
rules as we have discussed are without meaning. Another way of
saying this is that the logic of " m e a n i n g " and the grammar of
" m e a n i n g " are not the same; and the same holds for " r u l e s . "
Their being meaning rules implies that they are about meaning
but it does not imply that they have meaning. These being cups
for drinking implies that we drink from them but it does not
imply that we drink them.
 Part II

Ritual

Introductory Note

The account of ritual that will be outlined here is based upon

the study of a specific type of ritual: Vedic ritual. This raises a
number of questions. For example, to what extent does the
nature of this particular ritual influence or limit the nature of the
theory? I shall revert to these questions later when their an-
swers, given the proper context, will be more easily recognized
(see Chapter 30C). H o w e v e r , a related problem should be briefly
referred to at the outset. What is ritual? Don't we have to start
with a definition of ritual before we can understand it or pave the
way for a theory of ritual?
The demand for such a definition seems straightforward and
uncontroversial; and yet, it is based upon confusion. " R i t u a l " is
a fashionable term, and many things have been called ritual. The
power of Shakespearean drama is called " r i t u a l , " curses and
dances are regarded as ritual, and in any museum of ancient art,
items of which the use or function is not known are dubbed
"ritual o b j e c t s . " Jack Goody (1977) has shown that anthropol-
ogists have called almost anything ritual. He refers to an author
who includes among rituals, " c o r o n a t i o n s , funerals, Christmas,
d a n c e s , football, theatre, gymnastics, brass-bands, pop festi-
vals, student d e m o n s t r a t i o n s . " Adds Goody: " I n c l u d e elec-
tions, schools, work groups and the rituals of family living, and
you have covered much of social life in Britain t o - d a y " (1977:
26). Goody concluded that the term " r i t u a l " is best avoided.
The confusion I referred to does not lie in the uncertainty that
surrounds and almost seems to characterize the extension of the
concept of " R i t u a l . " The confusion lies in demanding first, a
62 Ritual

définition and assuming that a theory of the defined object can be

given only afterwards. It is possible that there exist areas of
experience where such a relationship between definitions and
theories prevails. Examples come to mind from mathematics:
having defined a circle, we may ask what its properties are or
derive theorems about its relation to other similarly defined
figures. In general, however, especially in an empirical science,
the situation with regard to an unknown area of experience is
different. We have intuitions about what rituals are but we do not
precisely know what they are. In order to know it we need a
theory first, and not the other way round. Different people have
different intuitions even though there may be a common core or
a family resemblance upon which linguistic usage is based.
When we read: " t h e president's speech was ritual," we vaguely
know what is intended here: the president went through the
movements of giving a speech because the circumstances re-
quired it but did not say anything important or significant. It is
not clear that it will be profitable to add all such cases of
linguistic usage onto a heap, and seek to extract a definition for
what they have in common. My guess is, that they have nothing
nontrivial in c o m m o n , that even a family resemblance is hard to
establish and that the definition, if it can be derived, will turn out
to be uninteresting. Dictionaries, which use such methods,
illustrate this lack of insight. The Concise Oxford Dictionary, for
example, defines " r i t u a l " in terms of " r i t e , " and says about
" r i t e " : " a religious or solemn o b s e r v a n c e . " What then are we to
do?
First of all, we should be led by our intuitions for those are all
we have; but they should be based upon familiarity with the area
in which we expect to find something. We should not concen-
trate on borderline or doubtful cases; these can be dealt with
later. We will, accordingly, reject most of the cases discussed by
Goody: a few people have called these rituals, many have
rejected them and we don't know our way about. We should
then select some clear and üncontroversial cases about which
everybody agrees that they are ritual. For these we should try
develop a theory, and this can be done by any means within our
 Ritual 63

power. For theories cannot be " d e r i v e d " from data as in the

caricature of Chapter 2; they are inspired by ideas, musings, and
other exercises of the imagination. Once formulated, they
should be clear enough to be subjected to tests and experimen-
tation.
If we are unable to develop a theory, this may simply be due
to personal limitations; but there may be more interesting
reasons. For example, there may not be a single concept to
which our intuition refers. For even though we should start with
intuitions in an area where nothing else is available, intuitions
are not infallible. The concept of "religion," for example, for
which most Western languages have a word and for which we
possess, accordingly, an " i n t u i t i o n , " may not be a universal
concept; hence it may not be possible to set up an interesting
theory of religion (Chapter 28).
Once we have arrived at a theory which accounts for an area
of experience that is generally accepted as falling within the
domain of a word—say, " r i t u a l " — w e are in a position to check
our intuitions and consider the borderline cases. If we find that
some are accounted for by our theory, we include them and
accept them as rituals; if not, we reject them. For example, if we
find that, according to our theory, commencement exercises are
rituals, but football matches are not, we are in a position to
discuss some advantages and disadvantages of our theory. If the
final result is satisfactory and exhibits general features or
structures that are interesting and lead to further investigations,
insights or predictions, we have reason to be optimistic. Perhaps
we can, at that stage, set up a definition of " r i t u a l " that is in
accordance with the theory and that enables us, if it is " o p e r a -
t i o n a l , " to recognize whether any object presented to us is or is
not a ritual.
Bertrand Russell has summarized this approach to theory
formation in simple terms: "Instinct, intuition, or insight is what
first leads to the beliefs which subsequent reason confirms or
confutes" (1953:19). This implies that definitions do not come in
the beginning, but in the end. Definitions, moreover, may be left
implicit in a theory: for to explicate them—which is always
64 Ritual

possible in accordance with a logical theorem due to Evert Beth

and generalized by William Craig (see Beth 1959:288-293)—may
yield nothing but unmanageable phrases.
So we shall proceed as follows. We will start with Vedic ritual
which comprises data of which no one has denied that they come
under ritual. There are, moreover, Indian terms which demar-
cate this domain and distinguish it from other things (e.g.,
Sanskrit yajna). We shall discuss characteristics and properties
of Vedic ritual and then sketch a proposal for a theory. Subse-
quently we shall try to find out whether the proposed hypothesis
is applicable to other undoubted rituals, and to things of which
some people have claimed that they are rituals, although others
have been doubtful. This latter undertaking will be an extensive
enterprise, and we shall be concerned only with a first beginning.
N e w data will be needed to test our hypothesis and, if necessary,
change or abandon it.
 Vedic Ritual

More than three thousand years ago, small groups of semi-

nomadic peoples crossed the mountain regions that separate
Central Asia from Iran and the subcontinent of India. They
spoke an Indo-European language, which developed into Vedic,
and imported the rudiments of a social and ritual system. Like
other speakers of Indo-European languages, they celebrated
fire, called Agni, and like their Iranian relatives, they adopted
the cult of Soma—a plant, probably hallucinogenic, which grew
in the high mountains. The interaction between these Central-
Asian adventurers arid earlier inhabitants of the Indian subcon-
tinent gave birth to Vedic civilization, named after the four
Vedas, oral compositions that have been transmitted by word of
mouth to the present day. The Vedas depict Vedic religion, in
the words of Louis Renou (1953:29), as "first and foremost a
liturgy, and only secondarily a mythological or speculative
system."
The four Vedas are directly connected with the ritual they
depict because portions of three of them are recited and chanted
as part of these ritual performances. The first of these three is the
Rigveda, a collection of poems, hymns and invocations re-
flecting myths, rites, battles and insights of many kinds; its
verses are recited. The second is the Sämaveda which consists
mostly of parts of the Rigveda set to music; its songs are
chanted. The Yajurveda consists of verse and prose: both forms
are recited in the ritual, the latter being largely devoted to the
description and interpretation of its acts. The three Vedas were
transmitted orally (Chapters 5 and 27). The fourth or Atharva-
veda, in some respects similar to the Rigveda, probably reflects
66 Ritual

a different stratum of society. It is only distantly related to Vedic

ritual.
There are two basic types of Vedic ritual: Srauta and Grhya.
The former is generally older and more complex; it is sometimes
called " t r a d i t i o n a l , " " p u b l i c " or " s o l e m n . " The latter com-
prises domestic or life-cycle rites such as the ceremonies per-
formed at birth, initiation, marriage or death. The Srauta rituals
were primarily dedicated to Agni and Soma. Offerings of clari-
fied butter and other vegetable and animal substances, including
libations of juice extracted from the stalks of the Soma plant,
were poured or thrown into sacrificial fires installed on altars.
The recitations and chants that accompany the ceremonies are
generally derived from sentences or bits and pieces from the
Vedas and are called mantras, a term originally used to distin-
guish the versified sections of the Vedas from the prose passages
called brähmana. They will be studied in Part III.
The simplest Srauta ritual is the Agnihotra, still performed by
several hundred brahmins in India. It consists of an oblation of
milk offered every morning and evening by a householder with
the optional assistance of a priest. Many Srauta performances
required the execution of multifarious activities by up to seven-
teen priests, each attached by birth to one of the branches or
schools into which the Vedas were subdivided. The rites were
performed by the priests on behalf of and for the benefit of a
Yajamäna or "ritual p a t r o n , " a distinguished personage (some-
times a king). The contribution of each priest or group of priests
to each of the Vedic rituals was formulated in precise detail in a
manual that belonged to his school. These manuals, the Srauta
Sûtras, composed between approximately the eighth and the
fourth centuries B . C . E . , are the primary source for our knowl-
edge of Vedic ritual. They will be studied in Chapter 26.
One of the most elaborate Vedic rituals was called Agnica-
yana, the "piling of A g n i . " The verbal root ci- from which
cay ana is derived means "pile up, arrange, order, c o n s t r u c t . "
During the performance of this ritual, which originated around
1,000 B . C . E . , a large altar in the shape of a bird was piled up from
more than a thousand bricks in five layers, called citi (from the
 Vedic Ritual 67

same verb), of approximately two hundred each. The bricks fell

into specific groups, had different names and shapes and were
each deposited in a specific order and " w i t h a specific m a n t r a , "
that is, while a specific mantra was recited. It we had to
characterize this ritual in terms of its historical origin, we should
say that it constituted a purely Indian development, distinguish-
able in this respect from the cult of Agni which was Indo-
European and from the cult of Soma which was Indo-Iranian.
Though Vedic ritual contained popular as well as more exotic
or extravagant elements, it was confined to an elite of profes-
sionals who spent much of their life learning and preserving its
oral tradition. These experts became the brahmins of India.
Performances must have been relatively common for at least five
centuries, from about 1,000 to 500 B . C . E . Then Vedic ritual
began to decline and new cultures and traditions rose to prom-
inence. Though Hinduism and Buddhism are replete with Vedic
elements, they belong to another age, " H i n d u i s m " (put between
quotes because it is not a single or unitary unit: see below, pages
397-398) continued to look upon the Vedas as its source,
formally transmitted by the brahmin members of its highest
caste, regarded as an eternal revelation, " o f nonhuman origin,"
and no longer understood. Among later philosophies, the
Vedänta or " E n d of the V e d a s , " reacted against the ritual and
subordinated it to philosophy, but it retained ritual elements and
continued to bask in the ancient glory of the Vedas. Buddhism
rejected the Vedic heritage, the authority of brahmins and the
supremacy of Vedic ritual. In due course it evolved its own
hierarchies and ceremonies, incorporating again Vedic rites.
Vedic fire ceremonies became part of the Buddhist tradition and
were exported all over Asia. The " p i l i n g " (cayana) of the Vedic
fire altar was transformed into the Buddhist caitya (from the
same root ci-) or stüpa (Tibetan: chöten). Originally a funeral
pile, it became the primary cult-symbol of Buddhism throughout
Asia.
Fragments and features of Vedic rites survive to the present
day in India in the Grhya or domestic ceremonies of high caste
Hindus, for example, in marriage ceremonies. Performances of
68 Ritual

the larger Vedic Srauta rituals were revived several times, e.g.,
during the Gupta period in North India and during the Chola and
Vijayanagara dynasties of the South. However, these extensive
rituals are heard of less and less. A text of the eleventh century
declared that they were no longer performed.
India, however, is a land of miracles. In Kerala, a distant
corner in the southwest of the subcontinent, far away from the
original home of Vedic civilization in the northwest, a few
families among the isolated and orthodox community of Nam-
budiri brahmins have maintained their traditional Vedic heritage
and continue to perform two Soma rituals: the Agnistoma, or
" P r a i s e of A g n i , " which lasts five days, and the Agnicayana or
" A g n i , " which lasts twelve days and continues through some of
the nights whence its other name: Atirätra or " o v e r - n i g h t . " The
Nambudiri performances are not artificial or scholarly recon-
structions (as have taken place in some parts of northern India),
nor are they the results of recent revivals. As in other parts of
southern India, the tradition is authentic and alive. The rituals
are performed as they have been learned from teachers, who
have learned it from their teachers, and so on. We are able to
reconstruct the lineage until we reach the ancient Srauta Sütras,
from which contemporary performances deviate only in detail,
thereby confirming what is also obvious from other sources:
namely, that the contemporary tradition is based upon the same
tradition as is reflected in these texts.
One such Atirätra-Agnicayana was performed in a small
village in Kerala in 1975. For the first time it was attended by
outsiders, filmed, photographed, recorded and extensively doc-
umented. Robert Gardner and I produced from 20 hours of rough
footage a 45-minute film, " A l t a r of F i r e . " A more extensive
version is presently being prepared in collaboration with Philo
Bregstein. With the help of the two chief Nambudiri ritualists,
C. V. Somayajipad and Itti Ravi Nambudiri, I prepared a
definitive account of the ceremonies as they took place in 1975,
illustrated with photographs by Adelaide de Menil. This account
was published in two volumes in 1983 under the title: " A g n i —
The Vedic Ritual of the Fire A l t a r . " This book, which will be
 Vedic Ritual 69

referred to as A G N I , contains the empirical data upon which

most of the discussions, proposals, hypotheses and theories, of
the following pages are based.
Vedic ritual is not only the oldest surviving ritual of mankind;
it also provides the best source material for a theory of ritual.
This is not because it is close to any alleged "original'' ritual.
Vedic ritual is not primitive and not an (/r-ritual. It is sophisti-
cated and already the product of a long development. But it is
the largest, most elaborate and (on account of the Sanskrit
manuals) best documented among the rituals of man. Hubert and
M a u s s , who noted these facts in 1899, used the Vedic animal
sacrifice as source material for the construction of a ritual
paradigm ( " u n scheme abstrait du sacrifice") in their " E s s a i sur
la nature et la fonction du sacrifice." H o w e v e r , they did not
know that these rituals are still performed, and many data were
inaccessible to them. They also erred in concentrating on
sacrifice, a term that I shall confine to a ritual in which an animal
is ritually killed. But all ritual is not sacrifice. The Vedic Srauta
rituals consist of a hierarchy: four basic rituals that involve
oblations of milk, rice or barley (the most important of these are
the Full- and New-Moon Ceremonies); one ritual that involves
animal sacrifice; seven varieties of Soma rituals; and finally the
Agnicayana which incorporates one or more occurrences of
most of these others (AGNI 1:40-54).
There are other ritual traditions in which animal sacrifice plays
a relatively subordinate role. Foremost among these is the Taoist
ritual tradition, which vies with the Vedic in its degree of
elaboration and complexity and is also equally well documented.
Although not as old as the Vedic, it preserves and incorporates
earlier ritual traditions of China. It also survives to the present
day, in particular in Southern Taiwan where it was discovered
and studied by Kristofer Schipper. This tradition is compared
with the Vedic in other publications (Schipper and Staal 1988
and forthcoming).
Because of the Indian and Chinese facts, I have adopted the
general term " r i t u a l " rather than the more limited notion of
"sacrifice." This is in accordance with the prevailing usage in
70 Ritual

anthropology, at least in most of its English manifestations.

Contemporary theorists on ritual from other backgrounds some-
times use the term "sacrifice." Walter Burkert has explained
this preference but is aware of its possible limitations:

Turning to the special problem of sacrifice, and thus limiting the scope of
inquiry, let us concentrate on the area of study with which most of us are
famiiiar—namely animal sacrifice in Jewish, Greek and Roman religion,
including its sublimation in Christian theology. This does not forbid us to
look for similarities and divergences in other cultures, though such
evidence will not decisively corroborate or falsify the findings. Whether
the sacrifice we choose is a typical or a very special case is another
question. Ï do not wish to contradict Frits StaaFs claim that the most
promising field for ritual studies is the Indian and Chinese tradition
(Hamerton-Kelly 1987:162).
 8

The Agnicayana Ritual

Since the Agnicayana-Atirätra will provide us with the bulk of

our empirical data, a brief description of this ritual is in order (cf.
the " B i r d ' s - e y e view of the Agnicayana" in A G N I 1:55-58).
The ritual takes place inside an enclosure within which various
altars are constructed:

J\ / /
H
M

Y B

Figure 1
The Ritual Enclosure

The smaller enclosure to the west contains at its western

extremity the circular domestic altar on which the domestic fire
will be installed and on which the oblations are cooked. At its
eastern end is the square offering altar on which the offering fire
will be installed: this is the fire in which the oblations are offered.
72 Ritual

T o the south is the semi-circular southern altar 'on which the

southern fire will be installed.
The larger enclosure to the east, trapezoid in shape, is
constructed later; it will be referred to as the Great Altar space.
At its eastern end, a new offering altar will be constructed in the
shape of a bird with its head facing east. The old offering altar in
the western enclosure will now become the new domestic altar,
and fire will be carried from the old to the new offering altar and
installed there. The two rectangular areas inside the Great Altar
space have specific functions: the larger one, to the west, is the
hall of recitation; and the smaller one, to the east, is connected
with the preparation of Soma and called the Soma hall.
The Agnicayana performance of 1975 took twelve days and
the main rites were distributed as follows.

First Day
The Yajamâna and his priests enter the ritual enclosure in an
introductory procession, carrying in clay pots the three sacred
fires the Yajamâna has kept burning in his home. The ukhä pot,
main ritual vessel of the Agnicayana, is prepared from clay,
along with the heads of a horse, a man, a bull, a ram and a
he-goat, as well as a few of the bricks that will be used for the
piling of the alter. An animal sacrifice is performed for Vâyu, the
wind. The five chief priests are selected:
the Adhvaryu who performs the majority of rites;
the Brahman who, generally, sits by the side of the
Yajamâna, supervising the rites but without partici-
pating;
the H o t â who recites from the Rigveda;
the Udgätä who chants from the Sâmaveda;
the Sadasya who is in charge of the hall of recitation.
The priests formally enter the smaller enclosure; this is called
adhyavasäna. Fire is produced by friction from two pieces of
wood. An isti rite is performed: this rite reflects the basic pattern
of the full- and new-moon ceremonies, and is used on various
occasions (Chapter 9A). Here it marks the consecration (dïksâ)
 The Agnicayana Ritual 73

of the Yajamäna, which follows immediately after it. During his

consecration, the Yajamäna crawls onto the skin of a black
antelope, a turban is tied around his head, he is given a staff, and
he closes his fists. F r o m the consecration on the first day until
the final bath on the twelfth day, the Yajamäna should sit on the
antelope skin and carry the staff. Neither he nor his wife should
go outside the ritual enclosure. The Yajamäna should generally
keep his fists closed and refrain from speaking (except for the
prescribed recitations), from bathing, sexual intercourse, and
certain kinds of food. At the end of the consecration, the
Yajamäna picks up the ukhä pot, which is filled with fire, and
takes three steps with it. H e wears a golden breast plate.

Second Day

The mahavïra pot is prepared from clay. This vessel will be

used during the repeated performances of the Pravargya rite, a
pastoral ceremony during which hot milk is prepared and offered
to the As vin twins, divine young men w h o ride the wind.

Third Day

A tree is cut and a sacrificial pole is made from it. Animal

victims will be tied to it and other rites performed around it. T o
the east of the old enclosure, in which the three altars have
already been made, the measurements of the Great Altar space
and of the bird-shaped offering altar are laid out (see Figure 1).

Fourth Day

In the place of the old offering altar, the new domestic altar is
piled up from five layers, each consisting of 3 x 7 rectangular
bricks. The Adhvaryu consecrates each brick on behalf of the
Yajamäna with mantras. T h e offering fire is installed on the
domestic altar and fire from the ukhä pot is added to it. An
introductory isti (prayanïyestï) is p e r f o r m e d / O u t s i d e the enclo-
sure, Soma stalks are bought from a merchant. They are
74 Ritual

measured, placed on a bullock skin, and transported on a Soma

cart. King Soma is installed on a throne to the south of the new
domestic = old offering altar, and a guest isti (ätithyesij) is.
performed in his honor. The Yajamäna and his priests touch a
plate with clarified butter, ceremonially pledging not to harm
each other (tänunäptra). The Yajamäna and his priests, except
the chanters of the Sämaveda, sprinkle the Soma ^vith mantras;
that is, they recite mantras over it which mention the act of
sprinkling. The Subrahmanya priest of the Sämaveda recites for
the first time his invitation to Indra, king of the gods, other gods
and brahmins, urging them to attend the forthcoming Soma
pressing on the tenth day. The first Pravargya is performed,
followed by a sequence of rites called Upasad, followed in turn
by lesser rites including Nihnava or ''Hiding R i t e " (Frontis-
piece, pages xv, 85 and 93). The ground is prepared for construct-
ing the new offering altar. It is plowed and seeds are sown.
Several items are buried in the ground, including a replica of a
tortoise, the ukhä pot, a golden image of a man and the five
animals heads. Two hundred bricks, of various shapes and sizes,
are placed on the ground in specific order, making up the first
layer of the bird altar. A 4 'naturally perforated" or " p o r o u s "
stone is placed in the center. All bricks are consecrated with
mantras by the Adhvaryu on behalf of the Yajamäna. The
Pravargya and Upasad ceremonies follow.

Fifth Day

After the morning Pravargya, Upasad, Nihnava, etc., the

second layer of 200 bricks is laid and consecrated, followed by
the evening Pravargya, Upasad, Nihnava. etc. The shapes and
configurations of the bricks have one pattern on the first, third
and fifth layer; and another on the second and fourth.

Sixth Day

After the morning Pravargya, Upasad, Nihnava, etc., the third

layer is laid and consecrated, with a naturally perforated stone in
 The Agnicayana Ritual 75

the center, followed by the evening Pravargya, Upasad, Nih-

nava, etc.

Seventh Day

After the morning Pravargya Upasad, Nihnava, etc., the

fourth layer is laid and consecrated, followed by the evening
Pravargya, Upasad, Nihnava, etc.

Eighth Day

After the morning Pravargya etc., the fifth layer is laid with a
"naturally perforated" stone in the center. The Yajamäna
recites a sequence of mantras which express the wish that the
bricks turn into cows. A long, continuous oblation of milk,
sesame seed, and other substances is made for Rudra and the
Rudras on the westernmost brick of the northern wing of the
bird. T h e Udgätä sings chants around the bird, and a strong man
pours a continuous stream of water from a pitcher three times
around the altar. The evening Pravargya, Upasad, and Nihnava
follow.

Ninth Day

After morning and evening Pravargya, Upasad, and Nihnava,

now performed for the last time, the mahâvïra pot and other
implements used in the Pravargya are placed on the new offering
altar in the shape of a man. Agni fire is carried forth in a pot from
the new domestic to the new offering altar and is installed in its
center. A long, continuous oblation of clarified butter is made
into this new offering fire through a large bamboo ladle. This
oblation, called " F l o w of W e a l t h " {vasor dhärä), is followed by
numerous other offerings and oblations. The hall of recitation is
prepared along with its six hearths, each associated with a priest.
N o w Agni a n d . Soma are carried forth together and the
Agnïsomïya animal sacrifice is performed. The Subrahmanya
priest, standing between the Yajamäna and his wife, chants his
invitation to the Soma pressing for the last time.
76 Ritual

Tenth to Twelfth Day

The tenth day is the Soma pressing day, and from now on
ceremonies will continue throughout the next two days and
nights. After the H o t ä ' s morning recital (prätaranuväka), which
starts long before sunrise, the morning pressing of Soma begins
and the first Soma oblations are offered. T w o chanters, four
other priests and the Yajamâna crawl in snakelike procession
onto the altar, where they make an oblation into the offering fire.
They then move to a place to the north of the altar where three
chanters sing their first chant, " t h e ouside chant for the purified
S o m a . " N u m e r o u s rites are performed, many simultaneously,
overlapping and/or temporarily interrupted by others. Fires are
installed on the hearths in the hall of recitation. Eleven goats are
sacrificed.
The priests enter the hall of recitations where the Soma juice
is drunk. The H o t â recites his first sastra recitation from the
Rigveda. Four " S o m a s e q u e n c e s " follow, each consisting of a
Sämaveda stotra chant, a Rigveda sastra recitation, Soma
offering and Soma drinking (Chapter 9B). From now on, all
recitations and chants, each marked by formal features of
considerable complexity, take place in the hall of recitation.
Soma offerings are made into the offering fire and the remnants
are drunk in the hall of recitation.
At the midday pressing, the Grävastut priest recites, blind-
folded, Rigveda verses in praise of the Soma pressing stones
(page 374). Five Soma sequences follow. Then the priests are
offered daksinâ, a ritual fee, and the Yajamâna is anointed as in
the royal consecration. After the third pressing, another pair of
Soma sequences is gone through, then three more, then a single
especially mysterious one ("the sixteenth"), three nocturnal
rounds of four, and a finale in the early morning of the twelfth
day, bringing the total number up to twenty-nine. On this last
day, two copious Soma offerings are made for Indra. Ancestral
rites are performed as well as expiation rites for errors or
omissions that were made or might have been made. The alliance
between the Yajamâna and his priests is dissolved (sakhyavisar-
 The Agnicayana Ritual 77

jana). The Yajamâna, his wife, and the priests have the final bath
{avabhrtha), a concluding isti {udayanïyestï) is performed and a
final goat is sacrificed. This is followed by a departure rite
(udavasäna). The Yajamâna returns home in ceremonial proces-
sion with his wife and the three fires which he installs on his
home altars so as to perform the evening and morning Agnihotra
rites for the rest of his life as a householder.
 9

Basic Rites

Among the numerous ritual sequences that occur repeatedly

in the Agnicayana and in other Vedic Srauta rituals are the Isti
and the Soma sequence. In the preceding section, we have
briefly indicated where, in what surroundings and under what
circumstances some of these occur. The Isti is important not
only as a structural element but also because its basic manifes-
tation occurs in the Full- and N e w - M o o n Ceremonies. The
number and nature of Soma sequences defines the Soma rituals.
Because both of these are important structural elements we shall
describe the Isti and Soma sequence in greater detail. They may
be called the elements (crroi^eta) of Vedic ritual (page 33).

9A. The Isti

The isti culminates in an offering of rice or barley cakes. The

Yajamäna and his wife are assisted by four priests: the Adhvaryu
of the Yajurveda who performs the ritual acts; the Hotâ who
recites from the Rigveda; the Brahman who supervises the rites
without participating; and the Agnïdhra or " k i n d l e r . " An isti is
preceded, accompanied and followed by numerous accessory
rites, but its basic structure consists in a brief series of acts that
follow each other in rapid succession. The subdivision of these
acts into elements is fixed, but their numbering and grouping
together is to some extent arbitrary. In the following description
of the isti paradigm, six elements are grouped in three episodes
of two elements each, which is a natural manner of subdivision:

EPISODE I.
Element 1. The Adhvaryu commands the Hotä to address the deity,
e.g., Agni, by saying: "Address Agni!"
80 Ritual

Element 2. The Hotä addresses or invites Agni by reciting verses from

the Rigveda.
EPISODE II.
Element 3. The Adhvaryu exclaims to the Agnîdh: "Make (him) hear!"
Element 4. The Agnîdh shouts: "Be it so! May he hear!"
EPISODE III.
Element 5. The Adhvaryu commands the Hotä to recite his main
recitation, the yäjyä ("offering verse") by saying: "Say the
yäjyä for Agni!"
Element 6. The Hotä begins the yäjyä by murmuring: "Earth! Air! We
who say the yäjyä . . . " and then recites verses from the
Rigveda, ending with the exclamation: "May (Agni) lead
(the offerings to the gods)!" At the last syllable, which the
Hotä shouts at the top of his voice, the Adhvaryu makes
the offering by throwing or pouring it into the fire, and the
Yajamäna pronounces his "renunciation": "This is for
Agni, not for me!"

Some of the elements or parts of elements in this structure are

fixed (e.g., 3, 4, and the exclamation in 6). Others contain
variable units, or are themselves variable. In 1 and in the
renunciation in 6, the name Agni may be replaced by another,
but in the exclamation in 6 it is always Agni who is understood.
The verses from the Rigveda in 2 and 6 vary. As a result of these
and other variations, there are many different istis. All have the
same structure, but they have different names and address
different gods with different verses. The large rituals contain
many istis, which occur either in sequence, or inside other units
themselves inserted within the larger rituals.
The structural relations between the six elements in the three
episodes can be described in various terms and from various
points of view. From the point of view of the H o t ä , who recites
twice, the structure of the configuration of six elements is:
AR - B C - DR'.
From the point of view of the Adhvaryu, who issues a command
three times, it is:
RA-R'B-R"A'.
 Basic Rites 81

The istis constitute the paradigm for one type of ritual

performance. Other forms are obtained by extending or modify-
ing the basic form. In the animal sacrifice, for example, there is
an additional officiating priest, the Maitrâvaruna, and Episode I
of the basic structure is extended with one element, as follows:

Element 1. The Adhvaryu commands the Maitrâvaruna to command

the Hotâ to address the deity.
Element 1'. The Maitrâvaruna commands the Hotä to address the deity.

The rest is the same.

9B. The Soma Sequence

All the larger Soma rituals are characterized by sequences of

rites that I have called " S o m a s e q u e n c e s " (AGNI 1:49, 54, 599,
608, etc.). Each Soma ritual is defined by a specific number of
these sequences. The Agnistoma consists of twelve Soma se-
quences: five at the morning pressing, five at the midday
pressing, and two at the third (or evening) pressing. The Atirâtra
consists of 29 such sequences. The number is not symbolic, but
arrived at by computation: the Atirâtra is constructed from the
Agnistoma by first modifying some of its twelve Soma se-
quences; and then adding to them: three, one (the " s i x t e e n t h " ) ;
three nocturnal rounds of four; and one final sequence. Thus
12 + 3 + 1 + 12 + 1 = 29 (above, page 76).
Each Soma sequence consists of a chant (stotra) from the
Sämaveda; a recitation (sastra) from the Rigveda; Soma offer-
ings to the deities; and Soma drinking by the Yajamana and some
of his priests. Each of these four episodes consist of several
smaller rites and other elements, some of them fixed, and others
variable. Here follows a simplified description of the four
episodes and their constituent elements:

EPISODE I. CHANT (by three chanters)

Element 1. The Adhvaryu (or his assistant, the Pratiprasthâtâ, depend-
ing on certain circumstances: see AGNI 1:625) hands two
blades of darbha grass to the Udgätä, main chanter of the
82 Ritual

Sämaveda, while reciting: "You are the bed for the coupling
of Rk and Sàman — for the sake of procreation!"
Element 2. The Adhvaryu or Pratiprasthâtâ continues with a recitation
in which the sound HIM, certain gods and chanters occur,
and which ends with: "OM! Chant!"
Element 3. The three chanters, facing west, north, and south, intone the
chant which begins with HIM and consists of three se-
quences in each of which are triple repetitions of certain
lines, depending on which chant it is. This complex pattern
is marked with the help of sticks placed on a piece of cloth.
Each chant consists of an addition of five pieces, in some of
which the original syllables of the verse are "hidden" by
lengthened " o " 's. Numerous other rules are followed (see
AGNI 1:602, etc.).
Element 4. Yajamâna and Adhvaryu recite together a piece called
"Chant Milking" (stutadoha).
EPISODE II. RECITATION (by Hotâ, Maitrâvaruna, Brâhmanâc-
chamsin, or Acchâvâka)
Element 1. The Hotä (or one of the other priests) recites a piece called
"prior light" (puroruc). Without making any pause, in fact,
without taking breath, this leads into:
Element 2. The main recitation. This is marked by triple repetition of the
first and last verse, taking breath only at certain junctions,
etc. There are other insertions by the reciter (e.g., "Let us
both recite!") and by the Adhvaryu. The latter, called
"salutation'' {pratigara) are inserted when the reciter pro-
duces his lengthened " o " 's.—Caland and Henry (1906, I:
232, note 8), generally dauntless, referred to these as "bi-
zarres contorsions liturgiques."
Element 3. The Hotâ adds a piece called "Recital Strength"
(ukthavîrya).
Element 4. Yajamâna and Adhvaryu recite together a piece called
"Recitation Milking" (sastradoha).
Element 5. The Yajamâna adds another piece, which has no name but
which I shall translate for the sake of illustration:
The ritual has been, has been produced,
it is born, it has grown,
it has become king of the gods.
May it turn us into kings,
may we be masters of wealth!
EPISODE III. SOM A OFFERING.
This is in some respects like an isti. It involves an offering
 Basic Rites 83

an offering verse (yäjyä), and ends in the exclamation:

"May (Agni) lead (the offerings to the gods)!"
EPISODE IV. SOMA DRINKING
Element 1. Each priest who is about to drink addresses the Adhvaryu
with: "Adhvaryu, invite me"
Element 2. The Adhvaryu replies: "You are invited!"
Element 3. The priest drinks and recites from the Rigveda, touching his
face and heart at certain points.
Element 4. The Adhvaryu recites a long piece from the Yajurveda,
called "Long Drink" (dirghabhaksa).

This concludes the simplified description of a Soma sequence.

Again, some elements are fixed, others are variable, and there is
a certain amount of overall variation which it is not necessary to
describe in the present context. Some Soma sequences are more
complicated than the type outlined here, others are simpler.
Some deviate considerably, e.g., the sixteenth (Chapters 17 and
18: (5)).
 10

T h e Structure of Ritual

The description of the Agnicayana in Chapter 8 is consider-

ably abbreviated and simplified; it omits many rites and ceremo-
nies which are less important although we shall mention some of
them later. Of the more important ceremonies, only the Isti and
the Soma Sequence have been described in greater detail (in
Chapter 9). In A G N I , a fuller description of the entire Agnica-
yana or Atirätra takes up about 500 pages, and even there, -
various rites and rituals have been omitted because they had
already been described in other works. The summaries in
Chapters 8 and 9 should suffice, however, to gain a first
understanding of the structure of the ritual.
A cursory inspection of the description shows that some rites
occur at the beginning and again at the end of others. For
example, the construction of each layer of the altar is " s u r -
r o u n d e d " by performances of Pravargya and Upasad in the
morning and in the evening, as follows:
Pravargya - Upasad - Layer - Pravargya - Upasad. (1)
In the process of this " s u r r o u n d i n g , " Pravargya and Upasad
are regarded as a unit (the Nihnava and other lesser rites are
included but I shall disregard them). For if they were considered
as two independent rites, the process could be executed in two
steps, first surrounding the layering by one rite, and then
surrounding the result by another, as in:
(Pravargya - (Upasad - Layer - Upasad) - Pravargya) (2)
Both structures, exemplified by (1) and (2), are frequent, but the
second is more common. These same structures are found
among recitations. For example, at the beginning and end of
86 Ritual

each consecration of a layer of the Agnicayana altar, the

Adhvaryu recites Taittirïya Samhitâ 5.7.9.1a followed by
Taittirya Samhitä 5.7.8.1a. The structure is of the same form as
(1) and the two recitations, o b v i o u s l y , b e l o n g together in that
order. We know this without taking the content of the recitations
or rites into account; a mere inspection of the structure of (1)
suffices. Such an argument is, as we shall see, a syntactic
argument.
If we refer to the two recitations from the Taittirïya Samhitâ
together as B , and to all the intervening recitations (adding rites,
if we wish) as A, we would notice that the recitations called B
are the same on each layer, but the sequences of A are different,
at least in part. The general structure, however, is always the
same and may be expressed as:
BAB. (3)
This accounts for (1) if " B " is interpreted as " P r a v a r g y a -
U p a s a d . " It accounts for (2) if applied twice, interpreting " B "
first as " U p a s a d " and then as " P r a v a r g y a . " (3) is recursive and
exhibits the feature of self-embedding (page 53). Such structures
have long been recognized by students of Vedic ritual although
they have not been expressed in this form. Heesterman, for
example, describes the odana ritual of preparing a rice stew
which is subsequently eaten by the priests, and occurs repeat-
edly: " w e find the odana prescribed both at the beginning and at
the end of the horse sacrifice" (1983:88). The structure
" o d a n a — i n t e r v a l — o d a n a " is again of the form (3).
In all such cases we have an identical element at the beginning
and at the end of a rite. There are numerous cases where the
activities at the beginning and end of a rite are related to each
other but are not identical. Hubert and Mauss (1909) drew
attention to some facts that are well known to ritualists, viz.,
« that the final bath (avabhrtha) at the end corresponds to the
consecration (dïksâ) at the beginning, and the concluding of-
fering (udanïyesti) similarly corresponds to the introductory
offering (prâyanïyesti), the departure (udavasäna) to the en-
trance (adhyavasâna), the dissolution (sakhyavisarjana) to the
 The Structure of Ritual 87

alliance (tänünaptra), and so forth. In all these cases, a large

number of rituals intervene between these initial and final rites.
In lesser rites, we often find the same structure. For example,
within the Full- and New-Moon Ceremonies, the main oblation
(pradhänahoma) is preceded by oblations called äjyabhäga and
followed by an oblation called svistakrt. The äjyabhägas are in
turn preceded by "fore-offerings" (prayäja) and the svistakrt is
followed by "after-offerings" (anuyäja). In the Soma sequences,
each sastra recitation is preceded by puroruc and followed by
ukthavïrya (see Chapter 9B, Episode II: Element 2 occurs
between Elements 1 and 3). The puroruc is often preceded by a
recitation called ähäva ( " L e t us both r e c i t e ! " ; omitted in Chap-
ter 9B because it does not occur in all cases), and the ukthavïrya
is followed by sastradoha (Element 4). Similarly, there are
recitations preceding and following each stotra chant (Elements
2 and 4 surround 3 in Episode I). These examples can be
extended almost indefinitely.
A m o m e n t ' s reflection will show that numerous simple human
activities are of the same form. If a rite takes place within an
.enclosure-;'the priest performing the rite has first to enter the
enclosure, and after completing it, leave. Similarly, when going
to a concert, one first goes to the concert hall, which is followed
by entering the building, listening, leaving, and returning, in that
order. In ritual, such structures are repeated and extended
indefinitely.
As in this nonritual example, both the main oblation of an isti
and the sastra recitation function as a center for two pairs of
initial and final acts. The sequence in which the two initial rites
or recitations occur is reversed or mirrored in the two final rites
or recitations—as in (2), above page 85. This general structure is
found on many levels and can be applied to units of various
sizes. Let us represent five initial rites by A , , . ... , A 5 . They
occur in the following order at the beginning of the Agnicayana:

A, introductory procession
A 2 adhyavasäna
A 3 dïksâ
88 Ritual

A 4 prâyanïyesti
A 5 tänünaptra
N o w let us denote each corresponding final rite by the same
symbol, adding an asterisk, as follows:
Af return home
A* udavasâna
Af avabhrtha
A% udayanïyesti
Af sakhyavisarjana
The order in which these rites are gone through is the following:

This suggests a tendency, not quite successful, to establish the

regular " m i r r o r - i m a g e " pattern, viz.:

^ 1 •**• 2 3"*^AA-5 • • • **• 5**• A H> '2.'^\

All these nesting or self-embedding structures can be repre-
sented by recursive rules of the following form:
B-+ABA (4)
This generates structures AABAA, AAABAAA, AAAABAAAA,
. . . by applying the rule again and again to its own output.
The occurrence of such recursive rules, viz., rules that gen-
erate infinitely many structures by applying and reapplying finite
mechanisms (in our case, one single operation) is significant, for
it shows that the ritual can be extended indefinitely. T h e Indian
theorists were aware of this. The ritualists constructed rituals of
indefinitely increasing complexity, the sattra rituals. These ritu-
als are often purely theoretical, but this does not diminish their
significance as both actual and possible rituals exhibit ritual
structure. Hillebrandt did not take these theoretical rituals
seriously: "Diejenigen S a t t r a ' s , welche länger als zwölf Jahre
dauern, heissen mahäsattra's . . . und hier versteigt sich Mythus
und Phantasie der Yäjnika's zu den sechsunddreisigjährigen
Opfern der S ä k t y a ' s , den hundertjährigen der S ä d h y a ' s , den
tausendjährigen der V i s v a s r j " (Hillebrandt 1897:158). The In-
 The Structure of Ritual 89

dian grammarian Patanjali, on the other hand, took these rituals

quite seriously, because he detected in them the same recursive-
ness that governs the structure of language. When discussing the
infinity of language, which grammar must describe by finite
means (cf. page 40), Patanjali refers to these sattra rituals:
' T h e r e are indeed linguistic expressions which are never
used. . . . Even though they are not used, they have of necessity
to be laid down by rules, just like protracted sat tras" {s anti vai
sabdä aprayuktàh . . . yady apy aprayuktä avasyam dîrghasat-
travallaksanenänuvidheyäh; Mahäbhäsya, Kielhorn, ed.. I:
8,23; 9,15).
 11

E m b e d d i n g and Transformation

The indefinite complexity of Vedic ritual is not due solely to

the recursive rule (4) we have just reviewed, but primarily to two
others. The first of these may be called inserting or embedding.
Structure (4) illustrates the special case of self-embedding, as we
have seen. The general case of embedding is most easily
illustrated when different rituals are compared with each other.
We have seen, for example, that the number of Soma sequences
in the Atirätra-Agnicayana is twenty-nine, a number that is
reached by adding sequences to the original two of the third
pressing (Chapter 8, tenth to twelfth day).
I shall discuss three further examples of such embeddings: the
first are the kindling verses (sâmidhenî), verses from the Rigveda
recited by the Hotâ when the firewood sticks are placed upon the
altar; the second, rituals that are embedded in the Agnistoma
and help transform it into the Agnicayana; and the third, the
construction of the twenty-ninth or final sastra from the H o t ä ' s
morning recital on the tenth day.

11 A. The Kindling Verses

During the Full- and New-Moon Ceremonies, sticks of fire-

wood (samidh) are placed upon the altar by the Adhvaryu. This
act is accompanied by the recitation of kindling verses
(sämidhenf) by the Hotâ. There are eleven such verses, all from
the Rigveda. During the animal sacrifice, two verses are added.
During the animal sacrifice for Väyu on the first day of the
Agnicayana (above, Chapter 8), the â r a u t a Sütra of Baudhäyana
prescribed 24 verses which are obtained by adding another
eleven, some from the Rigveda and some from the Taittiriya
92 Ritual

Samhitä of the Yajurveda. During the 1975 performance of the

Agnicayana, however, only the 13 verses of the ordinary animal
sacrifice were recited. This is in accordance with Sänkhäyana
Érauta Sütra (9.23) but constituted one of the first differences
between the 1975 performance and Baudhäyana Srauta Sütra.
T h e s e differences in number illustrate the structure of inser-
tion o r embedding (from A G N I 1:311):

THE SAMIDHENÏ VERSES

Full- and new-moon Animal sacrifice and Vâyavyam Paéu

ceremonies 1975 Vâyavyam PaSu according to
Baudhäyana
RV 3.27.1 RV 3.27.1 RV 3.27.1
RV 6.16.10-12 RV 6.16.10-12 RV 6.16.10-12
• RV 3.27.13-15 RV, 3.27.13-15 RV 3.27.13-15
RV 1.21.1 RV 1.21.1 RV 1.21.1
RV 3.27.4 RV 3.27.4-6 RV 3.27.4
TS 4.1.7.1a-4i
RV 3.27.5-6
TS 2.6.11.1a
RV 5.28.5-6 RV 5.28.5-6 RV 5.28.5-6
TS 4.1.7.4 k

TOTAL 11 13 24
TABLE 1

11B. From Agnistoma to Agnicayana

Among the rituals that are embedded in the Agnistoma and

help transform it into the Agnicayana, the most important ones
are: measurement of the Agniksetra, the "field of A g n i " on
which the bird-shaped altar will be constructed; construction of
the new domestic altar; and setting up of the Agniksetra.
In general, the first two of these rituals take place on the third
and fourth day, respectively, prior to the introductory offering or
 Embedding and Transformation 93

prâyanïyesti. The third ritual is performed on the fourth day after

the Pravargya, Upasad, Nihnava and Subrahmanyä recitation.
In 1975, the menses of the wife of the Yajamäna began on
April 13, scheduled on the second day of the ritual. The rites of
the third and fourth ritual day, planned for April 14-15, could not
be executed because of the ensuing pollution; expiatory rites
were performed instead. Some of the ceremonies of the third and
fourth ritual days were combined and performed on April 16. As
a result, ceremonies that, in the Agnistoma, are performed
consecutively, but that are not consecutive in the Agnicayana,
were also performed consecutively on April 16, 1975, but this
time as a consequence of the menses of the wife of the
Yajamäna. The resulting structures are illustrated in Table 2
(from A G N I 1:386):

SEQUENCE OF RITES ON THE THIRD AND FOURTH DAYS

I AGNISTOMA II AGNICAYANA III April 16, 1975

Measurement of Ma-
hâvedi(3)
Measurement ofAgni-
ksetra(3)
Construction of New Construction of New
Domestic Altar (4) Domestic Altar
Introductory Offering Introductory Offering Introductory Offering
(2) (4)
Purchase of Soma (2) Purchase of Soma (4) Purchase of Soma
Guest Offering (2) Guest Offering (4) Guest Offering
Pravargya, Upasad, and Pravargya, Upasad, and Pravargya, Upasad,
Subrahmanyâ(2) Subrahmanyâ(4) and Subrahmanyä
Measurement of Mahâvedi Measurement of Ma-
(3) hâvedi
Measurement of Agni-
ksetra
Preparation of Setting up Agniksetra Setting up Agniksetra
Uttaravedi (3) (4)
TABLE 2
94 Ritual

In Table 2, rites specific to the Agnicayana have been itali-

cized and the ritual days are put in parentheses.
These structures exemplify embedding and illustrate in pass-
ing that each actual performance of a ritual may be different—an
obvious fact that does not affect but exhibits the underlying
ritual structure. This confounded Richard Schechner, a Profes-
sor of Performance Studies, who argued that the 1975 perfor-
mance of the Agnicayana was significantly different from all the
others that had preceded it (see Schechner 1986, 1987; cf. Staal
1987) instead of recognizing that each performance is a different
event and yet exhibits the same ritual competence (Chapter 21).

11C. The Final Sastra for the Asvin Twins

The twenty-ninth or final sastra recitation of the Agnicayana,

dedicated to the Asvin twins, is recited by the Hotä before
sunrise on the final day and consists of a thousand mantras. It is
a tour de force not only because of the length of the recital but
also because of its complexity. Its demands on the powers of
memory are extraordinary and we shall return to this topic
(Chapter 27; A G N I 1:683-686; Staal 1986b). It is referred to here
in order to illustrate that there are cases where embedding and
inserting are supplemented by omitting. In general, omitting is
merely the reverse of inserting: just as the Agnicayana can be
thought of as arising from the Agnistoma by inserting, the
Agnistoma can be thought of as arising from the Agnicayana by
omitting. However, in the case of the twenty-ninth sastra only
the combined operation of inserting and omitting can explain the
proceedings.

11D. Conclusions

All the insertions we have discussed operate at different

levels, and apply to large as well as small units. Insertions are
made into other insertions, and here their recursive character
becomes apparent. If we take a closer look at these insertions
within insertions, we shall find another type of recursive rule
 Embedding and Transformation 95

that contributes to the indefinite complexity of the Vedic ritual:

transformational rules.
Let us start with the Animal Sacrifice for Väyu. Call it B. It is
inserted in the Agnicayana after the Ritual Preparation of the
Ukhä Pots (A) and before the Election of the Priests (C). The
ritual rule that effects this insertion may therefore be written as:
AC-^ ABC (5)
N o w let us consider the internal structure of the Animal Sacrifice
(B). Confining our attention to the Sâmidhenï verses, which I
shall call £ , it consists of various rites preceding these verses,
which may be lumped together and referred to as D , and various
rites following them, together referred to as F. Thus the Animal
Sacrifice B may be represented by DEF,

B-+DEF (6)
In an Animal Sacrifice, there are thirteen Sâmidhenï verses, as
we have seen; the E in (6), therefore, consists of thirteen such
verses. We also know, however, that the Animal Sacrifice for
Väyu, which occurs in the Agnicayana, should contain (ac-
cording to Baudhäyana) twenty-four Sâmidihenï verses. This
group, which may be called G, arises from E by inserting another
eleven mantras, i.e., by an insertion similar to the insertion
represented by (5). It would not be appropriate, however, to
express this insertion by a rule of the form:
E-+G (7)
This would indicate that in all Animal Sacrifices there are
twenty-four Sâmidihenï verses. We have to express that E is
replaced by G only in the Animal Sacrifice that is embedded in
the Agnicayana. In other words, we must restrict the context, or
the configuration in which E occurs and which conditions its
replacement by G. The simplest way to do this is by a rule of the
form:
96 Ritual

ABC

(8)

D E F DGF

This rule is formulated with the help of a double arrow to

distinguish it from the rules with single arrows such as (5)-(7).
Rules of the form (8) are identical with the transformations of
linguistics (Chapter 6A).
Transformational structures are typical of Vedic ritual, and it
is easy to provide other examples. Let A denote ceremonies
preceding the consecration, B the consecration, and C ceremo-
nies following the consecration. The prototype of B in the
Agnistoma consists of a sequence of rites, beginning with the
consecration isti (dîksanïyesti), that will be referred to as D. This
is followed by the ceremonies with the antelope skin, the
mekhalä rope, the turban, and so forth (F). In the Agnicayana,
between D and F new rites are inserted relating to the ukhä pot,
in which fire originates (G). In other words, a rule of the form (9)
applies:

ABC ABC

(9)

D F DGF

Actually, D itself is also transformed. It is replaced by a new rite

called agnidîksanïyesti, o r D * . A transformation must apply that
is of the form:

A B C A B C

(10)

D . .
 Embedding and Transformation 97

Another example is the Carrying Forth of Agni (agniprana-

yana; B). It follows oblations to Visvakarman on the domestic
altar (A) and is followed by Adhvara oblations on the offering
altar (O-JThe prototype of the Carrying Forth in the Agnistoma
consists of the transportation of the fire by the Adhvaryu,
recitations by him and by the Hotâ, and chants by the Prastotä
(£>). Afterwards the fire is installed on the altar (F). In the
Agnicayana, another recital is inserted, viz., the recitation by
the Second Hotâ or Maiträvaruna of the Song to the Irresistible
Warrior. If this is referred to as G, the expression (8) expresses
the structure again adequately, but it should be understood that
DGF does not represent a simple sequence of three rites, but a
combination in which some of the rites overlap or are simulta-
neous.
In the Final Bath (avabhrtha) there is also the insertion of a
new rite, characterized by the recitation of Taittirïya Samhitä
4.6.2.6 r. This can be described by a structure of the form (9).
In these transformations the context is specified on the left and
on the right; in other words, both the following and the preceding
ceremonies are specified. Sometimes it is more natural to leave
one side unspecified, or to regard it as empty. We might refer to
this as the prefixing, suffixing, or mere " a d d i n g " of rites. For
example, in the Full and N e w Moon Ceremonies the Formulas of
Completion (samistayajus; A) consist of two mantras, Taittirïya
Samhitä 1.1.13.3 u-v. In the Animal Sacrifice, A follows the final
oblations (C) and precedes the burying of the heart-spit (D). But
in this sacrifice the two formulas A are replaced by three (A*),
or:
C AD-* C A* D (11)
In the Agnistoma the context is specified differently, as C* . . .
D*, and another nine formulas {B) are added:

C* A* / ) * 4> C* A* / ) * (12)

A* B
98 Ritual

In the Agnicayana in a context C** . . . D**, another nineteen

mantras (E) are added:

** £)** (13)

Such sequences of mantras can be extended indefinitely.

What is the significance of these structures? They show that
an apparently empiricist and purely behaviorist description of
the sequence of acts of the Agnicayana, A, B, C, D, . . . , as if
they had a linear structure (14), is inadequate:

A B C D E F G . (14)

Underlying the sequence A, ß , C, . . . there is in fact a

hierarchical structure arrived at through the reiterated operation
of embeddings and transformations, viz., something of the form:

K L V O . .. (15)

D E F M N

In the description of the performance in A G N I I we have

implicitly accepted that it has a structure of the form (15). This
idea underlies the subdivision into episodes and smaller units,
described in varying detail, and the references to prototypes of
the isti, the Agnistoma, and other components. For example, we
 Embedding and Transformation 99

have not described rites such as B as merely "following" A,

which corresponds to the structure expressed by (14). Rather,
we have described B as the first rite of a ritual P, which
corresponds to the structure expressed by (15). T h u s , the
agnidîksanïyesti is not described as merely following the rites in
the previous chapter of A G N I ("Episode 3 " : pages 313-317),
but as constituting the first rite of the next, the consecration
( " E p i s o d e 4 " : pages 317-333).
 12

Ritual and G r a m m a r

Indian ritualists have always stressed the hierarchical struc-

ture of the ritual, as have Willem Cal and and other Western
scholars. The Srauta Sütras describe the main rituals in a
particular order. I have mentioned several of these, but shall
now single out four and refer to them by capital letters:

D: Darsapürnamäsa
P: Pasubandha
A: Agnistoma
C: Agnicayana.

This sequence is hierarchical. There is increasing complexity. A

person is in general only eligible to perform a later ritual in the
sequence, if he has already performed the earlier ones. Each
later ritual presupposes the former and incorporates one or more
occurrences of one or more of the former rituals. Sometimes
these embedded rituals are abbreviated. In general, they un-
dergo modification. I shall single out three embeddings:

In P, performances of D are embedded when a cake of eight potsherds

is offered to Agni and when a cake of eleven potsherds is offered to
Agni-Visnu;
In A, two performances of P (agnïsomïyapasu and savanïyapasu) and
several performances of D (e.g., dïksanïyesti, prâyanïyesîi, avabhrthesti,
udayanïyesti, udavasânïyesti, not to mention performances of D embed-
ded in P) are embedded;
In C, a performance of A, fourteen performances of P and numerous
performances of D, some of which already embedded in A and P, are
embedded.
102 Ritual

Chapter 11 has provided an idea of the syntax of these

structures. I shall now describe that structure in more general
and abstract terms, abstracting from the specific structures we
have met before and constructing a model of ritual—a formal
representation corresponding to what Hubert and Mauss called
a " s c h e m e abstrait du sacrifice." In order to make this precise,
a series of artificial assumptions will be made, redefining Z), P
and A. The reason for these artificial assumptions and definitions
is that they constitute a model which exhibits abstract ritual
structures. The model is identical with respect to these struc-
tures to the actually existing rituals, but is less elaborate than
the latter. It may look more complex because it is formal; but it
can only be represented in this formalized form because it is, in
fact, simpler. Actually existing rituals can be analysed in the
same manner as the model with regard to the structures in which
I am here interested, but such an analysis would, again, be much
more complex.
Such a procedure is common in science but as yet, if we
except linguistics, rare in the human sciences. It is also easily
misunderstood. Hans Penner (1985), for example, criticized my
formalizations of syntactic analysis when they were first pub-
lished (1979a) by emphasizing that, by my own admission, they
" d o not correspond to any existing ritual" or " t o any actual
ritual." I responded (1986a:41): " R a t h e r , they correspond, if
they are adequate, to features that are abstracted from actually
existing rites, just as the laws of physics do not correspond to
specific events that take place in my garden, but to features
abstracted from such e v e n t s . " At the time of writing I had not
read Quine's much more succinct answer to similar criticism:
" T o complain of bare bones is like criticizing the physicist for
failing to capture the richness of the rain forest" (Quine 1981:
186).
But enough of mere words. Let us assume that there is an
Agnistoma A in which three occurrences of D and two occur-
rences of P are embedded. In each P, furthermore, two occur-
rences of D are embedded. Let us fix the order by assuming that
in the performance of A, there is first a performance of D , , then
 Ritual and Grammar 103

of Px, then of D 4 , then of P2 and finally of D7. In Px, let there be

occurrences of D2 and D3 and in P2, of Z)5 and D 6 , in that order.
Next we assume that each ritual consists of a sequence of rites.
We shall adopt the convention, that each ritual marked by a
capital letter (such as A, Z), P) will consist of rites denoted by
small letters, indexed with single numerals if the governing ritual
is not indexed with a numeral (e.g., al9 a2, . . . ;rf,,rf 2 , . . .), and
indexed with double numerals if the governing ritual is already
indexed with a numeral (e.g., ritual D3 will consist of rites rf31,
d32, . . . ; P2 of P 2 1 , ^22' • • •)• Now we can give a precise
description of A if we know the exact number and order of these
constituent rites. Let us assume that ritual D always consists of
three rites. Let us assume that P always consists of nine rites.
Since these include the rites of D , a simple assumption would
have the rites of F which are not rites of D at the beginning, in
the middle and at the end; and the rites of the two occurrences
of D in between the beginning and the middle, and the middle
and the end. For A, we assume a sequence of thirty-three rites
including the rites of P. A simple assumption would again be
that, within A, the rituals P and D are always separated by a
single rite a.
A linear description of Agnistoma model A will now look like
this:
ax dxx dl2 dX3 a2 pxx d2] d22 d23 pX2 d3X d32 d33 pX3 a3 d4X d42
d43 a4 p2X d5l d52 d53 p22 d6l d62 d63 p23 a5 dlx d12 d13 a6. (16)
It should be apparent that not only the artificial model A can be
represented by such a sequence of rites, but that the real
Agnistoma can also be represented by such a sequence of rites —
only a longer sequence. In addition we would need more letters,
more indices, and longer indices. What are the advantages of
such a representation? The answer is: none. A linear represen-
tation of this type is not only extremely cumbersome, but it
obscures all the elements of structure we have been so eager to
detect.
At this stage it should be noted that most of the modern
descriptions of Yedic ritual are basically linear representations
104 Ritual

of precisely this type. This verdict is exaggerated, because

Caland & Henry refer back to Schwab and Hillebrandt, Schwab
refers back to Hiilebrandt, etc. The verdict is even less applica-
ble to the Srautakosa, which adopts a type of description which
is closer to that of the Srauta Sütras. But to the extent that all
these works deviate from the Srauta Sütras, their deviation is in
the direction of linear description. Precisely for this reason these
works obscure the structures it has been our aim to make
explicit.
H o w then do we arrive at a representation which reveals these
structures? This can only be done by adopting a non-linear
method of description. We can arrive at such a description by
gradually transforming the representation (16). I shall first re-
place (16) by a series of expressions in which each ritual is
rewritten (which will be symbolized by a single arrow) as a
sequence of rites and rituals, as follows:

A —> ax Dx a2 Px a3 D4 a4 P2 a5 D7 a6

P2-> p2X D5 p22 D6 p23

Dx - * dn dX2 dx3
D2 -> d2X d22 d23
D3 - » d3X d32 d33
D4-+ d4X d42 d43
D5 —> d5x d52 d53, etc.

In these expressions there is still a great deal of repetition and

redundance. In order to eliminate this, we simply delete all
indices of rituals, and omit the first numeral of indices with
double numerals. The result is:

A -> ax D a2 P a3 D a4 P a5 D a6 (17)
— P\ D P2 *-* P3 \lo;
D->dxd2d3 (19)

At this point we have arrived at a representation of model A

which clearly exhibits the underlying structure and is as perspic-
 Ritual and Grammar 105

uous as the material allows. The reader can verify that the
assumptions we made when constructing A are expressed by

An equivalent representation of this same information can be

given with the help of trees:

(17')

D a2 P a3 D a4 P a5 D

(18')

(19')

d\ d2 d3

The embedding of P and D in A, and of D in P , is represented in

(20): see next page.
The structure of (20) can also be expressed by making use of
parentheses:
A - * a{ (dl d2 d3)a2 {px (d{ d2 d3)p2 {dx d2 d3 )p3)a3 {dx d2 d3)
a4 {px (dx d2 d3)p2 (dx d2 d3)p3)a5 (dx d2 d3)a6. (20')
At this stage it may be noted that representations of the form
(17)—(19) are basically equivalent to the organization of the
material adopted by the authors of the Srauta Sütras. A Srauta
106

a, D a2 P a3 D a4 P a5 D a6

K
di d2 d3 (20)

dx d2 d3

Sütra would convey the information as follows. First it would

have a section in which D is described as dx d2 d3—correspond-
ing to (19). The next section, corresponding to (18), would
provide the description of P, referring to D whenever it occurs,
but without repeating the information given in the first section.
The next section, corresponding to (17), would provide the
description of A, referring to P and D whenever they occur, but
without repeating the information given in the earlier sections.
Only the modifications which the rituals undergo when they are
embedded, are always specified.
H o w can we handle these modifications? Let us try to
represent the one example which I gave before. We introduce
this into our model by assuming that in ritual D , the first rite, d]9
represents the recitation of fifteen samidhenï verses. Let us
further assume that in the second occurrence of D in P9 rite dx
has to be replaced by a rite d*, in which seventeen samidhenï
verses are recited. We cannot simply represent this transforma-
tion by adding an expression:
d\ - * d* (corresponding to (7) in Chapter 11D, page 95),
 Ritual and Grammar 107

for the effect of this would be that all occurrences of dx are

replaced by occurrences of d*. What we must do is, replace by
df only the dx in the second occurrence of D in P. This can be
done by introducing a transformational rule using a different
symbol instead of the single arrow—», for example a double
arrow =>. We have to represent the entire configuration in which
dl occurs since it is not otherwise possible to single out the d{ we
wish to single out. This can be done as follows:

(21)

Another example of this kind of transformation is the following.

When I wrote that A and fourteen occurrences of P are embed-
ded in C, I was simplifying. Actually, A and two occurrences of
P are embedded in C. Two occurrences of F are embedded in A.
But in the occurrence of A which is embedded in C, the second
occurrence of P is modified: it involves the sacrifice of eleven
goats instead of one.
To represent this, we start with the following two expressions
(in which irrelevant rituals and rites are represented by dots):

C -> . . . P . . . A . . . P . . .
A ^ . >. P . . . P . . .

H e r e P is the animal sacrifice of one goat. N o w let us write P *

for the ritual which involves eleven goats. Then, abstracting
from all the other modifications which A undergoes when it is
embedded in C, the required transformation can be represented
as follows:
108 Ritual

(22)

. . . P . . . P f. .

These examples may suffice to represent two features of the

structure of the descriptions of the Srauta Sütras, and indeed of
the srauta rituals themselves: embedding and modification.
What is astonishing about this structure is that it is exceedingly
similar to the syntax of a natural language as described in
Chomsky 1965. As linguists will long have recognized, expres-
sions of the form (17)—(19) are phrase structure rules. (The rites
represented by small letters, which I have introduced tenta-
tively, correspond to some extent to lexical items). Expressions
of the form (22) are transformations. The phrase structure rules
are recursive, because the same symbols occur to the right and
to the left of the single arrows: P, which occurs on the right in
(17), occurs on the left in (18); and D , which occurs on the right
in (17) and (18), occurs on the left in (19).
The ritual modifications have another property, which is
common in Vedic ritual as Professor Heesterman reminded me
(in a letter of January 9, 1976): an embedded ritual may be
interrupted, once or several times, by the ritual in which it is
embedded, to be continued or completed afterwards. For exam-
ple, the embedding of P in A assumes the following form: some
rites of P are performed, followed by rites belonging to A, after
which subsequent rites of P continue, etc. (in a linear description
the sequence has to be interrupted: see, e.g., Caland and Henry
1906:125-8, 186, 188).
For some time it seemed to me that this alternation was a
 Ritual and Grammar 109

characteristic of ritual without a linguistic parallel. Several

linguists I contacted could not provide a similar structure in a
natural language. But when I phoned Professor J, R. R o s s , he
immediately produced a simple example from English. Let us
embed the sentence:
(51) Harry is sad
into another sentence, as follows:
(52) It seems to me that [Harry is sad].
51 51
Applying several transformations we can derive the following
sentence, in which constituents of the original sentences alter-
nate:

Harry seems sad to me.

51 52 51 52

The recursive features of the ritual allow the construction of

rituals of indefinitely increasing complexity. Such recursiveness
is exploited in some of the, partly theoretical, sattra rituals. We
have seen that the grammarian Patanjali compared the recursive-
ness of language to these ritual creations which also have to be
described by rules (page 40).
Before discussing possible explanations for the similarities
between structures in ritual and language, it should be empha-
sized that the occurrence of such structures does not imply that
the ritualists were any more aware of their precise form than
language users are conscious of the syntactic rules which they
employ. In linguistics, this has been often misunderstood and
Chomsky has repeatedly emphasized that " t h e structure of
particular languages may very well be largely determined by
factors over which the individual has no conscious c o n t r o l "
(Chomsky 1965:59). The same applies in ritual. It is not invali-
dated by the fact, that we have based our analysis largely upon
descriptions of ritual provided by the ritualists themselves. The
structures I have discussed should therefore be distinguished
110 Ritual

from such structures as Lévi-Strauss found in a native Yoruba

theory, about which he says: " B u t , as theories go, the Yoruba
seem to have been able to throw more light than ethnologists on
the spirit of institutions and rules which in their society, as in
many others, are of an intellectual and deliberate c h a r a c t e r "
(Lévi-Strauss 1966:133). In Vedic ritual, the descriptions by the
authors of the Srauta Sütras, too, have thrown more light on the
subject than any modern author could do without their aid.
H o w e v e r , this does not imply that the deeper structures which
underlie many ritual constructions have at any time been con-
ceived deliberately.
Despite the specific difference in recursiveness we have
found, the occurrence in the syntax of both ritual and language
of specific and unobvious rules, viz., phrase structure rules,
transformational rules and self embedding rules, is sufficiently
striking to demand an explanation. It may be premature to offer
such an explanation at this early stage. One reason is that the
data upon which these similarities are based are asymmetrical.
On the language side, the relevant structures have been shown to
be in all likelihood universal. On the ritual side, our data have
been confined to Vedic ritual. Though this is most probably the
best point of departure, we are only beginning to explore a
domain of ritual syntax which appears to be as complex as the
syntax of language. Even in the domain of Vedic ritual we have
taken into account only a few structures. For example, we have
not so far used the fact, that the ritual is executed by combining
the contributions of different groups of priests with distinct ritual
systems. This is comparable to a language which exhibits the
rules of several distinct grammars as do some Creole languages.
Also, we have not studied the metarules which are required
when different rules, apparently inconsistent with each other,
have to be combined (Chapter 26). Similarly, the structure of
sampad 'numerical congruence' (Heesterman 1957:35,53) awaits
syntactic treatment. Since there is so much left u n d o n e , I can
only offer speculation.
The similarities we found may require elaborate explanations.
But they might alternatively be explained by one of three simple
 Ritual and Grammar 111

explanations: either language is the cause, or ritual is the cause,

or there is a common cause for both. I shall discount the third
which could be almost anything about which we know little—
ranging from the 'essence of man' to the structure of his brain.
It would seem plausible to opt for the first explanation. In its
support it could be argued that it would not be surprising for
man, w h o has a specific structure of language anchored in his
brain, to exhibit this same structure in his rituals. I think this is
possibly correct, but I wish to point out that it is not the only
plausible view. First of all, it smacks of prejudice. It is we w h o
are obsessed by language and who have (despite survivals) lost
touch with ritual. H e n c e it is only natural that the view that
language is primary, would appeal to u s . But for Early Man,
ritual was at least as important as language is for u s . Ritual, after
all, is much older than language. Unlike language, it can origi-
nate on all fours. It is c o m m o n among animals.
If we adopt an evolutionary perspective, it becomes relevant
to consider animal systems of communication or 'animal lan-
guage.' Evidence for an animal syntax which is at all like the
human appears to be meagre or nonexistent. As for monkeys and
apes, their communication systems " h a v e little relationship with
human language, but much with the ways human beings express
emotion through gesture, facial expressions, and tone of v o i c e "
(Lancaster 1968:446). In a study which pays some attention to
syntactic problems, Reynolds, partly following Altmann, has
shown that in the communication among M a c a q u e s , " a se-
quence of displays will determine the probability of the next
d i s p l a y " (I quote from Hill 1972:314). But Hill has rightly
pointed out that this would precisely be a communication system
with the properties of a Markov process model, " w h i c h
Chomsky (1957) so convincingly demonstrated was not the sort
of system represented in human language." Hill has in fact
suggested that a basic distinction between animal communica-
tion and human language is that the former lacks recursiveness
(Hill 1972:313-315; cf. Revzin 1974:25).
T o evaluate the problem we also require detailed studies of the
syntax of animal ritual. Little seems to be available in the
112 Ritual

otherwise impressive work of scholars like Crane, Huxley,

L o r e n z and Tinbergen (see, e.g., Huxley 1966). There is more,
but still not enough, in Thorpe 1951 and Barlow 1977. Such
information, however, may clench the argument. If there is no
recursiveness in the syntax of animal ritual, the issue remains
inconclusive. But if there is, and if Hill is right, this suggests that
the recursiveness which is the main characteristic of the syntax
of human language has a ritual origin.
Another hypothesis is consistent with the view that syntax has
a ritual origin: the hypothesis that syntax is older than seman-
tics. This is precisely what I suspect to be the case. The contrary
view, that structured systems of meaning developed first, seems
to be accepted implicitly by many scholars. It is certainly more
logical (hence the original appeal of generative semantics, which
denied an independent level of syntax). In addition to the
emphasis on synchronistic structure, this view would help to
explain the readily accepted taboo which the Société de linguisti-
que of Paris imposed upon investigations into the origin of
language: for meaning was held to be mysterious and inaccessi-
ble to scientific treatment. The question that seems to lurk
behind such a view is: why should people wish to establish
language if not for the sake of communication? But such an idea
rests on flimsy grounds. Language was not deliberately and
consciously established, its emergence and growth need not be
due to pragmatic or functional needs, and it clearly exhibits a
great deal of structure which plays no role in communication.
N a t u r e , in brief, is not always logical.
There are many facts that support the view that syntax is older
than semantics. Vedic ritual itself provides such evidence. Vedic
ritual is replete with recitation and chant. These recitations and
chants are comprised of a highly structured mixture of natural
language and meaningless sound. But whether or not portions of
this mixture are meaningful in other contexts, in their ritual use
the only things which matter are the sounds and their configu-
rations. According to the ancient ritualist Kautsa, all mantras
are meaningless, and this view is indeed applicable to the
occurrence of mantras throughout Indian culture (page 234). In
 Ritual and Grammar 113

Vedic ritual, as in mantra meditation, the function of language is

phonetic and syntactic, not semantic. We shall return to this
curious fact in Part III.
Other facts support our hypothesis in a negative way. During
performances of Vedic rituals, the participants are not supposed
to communicate with each other through ordinary language. In
later times, communication is restricted (and effectively elimi-
nated) by prohibiting the use of any language other than San-
skrit. Linguistic communication of an ordinary sort is also
excluded when ritual recitations are prescribed as anirukta, 'not
enunciated', upâmsu, 'inaudible', or when the rites are executed
iusnïm, 'in silence' (Renou and Silburn 1954). In the ritual of the
chanters, aniruktagäna 'unenunciated chant' is a protracted
chant of o with the same structure and pitch variation as the
original mantra, which is represented mentally. In meditation,
linguistic meaning is also excluded. The meditation mantras,
which are recited elsewhere, increasingly come to consist of
meaningless syllables, which tend to become mental or vanish
altogether. Thus phonetic structure disappears and syntactic
structure is reduced to mere repetition, expressible by the single
syntactic rule A —> AA.
I am inclined to believe that what we witness here is not a
curious collection of exotic facts, but a remnant or resurgence of
a pre-linguistic stage of development, during which man or his
ancestors used sound in a purely syntactic or ritual manner. This
would be supported by the generally archaic features of ritual as
well as of mysticism, and the claim that there was a golden age
when ritual practices and mystical insights were common (cf.
Staal 1975a:58). The locus classicus is Rigveda 1.164.50 =
10.90.16:

yajnéna yajnarn ayajanta deväs

täni dharmäni prathamäny âsan
"With the sacrifice the gods performed the sacrifice.
These were the first ordinances"
(transi. Brown 1965:32; 1968:218).
114 Ritual

In his presidential address to the 1974 meeting of the Linguis-

tic Society of America, Morris Halle expressed a similar view—
the conception of language not as a means for communication
but as playful activity, a kind of game. H e quotes in this
connexion from Novalis' Sprachwissenschaftlicher Monolog of
1798:

Das rechte Gespräch ist ein blosses Wortspiel . . •. Wenn man den
Leuten nur begreiflich machen könnte, dass es mit der Sprache wie mit
den mathematischen Formeln sei.—Sie machen eine Welt für sich aus—
sie spielen nur mit sich selbst, drücken nichts als ihre wunderbare Natur
aus, und eben darum sind sie so ausdrucksvoll—eben darum spiegelt sich
in ihnen das seltsame Verhältnisspiel der Dinge . . . so ist es auch mit der
Sprache ... ." (Halle 1975:527-528).

The view that syntax has a ritual origin and is older than
semantics would explain why there is a syntax in ritual, why
there is an independent level of syntax in linguistics, why
language is so unlogical and—pace Novalis—why language
pictures the world in such a roundabout fashion.
 13

Interpretations of Ritual

The structural properties of ritual that we have so far consid-

ered require not only consummate skill and expertise but also a
lot of the priests' attention—not less than is required, say, from
the members of an orchestra, a ballet company or a team of
engineers set upon the execution of a common task. What then
about the meaning of the ritual, a dimension we read so much
about especially in contemporary works on ritual by anthropol-
ogists and scholars of religion? Are the priests also concerned
with meaning or is this merely a scholarly pastime?
A widespread assumption about ritual is that it consists in
symbolic activities which refer to something else. It is charac-
teristic of a ritual performance, however, that it is self-contained
and self-absorbed. The performers are totally immersed in the
proper execution of their complex tasks. Isolated in their sacred
enclosure, they concentrate on correctness of act, recitation and
chant. Their primary concern, if not obsession, is with rules.
There are no symbolic meanings going through their minds when
they are engaged in performing ritual.
Such absorption by itself, does not show that ritual cannot
have a symbolic meaning. However, also when we ask a
brahmin explicitly why the rituals are performed, we never
receive an answer which refers to symbolic activity. There are
numerous different answers, such as: we do it because our
ancestors did it; because we are eligible to do it; because it is
good for society; because it is good; because it is our duty;
because it is said to lead to immortality; because it leads to
immortality. A visitor will furthermore observe that a person
who has performed a Vedic ritual acquires social and religious
116 Ritual

status, which involves other benefits, some of them economic.

Beyond such generalities one gets involved in individual case
histories. Some boys have never been given much of a choice,
and have been taught recitations and rites as a matter of fact; by
the time they have mastered these, there is little else they are
competent or motivated to do. Others are inspired by a spirit of
competition. The majority would not be able to come up with an
adequate answer to the question why they engage in ritual.
Why ask such personal questions? It might be more proper
and fruitful to ask specific questions about the meaning of
particular rites. Some such questions do receive specific an-
swers, on which participants and scholars agree. The Yajamâna
must keep his hands closed "like a child in the womb of its
mother, ready to be r e b o r n . " The fire altar has the shape of a
bird because fire, as well as Soma, were fetched from heaven by
a bird. The priests do not go south if they can help it for the
southern direction is inauspicious. Certain bricks of the altar are
consecrated so that it may rain (see, however, page 154).
Such simple answers form a small minority. They are given
rarely, and only in reply to similarly simple questions. Most
questions concerning ritual detail involve numerous complex
rules, and no participant could provide an answer or elucidation
with which he would himself be satisfied. Outsiders and bystand-
ers may volunteer their ideas about religion and philosophy
generally—without reference to any specific question. In most
cases such people do not know anything about Vedic ritual.
There is only one answer which the best and most reliable among
the ritualists themselves give consistently and with more than
average frequency: we act according to the rules because this is
our tradition (paramparä). The effective part of the answer
seems to be: look and listen, these are our activities! To
performing ritualists, rituals are like dance, of which Isadora
Duncan said: " I f I could tell you what it meant there would be
no point in dancing i t . "
Ritual, then, is primarily activity. It is an activity governed by
explicit rules. The important thing is what you do, not what you
think, believe or say. In India this has become a basic feature of
 Interpretations of Ritual 117

all religion, so that we should refer, not to the faithful or

orthodox, but to the orthoprax (from Greek orthos, " r i g h t " and
praxis, " a c t i o n " ) . ' It is precisely this feature which is least
understood by English-writing Indian authors such as V. S.
Naipaul and N . C. Chaudhuri, who have recently taken on the
role of explaining India to Western intelligentsia.

13A. Indian Interpretations

If we wish to know the meaning or theory of ritual, we should

not confine ourselves to practicing ritualists; we have learned,
after all, that it does not pay to ask elephants about zoology, or
artists about the theory of art. Before asking anyone else,
however, let us take a look at what the Indian tradition itself has
to offer. Since in India ritual has always been a favorite topic for
speculation, there is an abundance of material. Even prior to
speculation we find suggestive ideas. In the earliest Vedic
literature, rituals, along with metres and chants, are used by
gods and demons to fight and conquer each other, and some-
times to create worlds. Even when the aims are not explicit,
gods and demons are frequently depicted as engaged in ritual.
Commentaries provide rituals with a great variety of interpreta-
tions, sometimes inconsistent with each other.
In Vedic literature, a specific class of texts deals with the
interpretation of ritual. These are the Brähmanas—the same
term that denotes the members of the highest caste who appar-
ently took a special delight in the thinking up of such interpre-
tations. One of the most important of these, the Satapatha
Brähmana, figures prominently in almost all Western works
about Vedic ritual because its translation is widely available and
provides such an enormous wealth of interpretations that almost
everyone can find something to suit his taste or support her
theory.
Julius Eggeling, the translator of the Satapatha Brähmana,
was not enthusiastic about the work to which he devoted a good
part of his life. In the Introduction to the first of his five volume
translation, he wrote about the Brähmanas in general: " F o r
118 Ritual

wearisome prolixity of exposition, characterized by dogmatic

assertion and a flimsy symbolism rather than by serious reason-
ing, these works are perhaps not equalled a n y w h e r e " (I, 1882:
ix). Others have expressed similar views (quoted in A G N I 1:62-
64). Eggeling wrote during the nineteenth century, when reason
was perhaps overrated and irrationalist trends had not yet
invaded the humanities. But his view is not a thing of the past.
Louis Renou, one of the most distinguished Indologists and
Sanskrit scholars of the twentieth century, who wrote exten-
sively on Vedic ritual as well as on literature, philosophy, and all
other aspects of Indian culture, expressed in 1953 a more
balanced and less extreme view that is not less damaging:
We must be content with very general theories if we are to avoid
arbitrary explanations such as those put forward in the old Brâhmanas,
where we find fabricated accountsvof the origin of various details in the
liturgical ceremonial. In these stories there is much that deserves
attention, but the nidäna or bandhu, the hidden connection that they try
to establish, cannot be accepted; it is too visibly the product of the
priestly mind. It is recognized in the texts that comprehension must
cease at a certain point: they declare "paro'ksakämä hi deväh," "the
gods love what is out of sight" (1953:16; with a modification in the
translation).

It is exceedingly, indeed embarassingly easy to provide exam-

ples from the Brähmanas to illustrate their arbitrariness and
inconsistencies. It would be difficult to find the opposite: an
interpretation of a rite that convincingly or persuasively explains
why it is performed. I shall, for the time being, give a single,
simple example of the prevailing ad hoc type of interpretation. It
concerns the puroruc or " p r i o r light" (Chapter 9B, Episode II,
Element 1):
Now, the puroruc is he yonder who gives out light (viz., the sun); for he
shines in (from the) front. Now, the puroruc is the vital breath, the hymn
the body (âtmâ, the person himself)- (Or) the puroruc is the body, the
hymn offspring and cattle (Gonda 1981:63, note 9 = Sreekrishna Sarma
in: AGNI 11:679).

Anyone who has spent any time with the Brähmana literature
will agree that this passage is quite representative. The fact that
 Interpretations of Ritual 119

such interpretations are arbitrary need not prevent them from

constituting a system within themselves. Mylius (1976) has
shown that many identifications given in the same Kausïtaki
Brähmana can in fact be systematically interpreted in social,
psychological, or ideological terms. But even if such interpreta-
tions refer to reality; they still fail to elucidate rites, or throw
light on Vedic ritual.
The three interpretations of the puroruc that are provided in
this interpretation are neither related nor consistent with each
other. The author hands them out without much concern for
their apparent validity. Interpreting in the Brähmanas may be an
art, a kind of game or entertainment; but it is not an attempt
seriously to elucidate or explain. It resembles, in fact, Gada-
m e r ' s hermeneutics (Chapter 3B) in being empty, arbitrary and
ad hoe. This is ironic because G a d a m e r ' s oracular judgment of
all Oriental thought regards it as utterly alien ("Adapting these
things by Western philosophy is out of the question. Only a
negative insight can be taken as confirmed: our own philosophic
concepts, constituted by the Greeks, change the alien in
s u b s t a n c e " : Editor's Note to Dilthey 1949:18).
Some interpretations of the Brähmanas are predictable. They
do not give an acceptable explanation either, but they conform
to a general pattern. Some are concerned with the structural
properties we have discussed, e.g., the structure " B A B . "
Although we have noted that this structure is a simple extension
of a c o m m o n feature of human activity, the Brähmanas interpret
it in terms of an ascent to heaven. Heesterman (1957:12-13)
refers to some of the relevant texts:

The word prâyanïya- not only denotes the introduction to a sacrificial

session or a festival, but it is also associated with the idea of going to
heaven: "through the 'proceeding-day' {prâyanïya-) the gods proceed
{präyäri) to the world of heaven: because they proceeded, therefore it is
called the proceeding-day" (Pancavimsa Brähmana 4.2.2, etc.). The
counterpart of the prâyanïya is generally the udayanîya, the concluding
rite . . . . Through the udayanîya the sacrificer is supposed to reach the
earth again (Kausïtaki Brähmana 7.8, etc.).
120 . Ritual

In India, too, the interpretations of the Brähmanas were

regarded as unsatisfactory. They inspired philosophic or pre-
philosophic speculation in the Upanisads, but were replaced by
systematic and rational analysis in the ritual philosophy of the
Mïmâmsâ, which postulated an unseen fruit as the effect of ritual
activity. We shall return to this in a moment, but must first note
two preceding stages of development, following each other not
unlike the sequence of functionalism and structuralism in twen-
tieth century anthropology.
At first, specific rites were thought of as fulfilling specific
desires: for health, power, offspring, victory and the like. The
list of wishes and desires is not different from that of the majority
of our contemporaries. It is neither entirely material, nor exclu-
sively spiritual as some modern visionaries have claimed. But
this trend receded again into the background, and a structural
point of view came to the fore, which conformed more naturally
to the structure of the ritual and joined the efforts of the ritualists
to create a science of ritual. This led to the codification of two
kinds of rites we have, already mentioned: the Grhya or " d o -
mestic r i t e s , " which are rites de passage, life-cycle rites or
" s a c r a m e n t s , " accompanying such events as birth, initiation,
marriage and death; and the Srauta or "traditional r i t e s , " which
provide the foundation for most of my analysis.
There are several general and formal differences between
these two kinds of ritual. For example, the traditional rites
require three fire altars and the services of several priests,
whereas the domestic rites require only one fire (the domestic
fire) and one priest (the domestic priest). While the function of
the domestic rites appears to be fairly straightforward, the
significance of the traditional rites is not obvious. The traditional
ritual, with its myriad ramifications, exhibits the unhampered
development of ritual construction and creativity. The domestic
rites might seem to be amenable to explanations along the lines
of, e.g., Van G e n n e p ' s Rites de passage (1909) or Victor Turner
(Chapter 14B). But since such explanations are clearly inappli-
cable to the older traditional rites, and domestic and traditional
rites are similar in structure, it follows that all such theories are
 Interpretations of Ritual 121

inappropriate. There are, moreover, traditional rituals which last

a thousand years, which shows that some rites were purely
theoretical. Such theoretical constructs, we have seen, (page
88), are as important for the theory of ritual as are concrete
ceremonies. Many rites have in fact an intermediate status. The
Agnicayana, which was performed in 1975, is a traditional ritual
which seems to have been always " r e a l " , though some of its
extensions, which the texts describe, smack of theory.
The Srauta Sütras of the late Vedic period offer several
definitions of ritual. One which is often quoted characterizes it as
comprising three things: dravya, " t h e substance (which is used
in oblations)"; devatâ, " t h e deity (to which oblations are
offered)"; and tyägä\ "renunciation (of the fruits of the ritual
a c t s ) " . The tyägä is a formula pronounced by the Yajamäna at
the culmination of each act of oblation. When the officiating
priest makes the oblation into the fire for one of the gods, for
example Agni, the Yajamäna says:
4
'this is for Agni, not for me" (agnaye idam na mama).

At this point a contradiction begins to appear, which becomes

increasingly explicit in the ritualistic philosophy of the
Mïmâmsâ. The reason for performing a specific ritual is stated to
be the desire for a particular fruit or effect. The stock example of
the Mïmâmsâ is:

"he who desires heaven shall sacrifice with the Agnistoma ritual"
(agnistomena svargakâmo yajeta).

But this fruit is renounced whenever the Yajamäna utters his

tyäga formula of renunciation. The effect, therefore, is not
obtained.
The resulting picture is further complicated by another appar-
ent contradiction. The rites are subdivided into two classes:
" o b l i g a t o r y " (nitya) and " o p t i o n a l " (kämya). Unlike the Agni-
cayana, which is kämya, the Agnistoma is a nitya rite: every
brahman has the duty to perform it. So here is a ritual which
appears to be optional, since it is confined to those who desire
122 Ritual

heaven (nobody's duty); but which is also not optional, because

it is a prescribed duty; and which moreover in the final resort
does not bear any fruit because its fruits are abandoned. The
texts reflect such contradictions. The Mïmâmsâ Sütra, basic
manual of the Mïmâmsâ, lays down that the rites lead to
happiness, but the subcommentary " Straight S p o t l e s s "
(Rjuvimalâ) observes that this does not apply to obligatory acts.
The Mîmâmsâ philosophers faced another difficulty. When a
ritual performance is completed, no fruit is seen. The Yajamâna,
on whose behalf the rites have been performed, does not raise up
and go to heaven. Rather the opposite: he returns home and is,
as the texts put it, the same as he was before. In particular, he
must continue to perform the morning and evening fire rites
(Agnihotra) for the rest of his life. The Mïmâmsâ concluded,
quite logically, that the fruit of ritual activity is—temporarily—
unseen. It will become apparent only later, e.g., after death. An
elaborate theory was devised to show that this is in accordance
with the mechanism of karman, according to which every cause
has an effect. A special logical theorem, called arthäpatti, was
invented in support of this theory. The followers of the
Mïmâmsâ were criticized by others (e.g., the philosophers of the
Advaita Vedänta) for postulating such unseen effects. For what-
ever cfur contemporary fads may suggest—in India, the unseen is
resorted to only under duress (see page 190). What the Mïmâmsâ
in fact ended up teaching is that the rituals have to be performed
for their own sake.
The notion of tyäga, 4 'renunciation," has attained an impor-
tant position in Hinduism through the teachings of the Bhagavad
Gîta. Here Sri Krsna advocates as the highest goal of life a mode
of activity, in which acts are performed as usual, but the fruit
(phala) of action {karman) is always renounced (karma-phala-
tyäga).

13B. Western Interpretations

The Indian tradition offers suggestive speculations but it does

not come up with a single consistent theory of ritual. The most
 Interpretations of Ritual 123

interesting Indian contribution is perhaps the term karman itself:

originally and primarily used for ritual and similarly pure or ideal
activity, it comes by extension to denote any kind of human
activity. So let us see what modern scholars have to offer. For a
long time it has been fashionable to believe that rites re-enact
myths. This idea was partly inspired by the Babylonian festival
of the N e w Year, which involves a recital of the myth of
creation. But this hypothesis is difficult to support and creates an
unsolved problem: why should anybody wish to re-enact a
myth? The same difficulty applies to several more recent theo-
ries, according to which ritual reflects social structure. It is true,
again, that there are some parallels which require explanation.
But the question remains: why should social structures be
represented or enacted ritually, and in a very roundabout
manner at that? Such unanswered questions, generated by the
theory, suggest that theories of this type are best abandoned.
A related theory, current among anthropologists, is that
rituals are used, in preliterate societies, to transmit "cultural and
social v a l u e s " to the younger generation. This would explain the
informants' emphasis on tradition. But the assumption is, of
course, unnecessary. Not only are rituals not confined to preli-
terate societies (it is anthropologists who tend to confine them-
selves to preliterate societies); but such values (e.g., gods,
myths, kinship systems) are most readily transmitted by grand-
mothers and through language, and there is no need for them to
be transmitted again by other means. The only cultural values
rituals transmit are rituals.
We have seen that " t r a n s i t i o n " or " l i m i n a l " theories of the
van Gennep-Turner type do not explain the Srauta ritual. A
subclass of these theories (found, for example, in Mircea Eliade)
hypothesizes that ritual effects a transition from the realm of the
profane to that of the sacred. (Instead of " t r a n s i t i o n " we also
meet with " c o m m u n i c a t i o n " : a weaker version of the theory).
This is very intriguing and unclear. Terms such as " t r a n s i t i o n "
or " c o m m u n i c a t i o n " do not pose too much of a problem; but
" s a c r e d " and " p r o f a n e " certainly do. One possibility is that the
theory expresses a tautology: the distinction between profane
124 Ritual

and sacred is the distinction between the status of a person or

object before and after a relevant ritual is embarked upon;
accordingly, if sacred and profane have been defined in terms of
ritual, ritual cannot be defined in terms of sacred and profane.
This is circular and uninformative.
On another interpretation, this theory would assume that the
distinction between sacred and profane is already established
and known from elsewhere. For example, in the realm of
divinity, " s a c r e d " might have been shown to be the domain of
the gods, and " p r o f a n e " that of men. But a satisfactory distinc-
tion of this kind is not easily found, especially outside the realm
of ritual. Moreover, the terms do not introduce anything new.
The theory would merely claim that ritual effects a transition
from the realm of men to that of the gods (or a communication
between the two). As a matter of fact, the Vedic ritual offers an
immediate contradiction. During the Soma rituals, a transition is
effected from the "Old H a l l " or small enclosure to the Great
Altar space (see above, page 71 and Figure 1). The former is said
to be the abode of man, and the latter that of the gods. Thus a
transition from the domain of men to that of the gods is effected
within the ritual, instead of characterizing all of it. The distinc-
tion therefore cannot serve as a concept in terms of which the
ritual itself may be defined. The distinction between " s a c r e d "
and " p r o f a n e " exhibits a form of dualism that is typically
Western. Dualisms are also found in Asia but they are surpassed
by monistic and pluralistic perspectives.
Why have Western scholars generally regarded ritual as a
symbolic activity? One reason is that the first Westerners who
speculated about ritual were Christians who regarded ritual
ceremonies as part of their religion and assumed accordingly that
it symbolized religious truths and values. In India, this course
could not be adopted because interpretations or speculations
like those offered by the Brähmanas were never taken literally
and rarely seriously outside the particular tradition of their
perpetrators. The ritual itself continued to be performed but it
was always regarded as more basic than any such speculation,
truth, value or religion (cf. Chapter 28).
 Interpretations of Ritual 125

The sources of the Christian preoccupation with the symbolic

interpretation of ritual lie in the Carolingian Renaissance, when
the idea was first formulated that the ceremonies of the church
express and symbolize the coming of Christ and the history of
salvation. With the Protestant emphasis on scripture and faith,
ritual receded to the background. In more recent developments
within Christianity, Protestant as well as Roman Catholic, the
place of ritual or liturgy continues to be a topic of discussion. In
India, too, ritual or karman has been debated ceaselessly by
philosophers and theologians. This goes back as far as the early
Upanisads and the beginnings of Buddhism in the sixth century
B.C.E.
The first Western philosophers and scholars outside the Chris-
tian tradition who recognized the importance of ritual continued
to place it unquestioningly within the domain of religion. For
Hegel, ritual stands at the center of the religious process, where
the subject "participates in the absolute and is united with i t . "
E. O. James (1917:215) expressed his views in less pompous but
equally vacuous terms: "Generally speaking ritual evolved long
before belief, since primitive man is wont to 'dance out his
religion' " (both quoted in Cassirer 1925, 11:271). The first
serious scholar who demonstrated the primacy of ritual over
belief was W. Robertson Smith, author of a nineteenth century
classic, the Lectures on the Religion of the Semites of 1889. This
view was incorporated in his philosophy by Ernst Cassirer, who
regarded man primarily as a "symbolizing a n i m a l " but accepted
at the same time that ritual was prior to " d o g m a , " both in a
historical and in a psychological sense (Cassirer 1925, 11:270—
285). For Cassirer, therefore, ritual is to be interpreted in
symbolic terms, but since the symbols cannot refer to features of
belief, the question arises what do they refer to?
Doubts were cast on the idea that ritual is part of religion and
on the symbolic character of ritual by the founders of the
"sociology of religion," Emile Durkheim and Max Weber.
Durkheim was clearer and more methodical than Weber, his
slightly younger contemporary, but he did not go as far. Though
he saw clearly that religion is related, through ritual among other
126 Ritual

things, to social facts and institutions, he did not question the

view that ritual is part of religion. On the contrary, he formulated
that view with classic succinctness at the outset of his book Les
formes élémentaires de la vie religieuse of 1912:

Religious phenomena are naturally arranged in two fundamental catego-

ries: beliefs and rites. The first are states of opinion, and consist in
representations; the second are determined modes of action. Between
these two classes of facts there is all the difference which separates
thought from action (Durkheim 1912; English translation of 1915:36).

Following this statement of principle, which he adopted in the

organization of his book, Durkheim briefly and tentatively
discussed the two categories, but did not question their relation-
ship: " I t is possible to define the rite only after we have defined
the belief."
We have seen in Chapter 2 that Durkheim, given his empiri-
cist/positivist bias, omitted mystical experience from his classi-
fication of religious phenomena. Here we are concerned with
another prejudice that underlies his approach.
Durkheirrf s view of the primacy of belief over ritual, which
was a step back from the position that had already been reached
by Robertson Smith and others, has remained the preponderant
view that underlies Western studies of ritual. One reason for this
preponderance is that this view combines more easily with the
belief that ritual is symbolic than that it is not: for it can then be
simply maintained that ritual is a symbolic representation of
what people believe. In his discussions of ritual, Durkheim did
not make much use of the category of " s y m b o l , " but he referred
throughout his book to rites in terms of the human mind, e.g.:
" R i t e s are a manner of acting which take rise in the midst of the
assembled groups and which are destined to excite, maintain or
recreate certain mental states in these g r o u p s " (page 10), or: t 4 At
bottom, all these different practices are only variations of one
and the same theme: everywhere their basis is the same state of
mind, interpreted differently according to the situations, the
moments of history and the dispositions of the w o r s h i p p e r s "
(page 388).
 Interpretations of Ritual 127

There is a curious consequence of this approach to human

ritual. Basing himself upon earlier ethnographic and ethnological
investigation, Durkheim paid much attention to the totemic cult.
Of his nine chapters on "elementary beliefs," seven deal with
" t o t e m i c beliefs." Totemic cults involve a close union between
men and an animal species. If anthropologists had tried to
interpret this phenomenon in the light of the natural relatedness
that exists between the human animal and other animals, they
would have welcomed the facts about animal ritualization when
these became known. But since they emphasized symbols and
the human mind, anthropologists have been resistant to the idea
that animal ritualization has anything to do with human rituals:
for this would then involve the assumption that animals also
have human minds and beliefs. To a logician, a different conclu-
sion follows through contraposition: since animals have ritual,
but do not have human minds or beliefs, ritual cannot be
symbolic.
Durkheim's own analysis of rites throws doubts on his theo-
ries about ritual. Discussing data from the Australian tribe of the
Arunta made available by Spencer and Gillen, he focussed on a
significant feature that the authors themselves did not seem to
have noticed. In their first book, these ethnographers had
analyzed a ceremony called intichiuma, which was "destined
exclusively for the assurance of the reproduction of the totemic
species, and it seemed as though it ought to lose all meaning, if
this unique function were set a s i d e " (Durkheim 1915:383). But in
a later work by the same authors, the same ceremony was
discussed as part of the initiation ritual, in which function it was
called by a different name: mbatjalkatiuma (page 384). Durk-
heim concluded that "according to the circumstances, one and
the same ceremony serves two distinct functions" (page 385).
H e referred to Hubert and Mauss, who had already pointed out
a functional ambiguity of the same sort in the " H i n d u sacrifice,"
and then expanded his conclusion as follows:

This ambiguity shows that the real function of a rite does not consist in
the particular and definite effects which it seems to aim at and by which
128 Ritual

it is ordinarily characterized, but rather in a general action which, though

always and everywhere the same, is nevertheless capable of taking on
different forms according to the circumstances, (page 386)

The idea that a single rite serves different ends would seem to
suggest that ritual is to some extent independent of the ends it is
supposed to serve; but Durkheim did not draw this conclusion or
even consider such a possibility. Neither did van Gennep, who
had made the same discovery two years earlier, when he
observed that: " t h e same rite, remaining absolutely the same,
can change its meaning depending on the position it is given in a
ceremony, or on whether it is part of one ceremony or another.
The aspersion rite . . . is a fecundity rite in marriage ceremonies,
but an expulsion rite in separation c e r e m o n i e s " (Van Gennep
1910, in: Waardenburg I, 1973:299).
Van G e n n e p ' s observation remained virtually unknown, since
it occurred in a lecture delivered at the University of Brussels,
which was published in French and Dutch in 1911, and in English
translation only in 1973. An English translation of Durkheim's
work, on the other hand, was published in 1915, three years after
the original had appeared. Durkheim's work was very influential
in France, and became widely known in the United States after
the Second World War. But Durkheim's observations on the
functional ambiguity of rites have not attracted any attention,
and nobody has detected their important theoretical implica-
tions.
Weber studied religion within the context of society and in a
similar spirit as Durkheim had done. However, opposing current
theories, he did not take it for granted that symbolization was
one of its important features. At the outset of his book on the
sociology of religion, he interpreted magic, which is closely
related to rites, as a direct manipulation of forces, which ai first
reflects some kind of naturalism, and is subsequently trans-
formed into a symbolic activity:

Various consequences of significance to magical art emerged from the

development of a realm of souls, demons, and gods. These beings cannot
be grasped or perceived in any concrete sense but manifest a type of
 Interpretations of Ritual 129

transcendental being which normally is accessible only through the

mediation of symbols and significances, and which consequently is
represented as shadowy and even unreal. Since it is assumed that behind
real things and events there is something else, distinctive and spiritual, of
which real events are only the symptoms or indeed the symbols, an effort
must be made to influence, not the concrete things, but the spiritual
powers that express themselves through concrete things. This is done
through actions that address themselves to a spirit or soul, hence done by
instrumentalities that 'mean' something, i.e., symbols. Thereafter, nat-
uralism may be swept away by a flood of symbolic actions. (Weber 1922;
English translation, 1963:6-7)

Waardenburg (1973, 1:49) has correctly emphasized that We-

ber's was u a new view of religion which worked against a
German tradition stressing the mythological and symbolic as-
pects of religion." An important implication of this view is that
ritual has to be primarily understood without reference to the
symbolization which was attached to it afterwards. Weber
himself, however, wavered, and did not draw any such conclu-
sion. Both he and Talcott Parsons in his influential Introduction
to the 1963 American edition of the Sociology of Religion were
more interested in the later symbolic derivations than in the
original "pre-animistic naturalism" somewhat murkily postu-
lated by the theory. Weber had become famous on account of his
ideas about the relatedness between Protestantism and Capital-
ism, welcomed especially in the United States where they were
interpreted as an endorsement of both. H e came to issue
pontifical statements no longer consistent with his original
insights, e.g.: " E v e n t s are not just there and happen, but they
have a meaning and happen because of that m e a n i n g . " Such
statements have become part of the credo of anthropology in the
United States, where they are highlighted in popular readers on
cultural anthropology with sentimental titles such as Man Makes
Sense (Hammel and Simmons 1970).
The fashionableness of such views should not obscure the fact
that they are wholly atavistic and metaphysical. N o scholar or
scientist has adequate grounds for assuming that the data he
seeks to interpret and explain are meaningful to begin with.
Nature and culture are governed by chance as much as by
130 Ritual

necessity. Openmindedness with regard to such questions is

required as a matter of course in the natural sciences. In the
human sciences, such openmindedness cannot be circumvented
by defining man a priori and dogmatically as " t h e animal that
makes s e n s e . " The unproductiveness of such an approach is
particularly clear in biology, where many of the most interesting
problems, which also involve the human animal, revolve around
the question whether some of the phenomena can be shown to
" m a k e s e n s e , " or are merely due to chance (see, e.g., Monod
1970). We shall return to this topic (Chapters 24 and 30C).
Durkheim and Weber created the sociology of religion, and it
is largely due to them that religion and ritual are nowadays
studied not only by scholars of religion and religious scholars,
but also by anthropologists and sociologists who should have
less of an axe to grind. This development constitutes a significant
and valuable widening of perspective. However, Durkheim's
return to a view that had already been abandoned by others for
good reasons, and Weber's inconsistencies have also created a
great deal of confusion. In the study of ritual, in particular, it is
not possible to derive much help from their insights. The real
issues continue to be obscured by the continuation of symbolic
interpretations, due not only to the facts that Durkheim and
Weber had wavered, and that the break with the past had not
been radical, but also because of the general search for meaning
as a substitute for religion that characterizes Western man in the
twentieth century.
The only scholars who freed themselves from these overbear-
ing preoccupations of their own culture were fieldworkers who
immersed themselves for extended periods of time in a foreign
culture without losing their rational outlook and common sense.
Malinowski, for example, originally a physicist and accordingly
equipped with a good measure of scientific training, spent
several years with the Trobriand Islanders and in other parts of
the Pacific. He recognized some of the functions of ceremonies
in terms that are reminiscent of Durkheim but that are much
clearer than anything either Durkheim or Weber had written
about ritual:
 Interpretations of Ritual 131

We may, therefore, lay down the main function of initiation ceremonies:

they are a ritual and dramatic expression of the supreme power and value
of tradition in primitive societies; they also serve to impress this power
and value upon the minds of each generation, and they are at the same
time an extremely efficient means of transmitting tribal lore, of insuring
continuity in tradition and of maintaining tribal cohesion. (Malinowski
1948, in Waardenburg 1973, 1:555)

This statement rings true, but though it makes use of the notion
of " e x p r e s s i o n , " it has nothing to do with symbols or meaning.
It is partly applicable to other animals than men, and is moreover
not only true of other rites, but applies to many human institu-
tions (universities, for example).
All these different unsuccessful attempts at characterizing
ritual teach a lesson already known to philosophers: symboliza-
tion requires minds and beliefs. Therefore, to grant that ritual is
prior to belief, but to persist in trying to interpret it in terms of
symbols is a hopeless task. If ritual is prior to belief, as it
happens to be in the scheme of evolution, it must be interpreted
in different terms.

13C. The Meaninglessness of Ritual

Why has it proved so difficult to define the meaning, goals and
aims of ritual? Why are there so many different answers and
theories, not only often contradictory between themselves, but
of such disparate character that it is difficult to even compare
them with each other? There is one simple hypothesis which
would account for all these puzzling facts: the hypothesis that
ritual has no meaning, goal or aim.
This is precisely what appears to be the case. Ritual is pure
activity, without meaning or goal. Let me briefly digress for a
point of terminology. Things are either for their own sake, or for
the sake of something else. If I were defending the view that
ritual is for something else, it would be necessary to distinguish
between such other things as meaning, function, aim or goal. But
since my view is that ritual is for its own sake, I shall not bother
about these differences. To say that ritual is for its own sake is
to say that it is meaningless, without function, aim or goal, or
132 Ritual

also that it constitutes its own aim or goal. It does not follow that
is has no value: but whatever value it has is intrinsic value.
Ritual exhibits its character of pure activity most readily when
it is contrasted with the applied activities of our ordinary,
everyday life. In ritual activity, the rules count, but not the
result. In ordinary activity it is the other way around. In Vedic
ritual, for example, an important ceremony is agnipranayana,
"transporting the fire (from the Old to the N e w Altar) (page
7 5 ) " . This is sometimes described as a transition from the abode
of men to that of the gods. But the priests do not first think of
men and then mediate on the gods. They think of neither, at any
time. What is essential in the ceremony is the precise and
faultless execution, in accordance with rules, of numerous rites
and recitations. The result is important, but it has only ritual use
and can only be reached in the ritually prescribed manner. I
could not come in and assist in the proceedings by picking up the
fire from the Old Altar and depositing it on the N e w . In fact, if
I did such a horrible thing, the entire ceremony would be
desecrated, interrupted, and expiation rites would have to be
performed. Similar disasters would result if anyone used the
sacred fire for any but a ritual purpose, e.g., to heat water for
tea.
N o w contrast this with an ordinary activity. I am about to
transport my suitcase from my house to the bus stop, which is
about a mile away. There are no rules I have to follow, provided
I obtain the desired effect. I may put my suitcase on a skate
board. Or my brother may appear on a bicycle, and the two of us
use this vehicle to transport my suitcase to its intended destina-
tion.
The two kinds of activity, ritual and ordinary, can be juxta-
posed without conflict or contradiction. After making fire for the
altar in the ritually prescribed manner by rubbing two pieces of
wood together, a priest leaves the sacred enclosure and lights a
cigarette with a match. Not so different, actually, from Arthur
Rubinstein back home after a concert, putting on a gramophone
record. But the two domains should not be mixed. If a priest
would light a cigarette from the sacrificial fire, it would be bad.
 Interpretations of Ritual 133

If he would light a cigarette from the fire which he had produced

by rubbing two pieces of wood together in the ritual manner, he
would be considered mad or very eccentric. The ritual and
ordinary ways of making fire are neatly demarcated.
A distinctive feature of ordinary activity is that it runs risks
which ritual activity avoids. In ordinary activity, the entire
performance may fail to have the desired effect. The bicycle
together with its load may fall into a canal, or the suitcase may
be seized by armed robbers. In ritual activity, the activity itself
is all that counts. Not only have we established the rules
ourselves, so that we are completely in control; we are also
assured of success. If one rite goes wrong, another takes it place.
This goes a long way to explain the curious fact that rituals, so
apparently meaningless and useless, are at the same time readily
engaged in. Eo ipso it explains that ritual activity has a pleasant,
soothing effect. If you give up desire, you will be happy. This
idea and the notion that ritual is performed for its own sake are
closely connected and clearly foreshadowed by the Indian
doctrine of tyäga, the teachings of the Bhagavad Gîtâ, and by
similar notions in other traditions, e.g., wu-wei, " a b s e n c e of
(effective) a c t i o n " in Taoism, or the categorical imperative in
Kant. It also accounts for the similarity between rites and
games, which are equally unproductive, as Huizinga (followed
by Caillois) pointed out (see below page 427). But ritual is one up
on most games because you cannot even lose.
Several anthropologists have detected features of meaningless
in ritual, without recognizing that these features express its
essence. Lévi-Strauss says that ritual "consists of utterances,
gestures and manipulations of objects which are independent of
the interpretations which are proper to these modes of activity
and which result not from the ritual itself but from implicit
mythology" (1971:600). If we remove the word "implicit" from
this sentence (which means forsaking the author's ideas about
the complementarity of myth and ritual) we approximate what I
believe to be the correct theory. Van Gennep came close to the
idea that ritual is meaningless. After completing his Rites de
passage, he noted that marriage ceremonies, in many societies,
134 Ritual

include an aspersion rite which he interpreted as a fecundity rite.

But identical aspersion rites are employed, in the same and in
different societies, when a slave is acquired, when a new
ambassador arrives in town, to make rain or to expel someone.
Like IncMan commentators, Van Gennep gave different interpre-
tations to each of these rites. H e concluded: " t h e aspersion rite
does not have any personal or basic meaning in the state of
isolation, but it is meaningful if seen as a component part of a
particular ceremony. The meaning of the rite can, consequently,
only be found by determining the relation it has with the other
elements of the whole ceremony"(1911:520; above page 128).
Aspersion rites are not confined to humans. In his Sather
lectures at Berkeley, Walter Burkert (1979;43) dealt with the
ritual pouring of liquids for marking a territory and observed that
this is quite common in mammals: " w e all know the dog's
behavior at the s t o n e . " — I n the development of our concepts
and theories of ritual it is only a small step from "changing
m e a n i n g " to: " n o intrinsic meaning" and "structural m e a n i n g , "
and from there to: " n o meaning."
If ritual is useless this does not imply that it may not have
useful side-effects. It is obvious, for example, that ritual creates
a bond between the participants, reinforces solidarity, boosts
morale and constitutes a link with the ancestors. So do many
other institutions and customs. Such side-effects cannot be used
to explain the origin of ritual, though they may help to explain its
preservation. They explain why rituals are preserved though
their meaninglessness is recognized, like the Jewish ritual of the
Red Heifer which baffled even Solomon and which was consid-
ered the classic example of a divine command for which no
rational explanation can be adduced.
These side-effects fail to explain the most curious fact about
ritual preservation: rituals are always guarded jealously and with
extreme conservatism. This is directly explained by the theory
that ritual has no meaning. A useful institution is open; it may
undergo change, because efforts are made to render it more (or
less) useful. A useless institution is closed; it is not understood
and therefore can only be abandoned or preserved. There are
 Interpretations of Ritual 135

parallels to this situation from outside the realm of ritual. In

India, during the last 3000 years, the Vedic language gave way to
classical Sanskrit which was in due course replaced by Middle
and Modern Indo-Aryan languages. During all these changes the
Vedic mantras were orally transmitted without any change.
Why? Because they were meaningless. Languages change be-
cause they express meaning, are functional and constantly used.
Meaningless sounds do not change; they can only be remem-
bered or forgotten.
Freud has drawn attention to similarities between ritual and
neurosis. The obsessiveness which pervades ritual has led
several anthropologists to emphasize the emotions and anxiety
which sometimes accompany ritual, and which they claim un-
derlie it. In U homme nu, Lévi-Strauss has located such anxiety
in the ritualists' fear that reality, which they have cut up ritually,
cannot be put together again. But it is apparent that the
obsessiveness of ritual is also an immediate consequence of its
meaninglessness. Nothing is more conducive to uneasiness than
to be entrapped in absurdity. If I detect a mistake in cooking or
calculating, I perceive the result and understand the reason. But
if I have made a ritual mistake, I don't notice any difference and
don't see any reason. I am not even sure whether I made a
mistake or not, and there is no way to determine it. It is like
being in a foreign culture where strange things happen and it is
not clear whether one has made a faux pas. The Agnicayana
performance of 1975 was followed by a long series of expiation
rites, for mistakes that might have been committed. Our anxiety
is greatest when we don't know why we are anxious.
The meaninglessness of ritual explains the variety of meanings
attached to it. It could not be otherwise. Ideal activity cannot fail
to resemble actual activity. Therefore rituals resemble other
things, including features of myth and social structure. How-
ever, though a ritual activity may resemble a meaningful non-
ritual activity, this does not imply that it must itself be mean-
ingful. This can be seen in the realm of animal ritualization, as
well as in the human domain. Among animals, ritualization often
implies that the goal of an activity has changed. Many ritual
136 Ritual

displays incorporate modes of action which originally had a

different function, e.g., fighting. Such ritual displays may ac-
quire a new function: they lead to copulation because they are
sexually stimulating, for example. Some of the same ritual
displays, however, are post-nuptial or post-reproductive, and
therefore not clearly functional. Biologists find them puzzling
(e.g., Huxley 1966:254 and below, page 287).
H u m a n ritualization often follows animal ritualization rather
closely. Fighting, simulated or real, is still sexually stimulating
among humans. But typical human forms of ritualization seem in
general to dissolve meaning, not replace it. One of the earliest
rituals originated in connection with the use of fire. During most
of its existence, mankind did not know how to use it. Subse-
quently, more than 250,000 years ago, man learned the use of
fire; but he could not make it. So fire was collected from natural
conflagrations and was carefully kept and carried around. Elab-
orate techniques were devised for the preservation of fire.
Finally, more than 50,000 years ago, man learned how to make
fire. At this point ritualization and the cult of fire came into
being. For instead of relying on his art of making fire, and
producing it whenever he needed it (which is easy at least during
a dry season or in a dry climate), man continued to carry fire
around. A distinction was made between such " e t e r n a l " fire and
the " n e w " fire which could now be made—a distinction we have
since abandoned as irrational. To ancient man, and in several
existing societies, fires have retained individuality. They should
not be mixed. Fires have to be extinguished, or newly made, at
set times by ritual experts. Alongside, the continued preserva-
tion of " e t e r n a l " fire reflects fossilized habits which had lasted
some 200,000 years.
A more recent example comes from the Agnicayana (cf.
Heesterman 1962:34-36). During the ceremony of agniprana-
yana, when fire is transported from the Old to the N e w Altar,
one of the priests engages in a long recitation. The recitation is
of an ancient battle hymn, the Apratiratha or " S o n g to the
Irresistible W a r r i o r " (Taittiriya Samhitâ 4.6.4, cf. Rigveda
10.103 and 6.75). Indra is invoked as a victorious warrior or
 Interpretations of Ritual 137

hero, "fond of slaughter, disturber of p e o p l e s " , who with the

help of his arrows, chariots and troups, destroys the enemies.
When the priest recites: " C o m r a d e s , follow in Indra's
footsteps!", he sounds less like an officiating priest than like a
gang leader or a commander-in-chief. And what is the origin of
all of this? At an earlier period, the Vedic Aryans fought their
way into the Indian subcontinent, moving from west to east and
carrying fire. In the agnipranayana rite, fire is still carried from
west to east. But the priests are not celebrating the ancient raids
of their ancestors, of which they need not even be aware. The
function of the hymn has not changed. It has become ritual, i.e.,
disappeared.
Can the meaningless of ritual be formulated in terms of
development or evolution? Necessarily so, but we must specu-
late about origins. Originally, ritual was a mere activity per-
formed by animals in accordance with rules. Among, h u m a n s ,
when contrasted with ordinary, everyday and purposeful activ-
ities, its meaninglessness became patent, and rationalizations
and explanations were constructed to account for its persis-
tence. The chief provider of meaning being religion, ritual
became involved with religion and through this association,
meaningful. Next, rites were attached to important events which
thereby acquired religious meaning, too. In the course of time,
rituals, instead of remaining useless and pure, became useful and
meritorious.
Through the history of m a n ' s speculation on ritual we find
inklings of this background. Just as the Indians mused about
Srauta rites, the Chinese theorized about //, " r i t e s , ceremonies,
etiquette; rules of good manners and proper c o n d u c t . " The
Confucian philosopher Hsün Tzü (third century B . C . E . ) ex-
plained the origin of the li as follows:

Man at birth has desires. When these desires are not satisfied, he cannot
remain without seeking their satisfaction. When this seeking for satis-
faction is without measure or limit, there can only be contention. When
there is contention, there will be disorder; when there is disorder,
everything will be destroyed. The early kings hated this disorder, and so
they established the // and standards of justice so as to set limits to this
138 Ritual

confusion, to satisfy men's desires, and give opportunity to this satis-

faction, in order that desires should not be stretched to the breaking
point by things, nor things be used up by desires; that both these two
should mutually support one another and so continue to exist. This is
how the // originated (Fung Yu-lan 1952, 1:297).

We have seen that there are structural similarities as well as

dissimilarities between ritual and language. The similarities can
be explained by assuming that the origin of syntax is at least
partly ritual. We have also seen that syntax is the part of
language which stands most in need of explanation. Language
relates sounds to meanings, and so it must necessarily comprise
a domain of sounds (studied in phonetics and phonology) and a
domain of meanings (studied in semantics). If language were
rational and adapted to its purpose, sounds and meanings would
be related by means of a one-one-correspondance. If this were
true of natural languages, assuming semantics to be universal,
different natural languages would also stand to each other in a
one-one-correspondence, and translation could be effected with
the help of dictionaries only. There would be no need for
artificial languages, and logicians would be out of business.
What we do find in language is something different. Meanings
and sounds are related to each other through a vast and
complicated domain of structured rules: syntax. The transition
between sound and meaning is unnecessarily complex, round-
about and mathematically absurd. " N o b o d y in his right mind
would attempt to build an artificial language based on phrase
structure plus transformations" (J. A. Fodor in: Smith and
Miller 1966:270; cf. Chapter 6A).
How are we to explain such apparent redundancies? It is not
enough to say, as communication theorists might, that redun-
dancies are necessary for communication because they decrease
mistakes in reception. That assumes that language is only for the
sake of communication, which it is not. More importantly, such
redundancies, to perform their alleged function, merely need be
random: which cannot explain syntax, a structured domain of
specific rules which in fact makes language unlogical and ineffi-
cient. These specific rules, which are without rhyme or reason,
 Interpretations of Ritual 139

must come from elsewhere. They look like a rudiment of

something quite different. This supports the idea that the origin
of syntax is ritual.
The ritual origin of syntax has implications not only in
language but also in religion. I shall mention three. Ritual is
replete with language, but it is very often meaningless language.
When a small golden image of a man is buried under the fire altar
of the Agnicayana, the Priest of the Sämaveda chants songs
which contain such sounds as:

Kä hvä hvä hvä hvä hvä

phal phal phal phal phal
hau hau hau hau hau
bham bham . . . (eighteen times).

Such structured sounds partake of the syntax of ritual, but do

not relate to meaning. This applies to most mantras. The topic of
mantras is large and fascinating and we shall return to it in Part
III.
A second feature is that mysticism is characterized by the
absence of language. It points to a pre-linguistic state which can
be induced by ritual, by recitation, by silent meditation on
mantras, or by other means, as I have shown in Exploring
Mysticism (1975). All these methods help to eliminate meaning,
sound and (ritual) structure.
Wittgenstein had an inkling of the place which language
occupies in religion when he remarked:

Is speech essential for religion? I can very well imagine a religion in

which there are no doctrines, and hence nothing is said. Obviously the
essence of religion can have nothing to do with the fact that speech
occurs—or rather: if speech does occur, this itself is a component of
religious behavior [the German original has: Handlung, "activity"] and
not a theory. Therefore nothing turns on whether the words are true,
false or nonsensical (Waismann 1965:16).

The ritual origin of syntax is connected with another curious

fact, which I mentioned in passing (page 117): the Vedic gods
140 Ritual

fought and created not only with ritual, but also with meters and
chants. What an extraordinary thing to do! But no, it is not.
Meters and chants are like ritual in that they fail to express
meaning, but reflect syntactic structure in its pure form, hence
pure activity.
We conclude that the unchanging syntactic structures of ritual
are consistent with a great variety of meanings, artificially
attached to them. In his studies of the development of Taoist
ritual through a succession of historical contexts, Kristofer
Schipper has used the expression accretion of meaning to refer
to this constant search for new significance. This abundance of
meanings is tantamount to the nonexistence of a single intrinsic,
basic and necessary meaning. Rituals, then, are different from
the expressions of a language: for the latter refer to meanings in
a manner that has become nonarbitrary, and language can only
function because of that fixation. De Saussure rightly empha-
sized that the sounds of signs are arbitrary. Meanings are
attached in haphazard ways. But this refers to origin and
etymology, and in the resulting system everything has its fixed
place. Language can only be effectively used because it refers to
meaning in a systematic manner.
 14

Anthropology Without Asia

H o w , then, is it possible to understand ritual without inter-

preting it in terms of symbols, meaning, or sense? In order to
achieve such an understanding we have to do three things, more
or less at the same time: first we must have an open mind with
regard to the conceptual question where ritual ' ' b e l o n g s . " We
should detach it in particular from those domains where our
culture and history have been predisposed to place it: in the
realms of religion and society. Second, we must study ritual in
much greater depth than is done by the professional students of
religion and society. And third, we should conceive of ritual in
more general and abstract perspectives than has ever been
attempted. The third requirement remains to a large extent a task
for the future. The first two are briefly discussed in the present
chapter.
Detaching ritual from religion, anthropology and sociology is
not merely a mental exercise, but a methodological prerequisite.
Since it amounts to a basic change of perspective, it requires
special effort. Fortunately there are two kinds of circumstantial
evidence that come to our assistance in attempting to achieve
such a turn of mind. The first consists in the realization that the
interpretation of particular rites by scholars of religion and
anthropologists are often inconsistent with each other, and are
sometimes different within each of these fields. This holds also
for the Indian Brähmana literature, as we have seen. The
"functional ambiguities" detected by Van Gennep and Durk-
heim provide more explicit illustrations of these difficulties.
Once such inconsistencies of interpretation are met with, we
142 Ritual

should be prepared to accept that we may be barking up the

wrong tree.
The second kind of circumstantial evidence that may help us
to detach ritual from religion and anthropology comes from the
concept of ritualization that has been adopted in the description
of animal behavior. This has led to a new science, ethology,
whose roots lie in a nineteenth century classic: Darwin's Expres-
sion of the Emotions in Men and Animals of 1872. Contemporary
zoologists have interpreted ritualization as a change in the
function of a pattern of behavior, often serving the purpose of
communication. We should be careful, however, that we study
the parallels between human ritual and animal ritualization
without necessarily adopting the interpretations of the scholars
and scientists who provide us with the data. Just as the rites of
men may have nothing to do with religion or social structure, the
rites of animals may have nothing to do with emotions or
communication. What is uncontroversial is only that the detailed
formal parallels between human and animal rites and rituals cast
doubt on all religious, anthropological and sociological interpre-
tations, since for any of these to be true it would be necessary to
assume that animals that perform similar rites as men have
similar religious or social characteristics. Such assumptions are,
of course, far-fetched outside the realms of mythology and the
fairy tale. Animals do have aspersion rites, for example, but
these express neither fecundity in marriage ceremonies, nor
expulsion in separation ceremonies.
We have argued that Indian and Chinese rituals provide the
best data for the general study of ritual (pages 69-70). But when
we try to place these data in their proper context, we have to
study the large and complex civilizations and societies of India
and China that have been made accessible successfully and
fruitfully by specialists who have only seen the trees. The
neglect of these societies by sociologists and anthropologists,
who are trained to see the forest, is largely due to the peculiar
way they originally divided up their object of study: sociologists
dealt with Western and Westernized (i.e., literary and advanced)
societies, and anthropologists with non-Western (i.e., nonliter-
 Anthropology Without Asia 143

ary and nonadvanced) societies. But the dichotomies do not fit

or coincide: India has been throughout most of its recorded
history an advanced nonliterary society, and China, though both
literary and advanced, is not Western. The main civilizations of
Asia thus fell between the stools or were left to philologists who
deal only with texts. Weber constitutes a rare exception to this
general state of affairs. But it will not do to neglect India and
China—the largest societies on the planet, the most important
for its future, and the richest in historical depth.
Anthropologists and sociologists have failed to study rituals in
depth, both with respect to their historical development and
their contemporary richness. Jonathan Smith, who refers in this
connection to " s n a p s h o t anthropology," makes an exception for
Lévi-Strauss, who insisted that a ritual ceremony be " r e c o r d e d
and transcribed in its e n t i r e t y , " which may "fill a whole volume,
sometimes a very large o n e " and take " d a y s " to perform (Smith
1987:193). Lévi-Strauss cannot be criticized for neglecting the
complexity of ritual, at least in its contemporary forms, but it is
striking that in his voluminous works, which scan the planet
from the Amazon to Paris, the civilizations of India or China are
hardly mentioned.
I shall illustrate snapshot anthropology outside Asia from the
work of two leading anthropologists, one British and one Amer-
ican, whose data are among the best 'data on ritual in contem-
porary social science. It is not my aim to criticize these scholars
individually, but to illustrate the inherent limitations of the
anthropological approach to the study of ritual.

14A. Raymond Firth

Raymond Firth, pupil and successor of Malinowski at the

London School of Economics, wrote several books and articles
on Tikopia, a small Polynesian island with a population of little
more than 1,200 and comparatively isolated from Western
influence.
Among the ceremonies the people of Tikopia perform for
children there is a "fire formula" recited to a new born baby by
144 Ritual

an elder while an assistant lights a torch so that the light falls on

the face of the child. The reciter than addresses the man who
holds the child and his principal deity as well:

Pa Fetauta! Assist me at the fire of Pa Tafua which is being set up.

Unfold welfare here.

This is followed by a long recitation in which there is some

variation, and which Firth has therefore described as a "free
formula":

Thy fire Pa Tafua has been announced to thee

Unfold welfare
That the child may sleep well.
Thy fire Pa Saukirima O!
Recite hither for welfare for her
To sleep well.
My own fire Pa Korokoro
Light be your eyes
Unfold welfare for her
Perfume a thing for her to eat
That the child may sleep well.
From its recitation that has been performed
Clear be its eyes for the work.
Wake up you for the taro
To be industrious in the clearing of the taro
To be healthy for your work.
Wake you up to go and clear for taro
The taro which stands overgrown
To be cleared on the moment and finished
Wake you up and the seedlings that I filled tie up completely to be
carried on your back
Hasten you and when your parents are hungry go to the woods to
gather food for your parents to come and cook it in the oven to be
done quickly for your parents who hunger to be filled.
To go and get food
To go to the water
To carry a water-bottle
For your parents to drink.
Be fit in your water-carrying
Stride off, stride back
We who are thirsty
Now have become filled.
 Anthropology Without Asia 145

Climb up there to the mountain standing there

That you may be fit simply to get food.
Cut the leaves of giant taro which stands, to lever it out and proceed
hither.
Light the oven
Cook it till it is done
Go and fetch the food-kits of your parents to put it in
Then we are filled.
Go and fetch the water-bottle
Then we have drunk
Then we are filled.
Indeed you are industrious.
Stand at the oven-border to uncover the oven at once to fill the food
baskets
Go and give them to your parents
Fill the baskets of the relatives
Clear be your eyes for the fishing
The fish goes to a distance but bar it to dash hither to rest in your net
We have eaten of your netting
The fish that goes to a distance be turned by you to enter the net
When your parents are hungry, go and catch fish
Take your torch, go and take your net to go and fish.
Go and parcel up; parcel up a package for your parents and give them
to eat
Parcel up a package for us relatives to carry
Parcel up a package for your brothers
Parcel up a package for your sisters
There they have eaten completely
There have eaten completely the relatives, and I have eaten of the
package
Go and roast, there it is cooked
Give to eat, there I am filled
And distribute then to the relatives
There they have eaten completely
Go and give to your grand-parents; go and give to your fathers; go
give a package to your mother; go and give a package to your
brothers.
Then if there is one left, go and give a package to your brother
There we have eaten of your food-procuring.
(Firth 1967:53-55)

Firth observed these ceremonies in 1928-29. By 1952-53, when

he visited Tikopia again, he found that they had been much
146 Ritual

abbreviated. The question which naturally arises is whether the

1928-29 performances themselves were abbreviations of earlier
forms. Firth has not raised this question, but he says in general
about the historical depth of his data: " F o r about two centuries
back from the present time the data are fairly clear, but beyond
this they become very imperfect and soon can be regarded only
as m y t h " (page 28).
Using data such as these, Firth has formulated his views on
ritual in tentative terms in his Introduction:

The notions of ritual and its close analogue ceremonial have been
much debated in social anthropology. My own view is that the most
convenient way to look at ritual is to consider it as a formal set of
procedures of a symbolic kind, involving a code for social communica-
tion, and believed to possess a special efficacy in affecting technical and
social conditions of the performers or other participants. The formality
of these procedures lies in the fact that they are directed not simply to the
solution of an immediate technical problem by the most economical
means, but are regarded as having in themselves a certain validity
irrespective of their technical concomitants. Because of their general
validity, apart from the individual situation, they tend to be given a
repetitive routinized character which in itself is regarded as strengthen-
ing their significance. This significance is basically symbolic in that the
physical behaviour of the participants 'stands for1 relations of another
kind, as between man and an aspect of Nature or man and putative
spirits. Many rituals are performed to maintain an existing situation from
degeneration; others to change the situation, if only to restore it to an
original state of well-being. Ceremonial (ceremony) I regard as a species
of ritual in which, however, the emphasis is more upon symbolic ,
acknowledgement and demonstration of a social situation than upon the
efficacy of the procedures in modifying that situation. Whereas other
ritual procedures are believed to have a validity of their own, ceremonial
procedures, while formal in character, are not believed in themselves to
sustain the situation or effect a change in it.
(pages I2—13)

In the ancient Indian Vedic ritual several rites and recitations

occur that are to some extent similar to the Tikopia ceremonies.
When the Agni fire is to be generated by rubbing two pieces of
wood together, the Adhvaryu, chief priest of the Yajurveda,
says to the Hotâ, chief priest of the Rigveda:
 Anthropology Without Asia 147

Address Agni who is to be churned!

The H o t ä now recites three verses from the Rigveda, the first
one three times:

We implore you for our share, god Savitr,

owner of all that is worthwhile, always assisting!
(Rigveda 1.24.3)
Great heaven and earth must mix this ritual for us;
assist us with their support!
(RV 1.22.13)
Atharvan churched you, Agni, from the lotus;
priests from the head of the universe.
(RV 6.16.13)

When fire flares up, the Adhvaryu speaks to the Hotä again:

Address Agni who is born!

The H o t ä recites:

And people will say: Agni, killer of the demon, has arisen,
who in each fight wins booty.
(RV 1.74.3)

When fire is about to be installed on the altar, the Adhvaryu says

to the Hotä:

Address Agni who is thrust forward!

and the H o t ä recites:

Agni whom they carry like a ring on their hand,

l like a new born baby; who performs successfully
-' rites for the clans!
(RV 6.16.40)

Next the Adhvaryu says to the Hotä:

Address Agni who is kindled!

148 Ritual

and the H o t ä recites a number of verses while the Adhvaryu puts

sticks of firewood on the fire (see Chapter 11 A, A G N I 1:307—
311). The number of verses and sticks of firewood depends on
the ritual in which this rite is embedded.
On a later occasion, when fire is about to be transported from
the first altar to another, placed further east, the Adhvaryu
addresses the Hotä as well as a second Hotä, and all, including
the Adhvaryu himself, accompany the fire on its journey east
with long recitations from Rigveda and Yajurveda (AGNI 1:551-
555).
In describing these Vedic rites and recitations I have not
specified contexts in the manner in which this was done by Firth
with regard to the Tikopia rites. I have not specified, for
example, whether these rites are connected with new born
babies, initiations, marriages, funerals, or other events. The
reason for this omission is that rites such as these may be
inserted or embedded in numerous ritual contexts. This corrob-
orates van G e n n e p ' s and Durkheim's observations referred to
before, and is a significant circumstance to which I shall return.
This circumstance is also related to the question whether these
Vedic formulas may be regarded as "free formulas" in Firth's
sense. These are not, though others are. However, this is a
matter too complex to be discussed in the present context (cf.
Renou, 1954, s.v. aha; Gonda 1977:565-567; A G N I , 1:310-311,
532-533, 683-685, etc.).
I have described these Vedic rites in some detail (omitting
many further details, for example those that pertain to the
technique of recitation) not because they are so similar to the
Tikopia ceremonies recorded by Firth, but in order to illustrate
the difference in kind between the data we are dealing with. The
anthropologist records what he finds on his island or in his
village or tribe; with luck, he is in a position to revisit the area of
his earlier fieldwork, and find out whether there have been
changes. "Anthropology is basically the process of trying to get
a story out of a snapshot, and it usually doesn't w o r k " (Smith
1987:207). The student of Vedic ritual, on the other hand, deals
with a very extensive ancient literature on ritual, which includes
 Anthropology Without Asia - 149

a complete inventory of all the recitations and chants accompa-

nying rites that have been performed more or less continuously
in different parts of the Indian subcontinent for about three
millenia. The formulas I quoted with which the Adhvaryu
addresses the Hotâ are in fact about 2,700 years old, and the
recitations by the Hotâ are even older.
How can we be certain about such unbelievably remote dates?
Because of the structure and continuity of the Indian tradition.
These recitations and the rites that they accompany have been
the subject of a continuous tradition of commentaries and
subcommentaries from shortly after the time of their first com-
position to the present day. While the works of this tradition
have been largely committed to writing, there are also extensive
oral traditions that supplement them. The rites themselves
continue to be performed as a function of these oral traditions,
which have recently become the object of what has been called
" V e d i c fieldwork" (a term introduced by C. G. Kashikar 1968).
A comparison of the oral traditions with each other and with the
texts shows that many of the original compositions survive in
their pristine form. Cross-references, quotations, and inscrip-
tions, together with their historical, linguistic, philological and
literary analysis, enable scholars to check and countercheck the
authenticity of texts and traditions, establish their approximate
date and origin, and discover what changes they have under-
gone.
In India, the Vedic rituals have not only been described in
painstaking detail, and subjected to careful discussions and
commentaries by Indian scholars from about the eighth century
B.c.E., but all these ritual texts and their accessories have been
studied and made accessible by several generations of modern
Sanskritists from about the middle of the nineteenth century.
Albrecht (not Max) Weber, Hillebrandt, and Schwab were the
founders, and Caland was the grand master of this branch of
Sanskrit scholarship. A bibliography of works and publications
on Vedic ritual (including surveys, manuals, dictionaries, etc.)
by these savants and by contemporary scholars from Europe,
America, Japan, and India itself would constitute a sizable
150 Ritual

volume. The only social scientists who recognized the impor-

tance of some of this work were Hubert and Mauss, who
published in 1899 in the Année sociologique their article " E s s a i
sul la nature et la fonction du sacrifice." This essay, to which I
referred several times, over-emphasized animal sacrifice, for
which it relied on Schwab's monograph of 1886. It still provides
the best starting point from which sophisticated and possibly
adequate theories of ritual can be derived. Unfortunately, it has
not induced other anthropologists or sociologists to consult the
ongoing literature on Vedic ritual. Durkheim referred to Hubert
and M a u s s ' article, as we have seen, in the context of observa-
tions that he himself failed to interpret and that were generally
ignored elsewhere. The only scholar who has been inspired by
the " E s s a i " is the Indologist J. C. Heesterman, who introduced
a sociological perspective into the classical Indological studies
on ritual in his book The Ancient Indian Royal Consecration of
1957 and in numerous articles, some included in Heesterman
1985.
What is true of India is also true of China, though the ritual
tradition is not equally ancient and contemporary scholars are
only beginning to tap its resources (see, e.g., Schipper 1975).
There is a voluminous literature in Japanese on ritual, Buddhist
and Shinto, by ancient ritualists as well as contemporary social
scientists. In contrast to the veritable oceans of data and
evidence that the classical civilizations of Asia provide and that
modern scholars continue to make accessible, the data brought
back by contemporary anthropologists, interesting and varied as
they are, must strike any unbiased observer as meager. N o
wonder that the conclusions derived from these data are tenta-
tive and unsteady. Firth in his observations on ritual, quoted
above (p. 146), refers on the one hand to the symbolic interpre-
tations that reflect the traditional Western perspective rein-
forced by German philosophers, and recognizes on the other
hand that rites and rituals have a certain validity of their own,
without however explaining in what this validity consists or
where it resides. The cause for such wavering indecision, which
is not very different from the inconclusive views espoused by
 Anthropology Without Asia 151

Durkheim, Weber, and even Malinowski, lies to a large extent in

the limited and inadequate nature of the source data. The first
conclusion that may be inferred from this state of affairs is that
no theory of ritual can be taken seriously unless it comes to
terms with the data provided by the classical civilizations of
Asia, in particular India and China.
A similar situation as we still find in the social sciences existed
not long ago in Western linguistics. At first it was thought that
the structure of human language could be studied adequately by
collecting data from so-called primitive and exotic languages.
Wilhelm von Humboldt in Über die Verschiedenheit des
menschlichen Sprachbaues of 1836 used Javanese, Burmese,
Chinese, Sanskrit, and " D e l a w a r e " in this spirit, though he also
made some use of concepts that originated with the Sanskrit
grammarians. This is reminiscent of those early ethnologists who
adopted terms such as totem or mana from the cultures they had
studied, the only difference being that the terms of the Sanskrit
grammarians are technical terms with very precise meanings,
whereas totem and mana are hazy notions, The next step of
development of Western linguistics is represented by Whitney,
who provided in his grammar of Sanskrit a virtually complete
inventory of the language, but who still regarded Pânini, the
greatest Indian grammarian, as an object of Indological investi-
gation, at best an informant, and did not recognize that he was in
fact: a deceased colleague of genius. This was not merely an
oversight; it went hand in hand with profound misapprehensions
that Whitney entertained with regard to the nature of human
language (above page 50). The final stage was initiated by
Leonard Bloomfield and has culminated in the work of Paul
Kiparsky, in which contemporary linguistics comes to terms
with the insights and results of the Sanskrit grammarians.
In the West, linguistics is now considerably ahead of the study
of ritual, which has not even reached a state of development
where it can be regarded as a scientific discipline. If any progress
is to be made, these studies will first have to incorporate the
ritual evidence from India, China, and Japan. Next they must
take account of the Indian science of ritual (Chapter 26). Given
152 Ritual

further incentives and new creative insights, what has long been
a grab-bag of metaphysical and other speculations may then
finally turn into a science.

14B. Victor Turner

Some of the comments that were made on the work of

Raymond Firth apply to Victor Turner. He reflects on his study
of the Chihamba ritual in the following terms:

At one time I employed a method of analysis derived essentially from

Durkheim via A. R. Radcliffe-Brown. I considered the social function of
Chihamba with reference to the structural form of Ndembu society. But
this method did not enable me to handle the complexity, asymmetry, and
antinomy that characterize real social processes, of which ritual perfor-
mances may be said to constitute phases or stages. I found that ritual
action tended thereby to be reduced to a mere species of social action,
and the qualitative distinctions between religious and secular custom and
behaviour came to be obliterated. The ritual symbol, I found, had its own
formal principle. It could be no more reduced to, or explained by, any
particular category of secular behaviour or be regarded as the resultant
of many kinds of secular behaviour, than an amino-acid molecular chain
could be explained by the properties of the atoms interlinked by it. The
symbol, particularly the nuclear symbol, and also the plot of a ritual, had
somehow to be grasped in their specific essences. In other words, the
central approach to the problem of ritual has to be intuitive, although the
initial intuition may then be developed in a logical series of concepts.
(Turner 1975:186-187),

Turner, until his recent death the leading and most influential
American anthropologist to write on ritual, did not accept the
traditional doctrine that rites always have to re-enact myths. He
distinguishes, for example, in Ndembu ritual, rites that have a
mythological character from others that do not (1975:177). In the
passage quoted, he distinguishes two features of ritual: symbol
and plot. Both of these, he says, belong to a religious domain and
cannot be reduced to, or explained by, secular behaviour. When
he says that these features have to be grasped intuitively, before
they can be studied logically, he formulates the general charac-
teristic of all scholarly work, whether in the humanities, in the
 Anthropology Without Asia 153

sciences, or in the social sciences to which I referred in Chapter

2 and in the Introductory Note to Part II. The dichotomy
religious/secular (or sacred/profane) is another matter: a West-
ern dualistic fetish to which we need not pay serious attention.
F r o m the viewpoint of Vedic ritual, T u r n e r ' s approach is a
step forward from any approach which advocates the derivation
of ritual from mythology or social structure. Moreover, Vedic
ritual can be described as a " p l o t " , or perhaps more appropri-
ately, as an activity. It is not so easy to see why it should be a
symbol. I suspect that this is another category which is imposed
upon it from the outside. The main reason for this suspicion is
that it is not easy to find anything that Vedic ritual could be a
symbol of. So let us look a little more closely.
Earlier, Turner had distinguished three orders of reference for
a symbol's use: "(1) its manifest sense(s), of which the subject is
fully conscious and which is (are) related to the explicit aims of
the ritual (to remove sterility, bring on rain, remove a dead
hunter's incisor tooth from the body of his living kinsman, and
so on); (2) its latent sense(s), of which the subject is only
marginally aware but could have become fully aware and which
is (are) related to other ritual and pragmatic contexts of social
action; and (3) its hidden sense(s), of which the subject is
completely unconscious and which is (are) related to infantile
(and possibly pre-natal) experiences shared with most other
members of his society, and perhaps with most other human
b e i n g s " (Turner 1975:176).
As for the third kind of reference, Turner considers that it is
outside the scope of anthropology: " I t is best left to the various
and by no means anonymous schools of depth psychologists".
As for the first, it is easily disposed of in the case of the
Nambudiri Agnicayana: this Vedic performance is typically
self-contained and self-absorbed. There are no symbolic mean-
ings going through their minds when they are engaged in
performing the rites (above, page 115).
There are within the Agnicayana single rites which have a
symbolic function of which the performers are aware. For
example the Yajamâna or ritual patron keeps his fists closed
154 Ritual

because he is like a child in the womb of its mother, ready to be

reborn. Others only appear to have such a function. For exam-
ple, among the thousand bricks, ten are called ''rain-bringing"
bricks (above, page 116). Their name is sometimes recognized,
but the ceremonies affecting them are not performed when rain
is needed; they are performed as part of the Agnicayana when
the Agnicayana is performed. Many bricks have names, and
some of these evoke associations; but the majority of them, and
the accompanying rites and recitations, lack any clear function
or meaning. We shall return to this in a different context in
chapter 24.
So we are left with the second possibility mentioned by
Turner: the latent sense of a symbol. I have nothing against
unobservables (Chapter 2) but the burden of proof rests on the
shoulders of the person who makes such a claim. A good
example to which I shall return (page 453) was provided by
Heesterman (1962:18) who drew attention to the peculiar man-
ner in which the victim in the Vedic sacrifice is bound to the
stake: the cord is fastened to the right foot, goes round to the left
side of the neck and is then wound round the right horn and
finally fastened to the stake. Heesterman pointed out that thus
room is left for the slaughterer's knife. Since the animal is never
killed in this manner, this suggests that it formerly was. One can
easily imagine that the performers could be said to be marginally
aware of this, or that they would immediately recognize it as true
if it were pointed out to them.
Such cases, however, are exceptional. The large majority of
rites of the Agnicayana point in a different direction: they are
performed for their own sake, and are not symbolic or reminis-
cent of anything else. Theories which reduce rites to something
else are therefore not applicable to this ritual. Even if they are
correct in historical terms, namely point to the historical origin
of a rite, this origin has nothing to do with the ritual significance
of the act. For example, the movement from west to east which
characterizes certain ritual activities is connected with the
eastward expansion of the Vedic Indians through the plains
north of the Ganges. But the ritual is not a celebration of this:
 Anthropology Without Asia 155

that background has long disappeared. It is as with the etymo-

logical explanation of a word: in most cases, this has nothing to
do with its function or meaning. It falls entirely within the
scholar's domain.
In India this is well-known: usage prevails over etymology
(yogäd rûdhir bally ah) (page 23). The. Mïmamsa philosophers
apply a similar argument to ritual: the ritual is to be performed
for its own sake. Ritual is not for the sake of gods, but gods are
for the sake of ritual. Ritual is not for wish fulfilment, since rites
have to be performed by those who are eligible. I suspect that
the stress on symbol and meaning is largely due to the preoccu-
pations of scholars. It is they who connect rites with myths and
with sense. It is they who want to understand, namely, explain
in words and language what is basically activity. The irreduc-
ibility of ritual shows that action constitutes a category in its own
right. To the Vedic ritualist, it is the action which counts, not
verbal elucidation or interpretation. Of course, in India there
have also been scholars, such as the authors of the Brähmanas.
But practising ritualists pay no attention to their theories.
It is possible that there is a further reason for this state of
affairs that is historical. Western religions stress "right opin-
i o n " , or orthodoxy. Indian religions stress "right a c t i o n " , or
orthopraxy. To Western scholars, Indian religions only make
sense if they express ideas. To Indian practitioners, ideas are
optional. A practising Hindu ritualist can be a democrat, a
communist, an atheist, a magician, or a rationalist. What counts
and makes him a Hindu is primarily what he does. This topic,
too, has ramifications to which we return in Chapter 28.
 15

Syntax, Semantics and Performatives

The argument to show that Pravargya and Upasad form a

single unit within which the two parts occur in the given order
(Chapter 10) paid no attention to meaning and was a purely
syntactic argument. Many of the observations made in the
previous Chapters amount to saying that in the study of ritual, a
syntactic approach is more fruitful and promising than the
semantic approach that is customary. We shall expand this
observation later, but it is high time to form a more precise
conception of the distinction between syntax and semantics
before we go any further and contemplate ramifications. The
definitions are simple: syntax is concerned with relations be-
tween (logical, mathematical, linguistic, or other) expressions;
and semantics is concerned with the relations between such
expressions and their meanings.
The clearest examples of syntax and semantics are found in
elementary mathematics. One difficulty that schoolchildren of-
ten experience when they are introduced to algebra is that its
letter symbols are devoid of specific meaning. They are
" a b s t r a c t . " This applies not only to the „v, which stands for the
unknown answer of a problem, and to other so-called "vari-
a b l e s , " but also to " c o n s t a n t s , " such as a and /?, in terms of
which the problems themselves are formulated. For example,
the expression:

(a + b) 2 - a 2 + 2 • a . b + b 2 (23)

applies to many values and interpretations given to a and /?, and

to operators other than " = , " " + , " and fcfc • " . The constants
may be interpreted as integers, e.g.:
158 Ritual

(2 4- 3) 2 = 2 2 + 2 • 2 . 3 + 3 2
or also:
(29 + 37) 2 - 29 2 + 2 . 29 • 37 + 37 2 ;
but the expression (23) also applies to real numbers, imaginary
numbers, or even geometrical configurations, as in:

a-b b2

a2 a-b

In this figure, the total area of the square is (a + b) 2 , which

consists, as mere visual inspection shows, of four areas: a 2 , b 2 ,
and twice a • b . Such wide and flexible applications of the
symbols of algebra are characteristic of all of mathematics, and
explain why it is at the same time abstract and powerful.
The importance of the syntactic approach is that it studies
expressions such as (23) without paying attention to any of the
interpretations or meanings that might be assigned to them. The
identity or equality that (23) expresses applies to all readings
assigned t o k ' < z " or " / ? " without distinction. So we are in a
position to do many things at the same time, and more things
than we can possibly imagine, by restricting ourselves to the
syntactic manipulation of such expressions. Moreover, we will
not be swayed or influenced by particular interpretations and
arrive at slanted or biased views that have nothing to do with
intrinsic structure.
I shall now illustrate the semantic and syntactic approaches
with respect to an even simpler symbolic expression:
A B C B A. (24)
 Syntax, Semantics and Performatives 159

In the semantic approach, the symbols A, B, and C are inter-

preted as meaning something or referring to something. In the
syntactic approach we pay no attention to such possible mean-
ings or interpretations, but study the configurations of the letter
symbols only.
Adopting first the semantic approach, (24) may be understood
in many different ways, for example:

I. referring to numbers, e.g.: 3 8 7 8 3.

II. referring to segments, e.g.:

III. referring to algebraic symbols, e.g.: (a + a).

IV. referring to words, e.g.: found sleepy on sleepy found.
V. referring to tones, e.g.:

VI. referring to movements, e.g.: three steps up a staircase

and then down again; or: making a pirouette, jumping
up, kneeling, jumping up, making a pirouette again.
VII. referring to activities, e.g.: unlocking a box, opening it,
putting something inside, closing it, locking it.
VIII. referring to rites, e.g.: making fire, pouring Soma,
chanting a hymn, pouring Soma, making fire.

There is literally no end to the possible interpretations of (24).

Adopting a syntactic approach, there are also indefinitely
many possibilities. For example, (24) is part of:

I. M N A B C B A N M, but also of:

IL M N A B C B A M N.
III. it is a mirror image of itself.
IV. it consists of five units, just as: P Q R S T.
160 Ritual

V. it has specific instances, such as:

A B B C C C B B A , but also:
VI. A B C B A - B C B - C - B C B - A B C B A .
VII. it is interspersed through: P A Q B R C R B Q A P ,
but also through:
VIII. P A B Q R S C T B U V W A .

And so on. The expression (24) may be looked upon as a specific

instance of a more general expression, viz.:
A, . . . A n _ , A n A n _ , . . . A, (25)
where " n " is a natural number. In the case of (24), n = 3, for (24)
can be rewritten as:

It is relevant in the present context that the elementary units

of mathematical theories, and anything that can be stated in
terms of such units, may be of any size or kind: these structures
operate on many levels. Mathematics applies accordingly to the
inside of the atom as well as to the universe.
The semantic approach often seems richer than the syntactic,
but that is based upon the idea that syntax studies nothing but
sequences of symbols that follow each other in linear order.
Chapter 12 made it clear that syntacticians are not behaviorists
who are only concerned with sequences of events. In the simple
sentence:
I gave Peter his book back,
the underlying syntactic structure explains that " h i s b o o k " is a
unit and " b o o k b a c k " is not. This may be expressed by " t r e e s "
or by simple parentheses, e.g.:
I gave Peter (his book) back.
We have studied the syntactic structures of ritual from Chapter
8 onward and have not yet paid any systematic attention to
semantics. We shall do so in Chapter 25, but will first approach
semantics through one of its sub-disciplines: the study of per-
 Syntax, Semantics and Performatives 1.61.

formatives. The first scholar who explored the idea that perfor-
matives may be relevant to the study of ritual was the anthro-
pologist S. J. Tambiah in his Radcliffe-Brown Lecture of 1979
entitled "A Performative Approach to Ritual/'
Tambiah's lecture is rich in facts and ideas, and I cannot do it
full justice in the present context. Moreover, it develops results
from an earlier study, Tambiah 1968b, and the reader is urged to
consult both. Tambiah's 1979 lecture also discusses ideas from
other disciplines that may fruitfully contribute to the anthropo-
logical study of ritual. Unfortunately, this richness is sometimes
misleading and some concepts are probably incompatible with
each other. For example, Tambiah's emphasis on rule-governed
activity in which the rules are constitutive contradicts his
continuing emphasis on "symbolic communication"—the pre-
vailing anthropological doctrine that is probably wrong on both
counts (for ritual is neither symbolic nor communication). Let
me try to disentangle some of these ideas.
When Tambiah refers to Austin's performatives he invokes
Searle's distinction between regulative and constitutive. Regu-
lative rules regulate a pre-existing activity, an activity whose
existence is logically independent of the rules, as for instance
when dinner-table manners regulate the eating of food. Consti-
tutive rules constitute (and also regulate) an activity the exist-
ence of which is logically dependent on the rules, as in the rules
of football or chess (Searle 1969:34-35). One of Tambiah's chief
theses is that many rites are constitutive acts:

To Austin's classical examples of constitutive acts such as greeting,

baptizing, naming a ship and marriage vows—all of which are created
and understood within the bounds of the conventions themselves—to
these we can add several anthropological examples: the installation of a
Tallensi chief, Ndembu circumcision rites, Lodagaa mortuary rites, a
Japanese tea ceremony, a Catholic mass, and a multitude of cosmic rites
and festivals which are self-constituting events and of which we have
several classic descriptions (Tambiah 1979:128).

Tambiah derives interesting conclusions from this important

insight but these are somewhat obscured because he also inserts
162 Ritual

a long section entitled "Application of Information Theory to

Ritual" which would be relevant if we were dealing with the
transmission of messages between a sender and a receiver.
Tambiah uses the latter idiom when criticizing Egan's contrary
contention that the Sinhalese edura healer " c a n be compared to
an individual stranded on a desert island with only a radio
transmitter, but no receiver" (Tambiah: 150). The idea is,
apparently, that healing and ritual are communicative acts in
which there are transmitters and receivers between whom there
is communication of information. I am in no position to comment
on Egan's comments on healing, but such criticism, whatever its
force, does not throw light on ritual, especially if we have agreed
that ritual consists of rules that constitute it, that is, on whose
forms it logically depends. Tambiah alludes to this insight later,
for example, when he observes that the Thai tonsure rites,
whenever and for whomever performed, "followed certain
obligatory s e q u e n c e s " (Tambiah: 157). But in contexts where
communication or information theories are invoked (as in
Geertz: see Chapter 25C), one feels that Tambiah has not fully
exploited the implications of his insights which may incidentally
explain why he has not made use of them in some of his later
work (Chapter 25A). The inapplicability of communication and
information theories to the study of ritual follows a fortiori from
the fact that such theories have been singularly unhelpful even in
explaining language, where meaning, communication and infor-
mation play a larger and more obvious role than they do in ritual.
The anthropological examples Tambiah added to Austin's are
constitutive but they are not performatives for the simple reason
that they are not utterances of language. Austin dealt only with
those: he was concerned with linguistic utterances that not only
say but also do something, or, more precisely: that in saying, do
something (above, page 22). Austin drew attention to the fact
that some utterances of language are acts. Searle's distinction
between regulative and constitutive has a wider application: it
applies to rules and rule-governed activities, language as well as
ritual. But to say that rites are performative, i.e., that they do
something, is nothing special or remarkable: they are acts to
 Syntax, Semantics and Performatives 163

begin with. To call an utterance performative is interesting

because it calls attention to a special class of utterances; to call
an act performative is trivial because most acts are performative.
To say that a rite is constitutive means that its existence and
meaning depend on rules. If the rite only depends on rules (the
condition implicit in Searle's use of the term) it cannot also be
explained because of its alleged symbolic or semantic reference
or value; it may possess such features, but they cannot, in that
case, be constitutive; they must be added. This is remarkably
often the case, as we have seen: all the symbolic explanations of
Vedic ritual in the Brâhmana literature are ad hoc, arbitrary or
rationalizations, i.e., after-thoughts. If this holds true of the rites
Tambiah refers to, their existing explanations in semantic terms
(for example, Turner's explanations of Ndembu rites) must be
wrong. Tambiah has not drawn this conclusion for otherwise he
would have been led to the analysis in syntactic terms that I am
advocating. In that case he would have discovered that syntactic
structures can be explained by postulating rules, some of which
are constitutive of the rites themselves.
I have mentioned that I cannot do justice to Tambiah's
important study; the above remarks should be taken with a grain
of salt. They mainly intend to demonstrate that Tambiah, by
advocating a rule-oriented approach, somewhat misleadingly
called performative, went significantly beyond the semantic
approach that is characteristic of almost all anthropological
work on ritual. Lévi-Strauss in his " s t r u c t u r a l i s m " took a
similar step but it was still inextricably connected with seman-
tics.
By walking with a more vigorous stride than he himself
realized, Tambiah showed that explanations in terms of meaning
are inadequate. He thereby paved the way for a syntactic
approach that may ultimately lead to the scientific study of
ritual.
 16

Music and Ritual

When looked at from a syntactic perspective there are numer-

ous similarities between music and ritual. I shall discuss seven
examples from Music in 16A, and seven examples from ritual in
16B.

16A. Seven Musical Structures

The examples of musical structures are taken from Western

classical music but include some references to classical Indian
music. The units range over a wide area, from single tones,
motives and phrases ("approximating to what one could sing in
a single b r e a t h " : Schoenberg 1970:3) to Sonatas and Sympho-
nies.

I. R E F R A I N . In poetry as well as music, it is common to end

each of a sequence of structures with the same form, as in:
A R - B R - C R - D R - . . . (26)
Music differs from poetry in that the refrain frequently occurs at
the beginning as well as at the end. In Gregorian chant, the
variable elements A, B, C, . . . are often sung as a solo, and the
refrain R by a chorus. In church music, the refrain may be Amen
or Alleluia, the latter closing with the jubilas, a long vocalization
on the final —a. Medieval and Renaissance music is known for
its formes fixes such as rondeau, ballatta, and virelai. In the
virelai, the refrain occurs at the beginning and serves each time
as the opening of the next unit, as in:

RARBRCR. (27)
166 Ritual

In all these forms, the refrains are each time exactly the same.
After the Renaissance, refrains tend to be varied in many ways,
making them musically more interesting. I shall express a
variation of a form 4 t R " as: " R " \ In the following example, the
second Nocturne for Piano by Erik Satie, the refrain consists of

NOCTURNE
(DKUXIKMK)

POUR PIANO Erik SATIE

 Music and Ritual 167

Plus Int! J _/«o

'I' J J J J U J j ;

Ralentir dr plus on plus

four measures. It occurs at the beginning, and recurs twice in a

varied form (where the score says: " R e p r e n d r e " ) :
The structure is:

R A R ' B R". (28)

168 Ritual

II. C Y C L E . I shall use this term, which is very general, to

refer to a structural property that is equally general, viz., one
that begins and ends with the same element, as:
A B C D . . . A. (29)
In musical phrases, it is extremely common to begin and end
with the same note, e.g., in Mozart's Nozze di Figaro, N o . 3
C c Se vuol ballare"):
Allegretto
Figaro
m i r r r pr ir r
S© vuol bal - la - re, 3i-gnor Con - ti - no,

Larger compositions often begin and end in the same key. For
compositions to begin and end with exactly the same phrase is
actually rather rare: the Harvard Dictionary of Music refers to
two seventeenth century examples, from Giovanni Maria Tra-
baci and Steffano Landi (Apel 1970:218a referring to Davison
and Apel, 1974, N o s . 191 and 208).
It is nearly universal, on the other hand, for musical compo-
sitions to return to their point of departure at the end by means
of a variation of the opening theme, viz., adopting a structure of
the form:
A B C D . . . A'. (30)
A very well known example is the first movement of Beetho-
ven's Symphony N o . 5, which begins as follows:
Alleçro con brio.

and ends:
 Music and Ritual 169

A particularly important simple form of (30) is:

ABA'. (31)

This is the basic form of many types of composition, including

the Sonata and the Symphony to which I shall return. The
musical term for " A " ' in such cases is recapitulation or restate-
ment.
III. P A L Y N D R O M E . 1 shall use this term for a special
type of cycle, also referred to as retrograde or mirror. It is a
musical illustration of the structure discussed in the previous
section:

A B C B A, (24)
or its generalization:

A, . . . A n _ ! A n A n _. t . . . A,.. (25)

The simplest example is ascending and descending a scale (or the

other way round). In a musical composition there is generally
some kind of variation. In classical Indian music, when the scale
is introduced by an ascent (ärohana) followed by a descent
{avaroharia), the latter often uses different notes and is therefore
not a mirror of the former. An identical retrograde movement
occurs in Debussy's Prelude N o . 10 ( ' T h e Engulfed Cathe-
dral"):
170 Ritual

Retrograde structures are used in most canonical, fugal, and

12-tone compositions. In the cancrizans (crab canon), the mel-
ody is accompanied by itself played backwards, as in the
following example from Bach's Musical Offering:

The palindrome occurs with variations in larger structures. The

structure of Fugue N o . 16 in Book I of Bach's Wohltemperiertes
Clavier can be expressed by the following scheme of expositions
(referred to by A v s) and episodes (referred to by B's and C's):
(cf. Green 1965:265-267)
A B A C - A' B ' A" - C' A'" B" A"" (32)
Bach's motet Jesu, meine Freude consists of eleven parts: the
music of N o . 11 (a 4-part Chorale) is identical with that of N o . 1,
and N o . 10 (a 5-part Chorus) is a shortened version of N o . 2
(Steinitz 1978, 34), so that the structure is:
 Music and Ritual 171

AB B ' A. (33)
A similar pattern, ranging over longer stretches and closer to a
full-fledged palindrome, occurs in Bach's Johannes Passion
(Steinitz, ibid.). If we omit the Evangelist recitatives from the
section consisting of parts Nos. 27 through 52, and regard Nos.
31 and 32 as parallel interpolations with No. 48, the structure
may be described as follows (the numbers of the parts are
written below the corresponding structural elements):

A B C D E D' C' B' A'

27 29, 31-2, 34 36 38 40 42 44 46, 48, 50 52. (34)

IV. O V E R L A P P I N G . This feature of structure applies only to

simultaneous occurrences, i.e., two (or more) tones, motives,
phrases or melodies played or sung at the same time. It cannot
therefore be represented by linear sequences of symbols, such as
I have used so far. Here is a simple example of overlapping
motives from Mozart's Sonata for Violin and Piano, K.402:

Allegro moderato

Overlapping occurs frequently in counterpoint and polyphonic

music, especially in fugues, canons, etc.
V. T H R E E S O M E S . Triple repetitions are common on many
levels, from simple triplets to the threesomes that were universal
in church music because of their symbolism (viz., the trinity).
For compositions to begin or end with a triple repetition of a
motive or theme is very common in Indian music and occurs,
e.g., in Mozart's Sonata for Violin and Piano, K.377-II:
172 Ritual

" tU i Ü J U i Ij

or in Schubert's String Quartet, Op. 29-III:

Menvetio

Throughout classical western music, the chief function of the

number t u t h r e e " is that it is the most common number of
subdivisions of many larger units. It is characteristic of the two
following structures, Sonata and Symphony.
VI. S O N A T A . This form is characteristic of classical Western
music since Haydn and Mozart. It has the following structure:
A B A', (31)
which originated from the simpler form:
ABA, (35)
which is found in the Rondo and Minuet. Often the elements
themselves are again of the same form. In the paradigm, A is.
called the exposition, B the development (German: Durch-
führung which is better translated as " e l a b o r a t i o n " : Schoenberg
1970:200, note), and A' the recapitulation. A and A' are usually
in the same key, while B uses various keys and therefore
modulation.
The prehistory and history of the Sonata form exhibit extraor-
dinary variety. Sometimes the structure is influenced by the use
of various instruments, or varying numbers of players. At first
the parts were numerous and brief, and followed each other
 Music and Ritual 173

without a break. Later they became fewer, longer, and more

separate from each other. The solo parts were played by violins,
flutes, oboes, etc.; the accompaniment was often on the harpsi-
chord. Before the eighteenth century, the principle was simple:
if someone was around who was able to play the part on his
instrument, he was welcome to do it.
VII. S Y M P H O N Y . The symphony started tripartite, in the
Sonata form, with more instruments and performers: a Sonata
for orchestra. Soon a fourth part was added, chiefly for enter-
tainment: the Scherzo. The two middle movements, Andante
and Scherzo, are frequently of the form (35), whereas the first
and the last movement are more often of the form (31). The form
was perfected by Haydn and Mozart, and enlarged and extended
by Beethoven. Beethoven lengthened the coda, a final piece
following A ' , and the development, introduced numerous vari-
ations and added introductions to the various pieces. The
increasing complexity of symphonies after Beethoven has from
time to time been counterbalanced by simplification of the
thematic structure. In Berlioz' first two symphonies, for exam-
ple, a single theme (the " i d é e fixe1') appears in each movement.
The opening theme of the first movement returns in the final
movement in B r a h m s ' Third Symphony (the same return is found
in his Clarinet Quintet). The finale of Bruckner's Eighth Sym-
phony includes a simultaneous statement of the principal themes
of the three preceding movements (Green 1965:176). (Some
recurrences are incidental, or have the character of reminis-
cences or quotations, as in Beethoven's Fifth and Ninth Sym-
phonies, and in his Piano Sonata, Op. 101.)
To express these structures adequately more complex math-
ematical expressions than the linear are required (for other
formalizations see Ruwet 1966, reprinted in Ruwet 1972:100—
134). Although the expression ABA (35) is linear, the expression
A B A ' (31) is only apparently so: it has an inner or underlying
structure that is not linear. The variation A ' depends on the
original statement A, and therefore on the context. However, it
does not depend on the immediately preceding context, in which
 174 Ritual

case it would have been possible to express it with the help of a

context-sensitive rule of the form:

BA->BA'. (36)

The variation depends on a context that is further away, which

can be expressed by using transformational rules—the kind of
rules that are characteristic of human language (Chapter 6). With
the help of such rules, the variation A' can be introduced by rule
structures like the following:

(37)

This expresses at the same time that the variation is a variation

of the first movement, and that it occurs in the finale.
I have purposely provided a medley of musical forms in order
to illustrate the variety of musical structures. Whenever such a
structure is given, it is possible to define a notion of "structural
m e a n i n g " that depends entirely on structure. The "structural
m e a n i n g " of the entire structure is that structure itself, and the
"structural m e a n i n g " of an element within the structure is
defined in terms of the position the element occupies in that
structure. For example, part of the structural meaning of a
phrase may be that it is a refrain because it occurs as a refrain.
The notion of structural meaning adds nothing to the existing
structures themselves.
Another type of structural meaning that is typical of classical
Western music may be defined in terms of the notions of
consonant and dissonant intervals. Consonant intervals are
'intervals such as the fourth and fifth that " s o u n d stable and
c o m p l e t e " (Piston 1962:6). Dissonant intervals deviate from
these and require a resolution into consonant intervals. Such
resolutions operate on the harmonic as well as on the melodic
level. The musicologist Schenker and the composer Hindemith
 Music and Ritual 175

have used the terms " t e n s i o n " and " r e l e a s e " to characterize
this fundamental opposition between the formation of disso-
nance and its resolution into consonance.
The resolution is only rarely introduced by a single move. It is
more often reached through intermediate steps that introduce
new dissonances and resolutions. An example occurs at the end
of the first Prelude of Das Wohltemperierte Clavier:

Consonance and dissonance are to some extent relative con-

cepts: their perception changes with history because ears get
used to dissonances. This was pointed out by the devil to the
composer Adrian Leverkühn in Thomas M a n n ' s Dr. Faustus
(English translation, 1949:239):

The diminished seventh is right and full of expression at the beginning of

Op. 111. It corresponds to Beethoven's whole technical niveau, doesn't
it?—the tension between consonance and the harshest dissonance
known to him. The principle of tonality and its dynamics lend to the
chord its specific weight. It has lost it—by a historical process which
nobody reverses. . . . Every sound carries the whole, carries also the
entire history in itself.

Notes possess structural meaning when their occurrence can

be explained because it contributes to the formation of disso-
nances or their resolution into consonances. Since some disso-
nant intervals have different resolutions, notes may have multi-
ple meaning. In other words, there is functional ambiguity. The
diminished seventh cord, for example, may be interpreted in
four different ways, each expecting a different resolution. The
four interpretations can be distinguished by sounding different
notes (in the left hand) against the chord (in the right hand) in the
following example constructed by Piston (1962:193):
S 76
Ritual

Wè

Philosophers and musicologists have not been satisfied with

structural meaning alone, even with functional ambiguity added.
They have mused and written extensively on the meaning of
music in general. In simplified terms, there can be said to be two
schools. According to the first, music is allegorical and ex-
presses nonrnusical elements, e.g., features of the universe or of
man, such as emotions, passions, or moods. According to the
second, music is to be understood on its own terms. This
somewhat academic distinction corresponds to a distinction
made by some composers, musicians and music critics between
"program m u s i c " and "'absolute m u s i c . " Program music is
inspired by a nonmusical program, for example, a poem or
picture, usually indicated in its title (like Debussy's "Engulfed
Cathedral" actually placed at the end of the music between
parentheses.). Absolute music is not characterized in nonmu-
sical terms. The Harvard Dictionary (s.v. " p r o g r a m m u s i c " )
elaborates instructively:

Although examples of program music are found in nearly all periods from
at least the 14th century, it was not until the 19th century that it became
a serious rival of absolute music, to the point of ousting the latter—at
least temporarily—from its dominating position. About 1900, many
persons, particularly writers on music, believed that in order to be
understandable music must 'express something' or 'tell a story.' . . .
Today such views are a thing of the past . . ." (Apel 1970:696a).

Sometimes music actually imitates the sounds of nature or

culture. Schoenberg (1970:95-97) quotes examples from Beetho-
ven and Smetana (the sound of a brook or river), Wagner
(flickering flames), Schubert (the weathervane) and Bach (the
tearing of the temple veil). The Harvard Dictionary refers to
 Music and Ritual 111

Beethoven's renderings of the cries of the nightingale, cuckoo

and quail, Wagner's imitations of the toad and serpent, and
Richard Strauss' of a flock of sheep. "The climax of this trend
(and, in a sense, the reductio ad absurdum of program music) is
Respighi's The Pines of Rome, where the problem of imitating
the nightingale is solved by simply using a recording of an actual
nightingale's song" (Apel 1970:697a).
It is obvious that all such imitations are anecdotal and have
little or nothing to do with "the meaning of music." Musical
compositions may be inspired or occasioned by nonmusical
things such as a cathedral, but no music conveys the meaning
cathedral in the systematic manner in which the word "ca-
thedral" of a natural language such as English refers to a
cathedral. What such titles indicate is that music may be
composed or played on almost any occasion. These titles are*
mere names, and therefore arbitrary. Architectural monuments
may also be referred to by names (e.g., "The Parthenon," "The
White House"), but such names convey nç architectural mean-
ing. Similarly, Sonatas may be referred to by names such as
"Spring," "Waldstein," "Moonlight," or "Appasionata," but
these convey no musical meaning. They give radio announcers
something to talk about for the benefit of listeners in need of
nonmusical entertainment. If the expression "the meaning of
music" must be used, it can only refer to the structure of music,
that is, to syntactic structures like the ones I have exemplified
under the headings I-VII, and expressed by formulas such as
(24H37).
What is characteristic of music, viz., formal structures, is not
equally characteristic of other arts. Schopenhauer expressed
this insight in the following terms:

Die Musik ist . . , darin von allen andern Künsten verschieden, dass
sie nicht Abbild der Erscheinung, oder richtiger, der adäquaten Objek-
tivität des Willens, sondern unmittelbar Abbild des Willens selbst ist und
also zum allen physischen der Welt das Metaphysische, zu aller Erschei-
nung das Ding an sich darstellt ("Music differs from all the other arts in
that it is not representation of appearance, or rather, of the adequate
objectivity of the Will, but a direct representation of the Will itself, and
178 Ritual

therefore manifests the metaphysical in all that is physical in the world,

and the thing-in-itself in all appearance").

The idea is buried in metaphysical notions such as K a n t ' s

"thing-in-itself' and Schopenhauer's i 4 Will". However, des
Pudels Kern is straightforward: "music differs from all other arts
in that it represents no appearance of anything else, but only
itself." This passage occurs in Schopenhauer's Welt als Wille
und Vorstellung of 1818 and is quoted by Nietzsche in Die
Geburt der Tragödie aus dem Geiste der Musik of 1872. Scho-
penhauer clarified his view later by adapting traditional scholas-
tic terms: concepts are the universalia post rem, music provides
the universalia ante rem, and reality the universalia in re.

16B. Seven Ritual Structures

I shall again begin with seven examples described in syntactic

terms, this time of ritual structures. The fiçst five correspond to
the first five musical examples discussed in 16A. The latter two
structures are the Isti and the Soma Sequence (discussed in
Chapters 9A and 9B). Here the correspondence is more round-
about, but not less significant. The structural elements range
again over units of different size, from small elements of single
rites that take less than a minute to perform, to larger rituals that
take several days or more to execute.
Most of these ritual structures involve recitations or chants. In
the discussion of Tikopia ceremonies (Chapter 14A) we have had
occasion to observe that recitations are often an important
feature of ritual. To the extent that the ritual recitations and
chants exhibit musical features, the similarity between ritual and
music is unsurprising. It will be found, however, that even if we
disregard musical features of rituals, the similarities remain.
Those interested in the purely musical structures that are also
features of Vedic ritual should turn to the chants of the
Sämaveda. Their internal structure is taken into account else-
where (see Staal 1961:chapter VII; Howard 1977; below, Chap-
ters 19 and 23).
 Music and Ritual 179

I. R E F R A I N . There are several reasons for the frequency of

ritual refrains. The ritual acts often culminate in an offering or
oblation made by the Adhvaryu priest into the fire. These
offerings and oblations are of several kinds, but they naturally
come at the end of acts which exhibit great variety themselves.
The refrains consist of two parts: an exclamation by the Adh-
varyu (either svähäl or vausatl) and the " r e n u n c i a t i o n " (tyäga)
recited by the Yajamâna or patron of the ritual, for example:
' T h i s is for Agni, not for m e ! " (above, pages 80, 121). The
general structure is therefore of the form:
APQ - B P Q - C P Q - D P Q - . . . (26')
In long recitations from the Rigveda, such as the sastra
recitations to which I shall return, the verses are recited unin-
terruptedly, without taking breath at the end of each verse.
Breath is taken at the caesura between the second and third
quarter of each verse. At each caesura, the final vowel or nasal
is lengthened. Each verse ends in a lengthened O M—the famous
mantra of later Hindu mysticism.
The bird shaped altar that is characteristic of the Agnicayana
is constructed from a thousand bricks, each consecrated with
three mantras. Of these, the first is generally specific and
described for each brick or group of bricks, but the second and
third are always the same, so that the structure is again the same
as (26') (see A G N I 1:399).
The mantras recited before the final exclamations that accom-
pany offerings, have often a similar internal structure. Here are
two examples. After the altar has been completed, there are six
oblations of curds mixed with honey. The accompanying man-
tras differ in their first lines, but all end with:

Him whom we hate and who hates us

I piace in your jaws!

which is followed by: svâhâl (AGNI 1:104, 381, 544-545).

Not long after this rite, which takes place on the ninth day of
the Agnicayana performance, there is a long recitation called
camakam after ca me, " a n d for m e ! " recited by the Adhvaryu
180 Ritual

after each word (AGNI 1:563-574). This is immediately followed

by seven butter oblations, made through a copper pipe attached
to a small toy cart. Each mantra ends with the same refrain:

May he protect brahman, ksatram,

may they protect brahman, ksatram!

followed by: svähäl (AGNI I: 574-575). Both rites have the same
structure, expressed by (26').
II. C Y C L E . I shall use this term again for a structure of the
form (29) (above, page 168). This is relatively rare in Vedic
ritual, but an interesting variant occurs in the Agnicayana just
before the camakam recitation. Seven oblations of cooked rice
are made on potsherds for the Maruts, storm gods and compan-
ions of Indra. Seven recitations are involved that will be referred
to a A , , . . . , A 7 . These mantras are recited in low voice,
cyclically, and in overlapping pairs, the first mantra always
ending in the long vocalic insertion - o , and the oblation made
with the last, as follows:
A, -A2 - svähä!
A2 - svähä!
A3 -A4 — svähä!
A4 -A5 — svähä!
A5 -A6 - svähä!
A6 -A7 — svähä!
A7 -A, — svähä!
(AGNI 1:562-563), This kind of pairwise overlapping, but with-
out the cycle, is also found in the kramapätha mode of Veda
recitation, used in the traditional transmission of Rig- and
Yajurveda (see Staal 1961:24, 44; below, page 372). ,f
HI. P A L I N D R O M E . This is a very basic structure in Vedic
ritual. It ranges, generally with variations, over small as well as
large units. The Agnistoma, a ritual performance that lasts five
days, has such a structure, which is also found in the
Darsapürnamäsa, or "Full and New Moon C e r e m o n i e s . " Many
sastra recitals and stotra chants are embedded in similar port-
 Music and Ritual 181

manteau patterns, i shall not describe these structures in greater

detail, since I discussed them in Chapter 10.
IV. O V E R L A P P I N G . A variety of this structure was de-
scribed in the cycle discussed in II. The basic structure is very
common and reflects a general principle that is metaphysical as
well as structural: horror vacui. There should be no gaps;
accordingly, everything has to be performed continuously
{samtatam). The next rite is therefore begun before the previous
one is over, and the same holds for recitations, chants, and their
similar units (e.g., bhakti, to be chanted with one breath, from
which the chants are constructed). The practice extends to forms
of transmission of the ritual material. For example, some reci-
tations are " p r o m p t e d " by a helper, line for line, " b e f o r e " they
are recited by the ritually chosen and qualified officiating priest:
but this priest starts each time before the prompting is over. It
might be felt that this kind of " o v e r l a p p i n g " is a practical matter
and not the outcome of a structural principle. However, it is
widespread and based upon principles. I have included it pre-
cisely in order to show that we are not in a position to pass such
kinds of judgment when we are only beginning to subject the
data to a structural analysis.

V. T H R E E S O M E S . Triplets are everywhere. All longer reci-

tations from the Rigveda are subject to the rule: " R e c i t e the first
thrice, recite the last t h r i c e , " resulting in the structure:
Aj Aj Aj A 2 A 3 ... A n _ 2 A n _j A n A n A n . (38)
The Yajamâna and some of the priests frequently begin their
recitations with the so-called vyährti: BHÜR-BHUVAH-SVAR
( " E a r t h - S k y - H e a v e n ! " ) . Many larger units are subdivided into
three, though four is also met with. We have already found such
triplets in the Isti and the Soma Sequence. Each of the bricks of
the bird-shaped altar is also consecrated with three mantras. All
these are structural threesomes and are different from unstruc-
tured groups such as the three fires or altars, the three assistants
of each chief priest, the three stotra chanters, the three Vedas,
etc.
182 Ritual

VI. The isti has already been described (Chapter 9A).

VII. The Sorna Sequence has already been described (Chapter
9B).
Some of the variations of Soma sequences exhibit dependen-
cies that may be described as thematic. In many Soma se-
quences, the chant of the first episode is composed by putting to
music three verses from the Rgveda. The recitation of the
second episode begins with the same three verses, the first
repeated three times. In the nocturnal rounds of the Atirätra,
there is an additional refinement: in the first nocturnal round, the
first quarter verse of each of these three verses (that occur both
in the chant and in the recitation) is repeated; in the second
nocturnal round, the second quarter verse of each of these three
verses is repeated; and in the third nocturnal round, the third
quarter verse is similarly repeated (see AGNI 1:663-680).

16C* Conclusions

Now let us return to our old question: what do music or ritual

mean? I hope it has become clear by this time that this question,
with respect to the structures we have discussed, does not make
sense. These structures do not mean anything apart from and
beyond the structural complexity they display. They do not
" r e f e r " to specific aims, and although their names (insofar as
they have names) may seem to hint at their longforgotten origins,
these names do not evoke anything. The complexities inherent in
these structures have to be learned and practised, and can be
expressed with the help of precise rules—but they are not
symbolizations of anything else and do not point to a realm
beyond themselves.
In all these respects, there is considerable similarity between
the musical and ritual categories we have considered. The
similarities are not only structural, but also circumstantial. Just
like Sonatas and Symphonies, istis are performed on différent
occasions (providing ample evidence in support of van G e n n e p ' s
and Durkheim's observations, quoted in Chapter 13B) and have
 Music and Ritual 183

different names (like the intichiuma and mbatjalkatiuma rites

referred to by Durkheim). Even the fact that a Soma sequence
ends with drinking Soma is reminiscent of what may happen to
music. Smoking and drinking of coffee and beer was quite
common during musical performances throughout the nine-
teenth century. In Amsterdam it took strong conductors like
Willem Kes and Willem Mengelberg to get rid of the cups,
glasses and ashtrays, and introduce the musical puritanism that
assimilated concert performances further to church services.
To dissociate rituals from religious services and group them
together with music performances is only a small part of a
general reclassification of the ritual phenomena. This becomes
clear only at a level of abstraction sufficiently high to enable us
to abstract from, or disregard, certain dissimilarities in order to
detect certain similarities. Of course, there are dissimilarities as
well as similarities between Sonatas, Symphonies, istis, and
Soma sequences. But all four structures are also different from
many other things, e.g., poems, epics, laws, theories, stories,
cults, commercial transactions, educational projects, etc.—and
similar to many others, e.g., dances, games, and certain sports.
Ritual is more rigid than music although this is partly due to
the nature of our evidence, as we shall see in a moment. Western
music is more rigid than Indian music although some cadenzas in
concertos are improvised. A parallel occurs in the Soma rituals
where a lecture in the local language (Màlayalam), originally
largely improvised, is addressed to the Yajamäna after his
consecration by a senior member of the community (AGNI I:
329-333, 698-702). It is noteworthy that this address is con-
cerned with the meaning (artha) of a command (praisa), which
explains the use of the vernacular and underscores that the uses
of Vedic Sanskrit elsewhere in the ritual are devoid of meaning.
The structural similarities between music and ritual discussed
under the heading I through V are more than mere similarities:
they are identities. However, the structures discussed under VI
and VII are not sufficiently similar for one to say more than: Isti
and Soma Sequence are to some extent to Vedic ritual what
Sonata and Symphony are to classical Western music. Why then
184 Ritual

compare them with each other? The true significance of the

comparison is this. Sonata and Symphony exhibit structural
categories that characterize classical Western music, and do not
refer to anything nonmusical. Similarly, Isti and Soma Sequence
exhibit structural categories that characterize Vedic ritual, and
do not refer to anything nonritual. Not even the defenders of
program music have e^er claimed that the Sonata form or the
Symphony as such (as distinct from particular Sonatas and
Symphonies) exhibit nonmusical features of the world—unless
these are themselves syntactic or mathematical structures. So-
nata and Symphony are obvious features of ''absolute m u s i c . "
In the realm of ritual, Isti and Soma Sequences are equally
obvious features of ''absolute ritual." It has been assumed
without question that every ritual structure must have meaning;
in other words, that all ritual must be "program ritual." How-
ever, it should be obvious from the structures I have discussed
that these can only be described as features of what may be
called "absolute ritual." This is especially clear in the cases of
the Isti and the Soma sequence.
There are important differences between data on music from
roughly the second millenium C . E . in Europe and data on ritual
from roughly the second millenium B . C E . in India. The Euro-
pean data, for example, are much more complete in historical
terms. We can trace the development of classical Western music
step by step. The Indian data are relatively exhaustive only for
the period during which the classical ritual was codified, i.e.,
approximately from the 8th to the 6th century B . C . E . , when the
Srauta Sütras were composed. There is much less information
on the pre-classical ritual, though there are hints in the Rig veda
and scholars like Heesterman have been able to reconstruct
some of its general features. About the post-classical develop-
ment of Vedic ritual we are informed to some extent, but
unevenly. These later developments show mainly that the struc-
tures of the rites have not changed; but some ritual substances
have been abandoned (e.g., the original Soma; some of the
sacrificial animals; an occasional human head), and there have
been new flurries of interpretation, from time to time. All this
 Music and Ritual 185

supports the thesis, that the most important feature of ritual is

structure, and not substance or interpretation. Finally, the facts
that we. know Vedic ritual in a relatively codified form, and
Western music throughout a period of dynamic and often
revolutionary development explain in part that our ritual data
appear to be more rigid than our musical data.
The structural similarities between Indian ritual of the second
millenium B . C . E . and Western music of the second millenium
C E . are surprising, and certainly more interesting and significant
than would be such similarities found between Indian ritual and
music at different periods, or between European music and ritual
at different periods of time. The similarities we have discussed
are obviously not due to influences and are independent of
historical relationships. The relations are purely structural,
exhibit intrinsic similarities, and entitle us therefore to use the
kind of synchronie treatment that the syntactic approach pre-
supposes and demands. At the same time, it remains desirable to
compare these findings with other similar specific data on music
and ritual from other civilizations and periods of history.
It is possible that the high level of abstraction and formaliza-
tion that characterizes both Vedic ritual and classical Western
music is not widespread in terms of space or time. One thing
seems certain: there was no highly developed music unconnec-
ted with ritual in India, and no highly developed ritual uncon-
nected with music in Europe during the periods we have
considered. What we know of classical music in India during the
Vedic period is confined to the chants of the Sâmaveda, which
are closely related to ritual. In Europe, the example of J. S.
Bach is instructive. During 1717-1723, when he was employed at
Köthen at a Calvinist court where music and ritual were not
encouraged, he composed mainly secular music, adopting the
same styles he had earlier used and was to use subsequently,
when he spent much of his time composing music for the church.
All the evidence we have considered points in one direction:
ritual and music have no meaning or content, and can be
provided accordingly with any number of different meanings or
interpretations. The preponderant and enduring characteristics
186 Ritual

of music and ritual are that they consist of formal structures of

sounds and acts that can be studied most effectively and
fruitfully by adopting a syntactic approach. These sounds and
acts are like the letter symbols of mathematics in that they are
abstract and can therefore be interpreted in any way we like.
There is additional evidence from Vedic ritual that leads to the
same conclusion. The rites are traditionally transmitted from
each generation to the next with the help of methods and
techniques that pay exclusive attention only to their form.
Meaning is not mentioned, and interpretations are not given. Of
course, interpretations, or rather: speculations, exist, but they
are no part of the ritual tradition. Such interpretations are found
elsewhere, for example, in religious and philosophical texts. The
recitations and chants that accompany ritual are also handed
down without paying attention to their meaning (in as far as they
possess meaning: for many chants and some recitations consist
of meaningless syllables). Whereas all recitations and many of
the chants are handed down in schools or houses, independent of
their ritual applications and uses (see Staal 1961), ritual knowl-
edge is primarily transmitted during ritual performances and
their rehearsals. This applies to a limited extent to ritual recita-
tions, as illustrated by my earlier reference to " p r o m p t i n g "
(above page 181).
The transmission of ritual is illustrated on the Frontispiece,
where an experienced ritualist, Cherumukku Vaidikan Vasude-
van Akkitiripad, instructs his son Vallabhan, during the 1975
performance of the Agnicayaha, how to perform the nihnava rite
(see page xi; A G N I 1:359; MacFarland-Staal 1988; and cf.
Brough 1950). The son, C. V. Vallabhan Nambudiri, was the
youngest among the seventeen officiating priests (he assumed
the office of Nestä). He was already an excellent reciter, and had
witnessed numerous rites and ritual performances, but at the
time of the Agnicayana he had never performed nihnava himself.
His father appeared at the right time, and showed him precisely
how to do it. N o words were used. Neither father nor son would
pause to think of the outsider's question, what the nihnava rite
is supposed to mean. This is how ritual is transmitted—success-
 Music and Ritual 187

fully, one might add, since it has gone on in India for about three
thousand years. What is remarkable about this case of ritual
transmission is that it disregards not only meaning, but does not
even use language.
The most important contemporary anthropologist to pay at-
tention to music is Lévi-Strauss. According to Lévi-Strauss
(1964), there exist close links between myth and music, which
are both " m a c h i n e s or instruments for the obliteration of t i m e . "
Music is also, according to him, '-'the-supreme mystery of the
sciences of m a n " (la suprême mystère des sciences de
Vhomme). It is remarkable, however, how a synchronistic
approach, valuable not only in its own right, but necessary for
the understanding of structure, may obscure simple diachronic
facts. In the scheme of evolution, music is obviously much older
than myth. If there is a link between the two, it can only be that
myth has adopted structures inherent in music. Since ritual
displays similar structures, this explains the parallelisms be-
tween myth and ritual that have often been mentioned but that
do not signify anything.
Lévi-Strauss sees music embodied in " t h a t God, Richard
W a g n e r , " who has always exercised a morbid spell in French
intellectual circles. But this is mere fashion, like Lévi-Strauss'
own presentation of Le cru et le cuit as a musical composition,
subdivided into Overture, Theme and Variations, Sonata,
Fugue, Three-part invention, Rustic symphony in three move-
ments. " T h e conceit is not n e w , " says George Steiner, " o n e
finds it in Baudelaire's theory of correspondence (to which
Lévi-Strauss implicitly refers), in Mallarmé, and in Broch's
Death of Virgil, a novel ordered in analogy with the changes of
mood and rhythm in a string quartet. Lévi-Strauss does little,
moreover, to enforce the musical mimesis. It remains a labored
jeu d'esprit " The passage that contains the phrase referring to
music as " t h e supreme mystery of the sciences of m a n , " says:
But since music is a language with some meaning at least for the immense
majority of mankind, although only a tiny minority of people are capable
of formulating a meaning in it, and since it is the only language with the
contradictory attributes of being at once intelligible and untranslatable,
188 Ritual

the musical creator is a being comparable to the gods, and music itself
the supreme mystery of the sciences of man, a mystery that all the
various disciplines come up against and which holds the key to their
progress (Lévi-Strauss 1964:26; English translation 1969:18).

Lévi-Strauss often uses concepts taken from linguistics or

communication theory. A distinctive feature of human language
is that it possesses, in addition to phonological and syntactic
components, a third component which is semantic and which
enables language to convey meaning or "transmit m e s s a g e s "
(Chapter 6). This semantic component is linked in a precise and
intricate manner to the other two components, so that meaning
is systematically expressed by the speaker and understood by
the hearer when structured sounds are transmitted through the
air. Misunderstandings between users of a language occur, of
course, but on account of this system, combined with redun-
dancy and other communicative devices, these are kept to a
minimum. Otherwise language could not function.
Music and ritual are not languages precisely because they lack
the corresponding third component. Music can convey so many
different things because it has no intrinsic meaning: there is no
systematic relationship between its structured sounds and these
many things. Systems of music differ not only between cultures
(which is true of language); it is also patent within a culture that
music means different things to different people who are equally
familiar with it (which is not true of language). The only features
of music that can be universally perceived are either the purely
structural and physical (corresponding to syntax and phonology
in language), or nonmusical accessories such as names or titles,
provided of course they are explicitly attached. For this reason,
Debussy's piece called after a cathedral does not convey the
meaning cathedral in the systematic manner in which the word
" c a t h e d r a l " of a natural language such as English refers to a
cathedral (see above, page 177).
Music, then, conveys no message. It is intelligible only in two
respects: objectively with reference to its structure and physical
properties, including labels arbitrarily attached to it; and subjec-
tively with reference to everything else it may or may not convey
 Music and Ritual 189

to various individuals, depending not only on their imagination,

but also on their individual past experience C'La musique est un
fait de m é m o i r e " as the French proverb says). Music is untrans-
latable because there is nothing to translate (for translation of a
sentence A in one language into a sentence B in another language
is, roughly, detaching the meaning expressed by the form of A
from A, and providing it with a new form, B). That the alleged
messages are sent out by a few and understood by many is not
only false (because there are no messages to understand) but
also irrelevant: with reference to language the same can be said
and would be tautologically true of prophets, but this throws no
light on the nature of language.
That in music, form is more important than content has been
expressed by musicians and musicologists in different ways. I
shall confine myself to two examples from Igor Stravinski:

Î regard music essentially as incapable of expressing anything whatso-

ever: a sentiment, an attitude, a state of mind, a natural scene, etc.
Expression has never been an immanent property of music . . .
The phenomenon of music has been given to us only for the sake of
ordering things. It requires for its realization, necessarily and uniquely,
a construction. When the construction is complete, when order has been
established, everything is said and done. It would be useless to seek or
expect anything else (quoted in Boulez 1966: 9-10).

Should we, then, abandon meaning altogether? It is difficult to

say, at this stage. On methodological grounds we should pursue
the syntactic approach that has been totally neglected in the
study of ritual in spite of the fact that it obviously works. In
order to do this, we cannot operate on a level of generalities. Our
data must be specific, exact, and as exhaustive as possible. We
have seen that such data are available in India and other parts of
Asia, and can be studied through literature as well as through
fieldwork. Other ritual traditions of mankind insofar as they are
still accessible should be studied exhaustively and with the same
concern for detail that the Indian and Chinese ritualists^ them-
selves already displayed, but that is rarely employed by contem-
porary Western anthropologists, sociologists, or scholars of
190 Ritual

religion. The resulting structural descriptions should be com-

pared with the structural descriptions of animal rituals, and will
then yield unbiased results with regard to the relationships
between humans and other animals. When we have done all that
and still feel that something is lacking, we are entitled, by all
means, to search for meaning. Even then it recommends itself,
again on methodological grounds, to adopt the Talmudic saying
that is the motto of the Harvard Dictionary of Music although it
is more appropriate to ritual (which is not only audible but also
visible):

If you want to understand the invisible,

look carefully at the visible.

The MImämsä philosophy of ritual supplements this:

drsie sambhavaty adrstakalpanâ 'yogâî

"for when a visible result is possible, it is improper to postulate an
invisible one" (MImämsä Paribhäsä, ed. Mädhavänanda, 1948:73-
74).
 Part III

Mantras

Introductory Note

We have seen that Vedic ritual consists of rites and mantras,

a term used for recitations and chants that accompany rites and
that are derived from sentences or bits and pieces of the Vedas.
The term mantra is derived from a verbal root man- which
means " t h i n k " and the suffix -tra which often expresses instru-
mentality. The etymology, therefore, suggests the meaning:
" i n s t r u m e n t to t h i n k . " However, this meaning fits only occa-
sionally, which is not surprising since meaning is determined by
usage, not etymology (above, page 155). Cognate words in other
Indo-European languages, which include English " m i n d " and
" m e a n i n g , " are suggestive but this fact is also without much
significance. If one wants a brief, approximate and practical
translation of the term mantra into English, " s a c r e d formula" or
" s p e l l " is often appropriate. But such translations do not
explain: they are soothing, at best. The question remains: what
are " s a c r e d formulas" and " s p e l l s " ? Modern associations are
unlikely to be helpful and it is more appropriate, therefore, to
inquire what the Indian tradition has to offer. This we shall do in
the next four chapters. That an interpretation like " s p e l l , " apart
from being vague, is not enough follows, for example, from the
fact that mantras are not only believed to conjure up deities
which is what a spell may be expected to do; in Tantrism, they
are deities. In other words, there must be more afoot.
The poets of the Vedas were fascinated by the power of
inspiration on which their poetic activity depended and to which
they referred with a variety of terms, including those derived
from the root dhi- such as dhih and dhitih (see Gonda 1963a).
192 Mantras

From this root, or its close relative dhyä-, derives the later
dhycina, the " m e d i t a t i o n " which Buddhism introduced into
China (as ch1 an) and Japan (as zen). Vedic poets knew that
inspiration is not enough unless it is manifest in language. These
manifestations were also referred to by several terms, one of
which is mantra. At an early period, the term accordingly
denoted the versified portions of the Vedas (including the
Rigveda in its entirety) as distinguished from their prose pas-
sages, called hrähmana, not considered similarly inspired and
consisting of comments and explanations of the poems given
either by the bards themselves or, more likely and increasingly,
by thinkers of a more speculative bend of mind: interpreters,
scholars and philosophers.
During the middle Vedic period—approximately, from 1,000
B.c.E. to 500 B.c.E.—ritual became the most prominent expres-
sion of Vedic culture and civilization. The ritual meaning of
mantra already referred to—a recitation or chant accompanying
a rite—was now established and became widespread. Mantras
were derived from all Vedic compositions, and may therefore
include re, " v e r s e (from the Rigveda)," säman, " c h a n t (from
the S ä m a v e d a ) , " yajus, "formula (from the Yajurveda; gener-
ally m u t t e r e d ) , " and nigada, "formula (also from the Yajur-
veda, but generally spoken l o u d ) . " Mantras are always the
elements that are recited or chanted; they are neither the ritual
acts themselves, nor their glosses or meanings, nor Vedic
expressions that prompt such acts. In the later systematization
of ritual, for example in the ritual philosophy of the Mïmâmsâ,
mantras are distinguished not only from brâhmanas but also
from the vidhis or "injunctions" that enjoin ritual acts.
I shall primarily study mantras in this second, ritualistic, sense
and in their Vedic contexts, but refer to the later Tantric
(Buddhist and " H i n d u " ) uses as well. With reference to these
latter traditions, i shall use the term " m a n t r a " in the more
general but as yet unspecified sense in which it is generally
employed in the study of Indian religion. We shall see that
mantras or similar entities occur in other Asian traditions as
well. In India, the emphasis is always on oral elements and
 Mantras 193

transmission; elsewhere in Asia, " m a n t r a s " need not be recita-

tions or chants but may be " w r i t t e n , " like the pseudo-Sanskritic
" c o s m i c writing" or "celestial script" of Taoism (Bokenkamp
1983:463; Schipper and Hsiu-huei 1986:193).
Many ritual structures consist largely of mantra recitation.
The Isti (Chapter 9A) consists, prior to the final act of oblation,
of mantras (Wheelock 1985:180, for this reason, calls it a
" m a n t r a cluster"). Other rites—for example, the Soma se-
quence (Chapter 9B)—consists of mantras (chants and recita-
tions) as well as acts. According to some Srauta Sütras there is
a one-one-correspondence between mantras and acts, expressed
as follows in Àpastamba Srauta Sütra 24.1.38:

ekamanträni karma ni S'each act is accompanied by one mantra."

There are exceptions (as the examples cited demonstrate) but

according to the sütra, they should always be marked:

vcicancid ekam karma bahitmantram, "when it is explicitly stated, one

act corresponds to several
mantras." (Äpastamba Srauta
Sutra 24.1.44).

In actual practice, exceptions are numerous, including many not

explicitly referred to in the sütras. Among the former, some
exhibit regular features and are formulated by rules. It is
common in Srauta ritual to repeat rites four times, three times
with a mantra, and the last time without. That is, if we refer to
ritual acts as " A " and to mantras as " M " , the resulting
structure may be represented by:
AM/AM/AM/A. (1)
This is often mentioned when such a sequence occurs, e.g., in
Baudhâyana Srauta Sütra 10.23:21.14:

trir yajusci tüsnirn caturtham, k'thrice with a yajus, the fourth time
silently" (cf. AGNI 11:524-525).

The corresponding practice in the domestic, Grhya ritual is a

triple repetition of the ritual act of which only the first is
194 Mantras

accompanied by a mantra. The rule, which I have not found in a

text, is: sakrn mantrena dvis tusnlm (C. R. Swaminathan, per-
sonal communication). Gonda (1980b:362), who provides similar
references, adds that we also find rites repeated thrice, " t w i c e
with different mantras and once silently." Thus we find two
structures:
A M / A /A (2)
and:
A M / A N / A (3)
(cf. Staal !988d).
As these examples indicate, deviation from the norm of
one-one-correspondence arises more easily through acts being
performed without mantras, than through mantras being recited
without acts. Exceptions are mentioned (Mïmamsa Sütra 12.4.1
refers to mantras that do not accompany acts as
akannasamyuktah) but they are rare. In general, mantras are
more important than acts. Baudhäyana illustrates this and dem-
onstrates in addition that acts may owe their existence to
mantras. In a case where the number of mantras is greater than
that of acts, new acts are introduced to make up the difference
and save the principle of one-one-correspondence. Fully de-
scribed in A G N I (1:493-503), it is of sufficient interest to be
briefly explained here.
Almost all the mantras with which the bricks of the Agnica-
yana altar are consecrated are taken from specific sections of the
Taittirïya Samhitâ of the Black Yajurveda. However, these
sections contain 118 more mantras than the fixed number of
1,000 bricks, made up of five layers of 200 each (actually, there
are five more since ten bricks are half as thick as the others: see
A G N I 1:202-203, 482-485). Such a discrepancy could be re-
solved in one of three ways: either the principle of one-one-
correspondence is violated or abandoned; or the extra mantras
are not recited; or new acts are introduced to make up the
required number. The last alternative is adopted by
Baudhäyana: 118 pebbles (called sarkara in Sanskrit and
 Mantras 195

kölipparan, "chicken-fish," in Malayalam) are inserted between

the bricks, like mortar or cement. Each pebble is treated like a
brick, that is, consecrated with three mantras, the first two
always the same (as in the case of the other bricks), the third
mantra specific to each pebble and taken from the 118 extra
mantras. Thus the number of mantras and the principle of
one-one-correspondence are both preserved. (The term sarkara,
incidentally, is pre-Indo-European: it is cognate with Latin
calculus and, via Arabic, with English sugar.) Another impor-
tant rule about the relationship between mantras and acts states:

manträntcüh karmadïn samnipatayeî

"one should let the beginning of the acts coincide with the end of the
mantras"
•

( Ä p a s t a m b a S r a u t a S ü t r a 2 4 . 2 . 1 ; t h i s t o p i c is f u r t h e r d i s c u s s e d i n

M ï m a m s â S ü t r a 1 2 . 3 . 2 5 a n d f o l l o w i n g ) . A n e x a m p l e o c c u r s i n

t h e Isti r i t e w h e r e t h e o f f e r i n g is m a d e w h e n t h e H o t â s h o u t s

- s a t \ , t h e l a s t s y l l a b l e o f v a u s a t , " m a y ( A g n i ) l e a d t h e o f f e r i n g s

t o t h e g o d s . "

It is l i k e l y t h a t t h e i d e a t h a t m a n t r a s a r e s u c c e e d e d b y a c t s is

r e l a t e d t o t h e p r a c t i c e w e find e l s e w h e r e o f " m a g i c a l r i t e s "

b e i n g s u c c e e d e d b y " t e c h n i c a l o p e r a t i o n s . " T a m b i a h h a s d r a w n

a t t e n t i o n t o M a l i n o w s k i ' s a n a l y s i s o f t h e r e l a t i o n b e t w e e n T r o -

b r i a n d m a g i c a n d p r a c t i c a l a c t i v i t y , w h i c h s h o w s t h a t " t h e

w h o l e c y c l e o f g a r d e n i n g o r o f c a n o e b u i l d i n g m u s t b e s e e n a s

o n e l o n g s e r i e s o f a c t i v i t i e s w h i c h f o r m a r e g u l a r p a t t e r n o f M —»

T , M —» T , M —> T , M —» T : w h e r e M s t a n d s f o r t h e m a g i c a l rite

a n d T f o r t h e t e c h n i c a l o p e r a t i o n t h a t s u c c e e d s i t " ( T a m b i a h

1 9 6 8 : 1 9 8 , f o l l o w e d b y d e t a i l e d e x a m p l e s ) .

T o w h a t e x t e n t d o m a n t r a s refer t o t h e a c t s t h e y a c c o m p a n y ?

It is i m p o r t a n t t o h e a r w h a t t h i s q u e s t i o n a s k s . It is c l e a r t h a t

b r ä h m a n a s r e f e r t o m a n t r a s : t h e y w e r e c o m p o s e d t o i n t e r p r e t

t h e m . T h a t m a n t r a s a r e s u i t a b l e c o m p a n i o n s t o r i t e s a n d t h a t

t h e y t w o b e l o n g t o g e t h e r is a l s o n o t o p e n t o d o u b t . T h e q u e s t i o n

is, d o m a n t r a s e x p r e s s r i t e s , o r e x p r e s s w h a t r i t e s e x p r e s s , o r

r e f e r t o r i t e s in t h e m a n n e r in w h i c h l i n g u i s t i c e x p r e s s i o n s r e f e r

t o t h e i r m e a n i n g s ? I n t h e p a s t , m a n y s c h o l a r s h a v e t h r o w n d o u b t
196 Mantras

on the nature of this relationship. Gonda has dealt with this

problem in a series of publications (1981, 1982, 1985), including
a thorough examination of the mantras of the Agnyupasthäna
and the Sautramanï (Gonda 1980a). Gonda has emphasized that
the scholarly literature on Vedic ritual generally refers to man-
tras without ever quoting them in full (S. Einoo, in an unpub-
lished review of Gonda 1980a, shows that Gonda himself con-
tinues to do the same in the majority of cases). It is clear from
G o n d a ' s work that many connections between mantras and acts
exist. H o w e v e r , most of these are historical connections: they
throw light on the original associations that may have existed
between mantras and rites. But this is not the same as the ritual
function of mantras or their function in the ritual system.
If we adopt the synchronistic perspective of the Srauta Sütras,
and look at the system of mantras and acts as a " système où tout
se t i e n t " (as did de Saussure with respect to language) we find,
in general, that mantras do not refer to the acts they accompany.
Absence of reference is what linguists call the " u n m a r k e d "
case; there is no general term to refer to it. But a technical term
exists for those rks of the Rigveda that refer to, or " a d d r e s s "
{ahhivad-) the accompanying act. They are called
rüpasamrddha, "perfect in f o r m . " The existence of such a
technical term shows that these mantras were regarded as
special and exceptional. The majority of mantras, by implica-
tion, do not " r e f e r " or " a d d r e s s . "
"Perfection of f o r m " means no more than that the mantra
contains a particular word. Aitareya Brâhmana 1.16 (3.5) applies
the term to the mantra Rigveda 1.74.3, which contains the word
ajani, " i s b o r n , " and is recited when Soma " i s b o r n . " It refers
to such cases in the following terms:

etad vai yajnasya samrddham yad rüpasamrddham yat karma

kriyarnänam rg abhivadati, "the perfection of ritual is when it is perfect
in form, viz., when the rk addresses the act that is being performed" (cf.
Kane V.II,1962:1097).

We shall study the relationships between ritual acts and

mantras in greater detail, but it is important that the reader
 Mantras 197

develop—pace Gadamer—a feel for mantras first. Mantra is not

exactly what is sometimes defined as such or what it is felt to be
in subcultures of the counterculture. I shall therefore first list six
representative examples of Vedic mantras. They will be given in
Sanskrit, together with a translation into English. I have omitted
the accents for the sake of simplicity even though they are
considered important (we shall return to them in Chapter 24).
Afterwards the six mantras will be placed in their ritual context
and we shall return to their general discussion.
 17

Six Vedic Mantras

(1) agnin ...

agninjyotismatah kuruta I diksita vacant yaccha I patni vacant yaccha I

"Kindle the fires! Consecrated one, control your speech!
Wife, control your speech!"
{Baudhäyana Srauta Sütra 6.6)

(2) mitro na ehi . . .

mitro na ehi sumitradhä I indrasyorum à visa daksinam I usann

usantam syonah syonam I
"Come to us as a friend, making good friends. Enter the right thigh of
Indra; you willing, it willing, you gracious, it gracious"
{Taittirlya Samhitä 1.2.7.1 0

(3) yo'sman dvesti . . .

y o' s man dvesti yam ca vayam dvisma I idam asya grïvâ api krntämi I
"He who hates us and whom we hate, here I cut off his neck!"
{Taittirlya Samhitä 1.3.1.1 c)

(4) devasya tvä savitah . . .

devasya tvä savituh prasave' svinor bähubhyäm pu s no hastäbhyäm

agnaye jiistam nirvapâmi I
"On the impulse of the God Savitr, with the arms of the two Asvins, with
the hands of Püsan, I offer you dear to Agni."
(Taittirlya Samhitä 1.1.4.2 m)

(5) indra jusasva . . .

indra jusasva pra vahä yähi süra haribhyäm I pibä sutasya mater iha
madhos cakänas cärur madäya 11 indra jataram navyo na prnasva
200 Mantras

madhor divo na I asya sutasya svarnopa tvä madäh suväco

aguh I lindras turäsän mitro vrtram yo jaghäna yatir na I bibheda
valam bhrgur na sasahe satrün made somasya 11

"Indra enjoy—drive on,

come, hero—with your two steeds,
drink of Soma—like a sage,
loving the sweet, pleased with inebriation!
Indra, your belly—like one to be praised,
fill it with sweet—like heavens,
with pressed Soma—like paradise,
well-spoken inebriants have gone to you!
Indra fast conquering—like a friend,
killing the demon—like ascetics,
he split the cave—like Bhrgu,
he conquers his enemies inebriated with Soma!"
{Atharvaveda 2.5.1-3)

(6) hä bu hä bu hä bu . . .

hä bu hâ bu hä bu bhä bham bham bham bham bham bhâ bham bham

bham bham bham bhä bham bham bham bham bham I hâ bu hä bu hä
bu brahma jajnänam prathamam purästät I vi sïmatas suruco vena ä
vât I sa budhniyä upamä asya va yi sthâh I satas ca yonim asatas ca vä
yi vah I hä bu hä bu hä bu bhä bham bham bham bham bham bhä bham
bham bham bham bham bhä bham bham bham bham bham I hâ bu hä
bu hä vu va I brahma devânâm bhäti parame vyoman brahma devänäm
bhäti parame vyoman brahma devänäm bhäti parame vyomän II

Here translation becomes more difficult even than in the prev

ous case; but it may be attempted, in free fashion, as follows:

"Hey hey hey! BANG bang bang bang bang bang BANG bang bang
bang bang bang BANG bang bang bang bang ! Hey hey hey! Born as
brahman first in the ea-east, Vena has shone out of the glimmering
horizon. He has revealed its highest and lowest positionemes, the womb
of being and of non-be-beying. Hey hey hey! BANG bang bang bang
bang bang BANG bang bang bang bang bang BANG bang bang bang
bang bang! Hey hey, hey man! Brahman shines in the highest heaven of
the gods brahman shines in the highest heaven of the gods brahman
shines in the highest heaven of the gogodeses!"
(Jaiminfya Aranyageyagäna 12.9)
 18

Ritual Context

The six expressions which we have listed are either mantras or

consist of mantras. I have listed them in a sequence in which the
formal elements gradually increase and the semantic elements
decrease: that is, they increasingly exhibit " m a n t r a " character-
istics. Let us now review these mantras again, and place them in
the ritual context which conveys their "ritual m e a n i n g . " We
shall find that in the simple cases listed first, the relationship
between mantras and acts is simple. In the first case, the mantra
may be said to " r e f e r " to the act. In the later cases, that
relationship becomes increasingly complex and indirect until it
virtually disappears.

(1) agnlh . . .

This mantra is a command and belongs to the category of

nigada. It is addressed by the Adhvaryu priest, shouting in a
loud voice, to the other priests, the Yajamäna, and the
Yajamäna's wife, after the Yajamäna's consecration has taken
place. Following the mantra, fuel is added to the fires, and the
Yajamäna and his wife "control their s p e e c h " (i.e., they pro-
nounce only what is prescribed, but do not chatter: see A G N I ,
1:333). It stands to reason to assume that this mantra is an
ordinary command, that has been understood as such by those to
whom it was addressed. This implies, among other things, that
the Adhvaryu priest is the kind of person who has the authority
to issue such commands.
202 Mantras

(2) mitro na ehi . . .

This mantra is a yajus, muttered by the Yajamâna after the

Soma plant has been purchased by the Adhvaryu from a
merchant. The Yajamâna mutters the first part of the mantra
(through . . . sumitradhä) when the Adhvaryu approaches him
with the Soma bundle. He then uncovers his right thigh, places
the bundle on it, and recites the remainder of the mantra
(Caland-Henry 1906, 1:46; Srautakosa, Sanskrit Section, II,
1:50). Here no command is given or followed. The mantras
accompany an act or acts, and may be interpreted as comments
on that act or on those acts.

(3) yo'smàn dvesti .

This mantra has a purely ritual use: it is recited when the soil
within a ritual enclosure is prepared with the help of the sphya,
a wooden knife. One of the brähmanas associated with this
mantra provides it with an interpretation which is, as is usual, a
rationalization: the enemy has to be excluded from the altar, for
making the altar is a cruel act. " L e t him think of anyone he
hates; he does truly inflict trouble upon him!" (Jaittirlya
Samhitâ 2.6.4.4). Another brähmana comments: ' T h e r e are two
persons: one whom he hates, and one who hates him. Surely, he
should cut off the necks of both, successively" (Taittiriya
Samhitä6A.$A; cf. A G N I 1:104).
When I call such interpretations rationalizations, I do not
intend to deny that there were real enemies in Vedic times,
whose necks could be and actually were cut off. There is ample
evidence for battles, sometimes intruding on ritual (see, e.g.,
Heesterman 1962). Such a background is reflected in the mantra,
and points to one of its possible origins. However, the
" m e a n i n g " of the mantra is its ritual use, that is, the act it
accompanies; in ritual terms, the mantra means that the soil is
scratched with the sphya.
 Ritual Context 203

(4) devasya tvä savituh ...

This mantra, recited frequently, accompanies and indicates

the measuring out of the material for an offering (nirvâpa). The
first three phrases (through hastäbhyäm) occur at the beginning
of many other mantras (see Bloomfield 1906:492-94). Character-
istically, the brâhmanas are unhelpful, e.g.:

He says wkOn the impulse of the God Savitr," in order to give impulse (to
his action). "With the arms of the two Asvins," (for) the two Asvins
were the Adhvaryus of the Gods. "With the hands of Püsan," for the
guidance (of his action). "I offer you dear to Agni," he offers it indeed
dear to Agni {Taittirïya Brâhmana 3.2.4; Dumont 1957:227-28).

That the ritual meaning is only "measuring out for an offering"

is obvious from a discussion in the Mïmâmsâ Sütra (2.1.46). The
purpose of this discussion is to establish that mantras always
consist of a single sentence because they express a single
meaning {arthaikatväd ekam väkyarn). The commentator Sabara
elucidated this as follows: ' T h e sütra is explained because
mantras fulfill a single purpose. Devasya tvä . . . , for example,
indicates 'measuring out for an offering'. The words that com-
prise the mantra express precisely this, and therefore consist of
a single s e n t e n c e " {ekaprayojanatvcid upapannam I yathâ tävad
devasya tveti nirväpaprakäsanam I tasya visistas^a vâcaka
etâvân padasamühas tadväkyam).

- (5) indra jusasva ...

These verses are mantras in the original sense: they are

inspired compositions and may, in fact, have been composed
under the influence of Sonia. This is rare, if not exceptional. In
the Rigveda, there is only one hymn (10.119) which describes the
effects of drinking Soma in detail. Even with respect to this
hymn, Brough (1971:341) judges: " S u c h a hymn cannot have
been composed by a poet under the influence of soma: the
artifice of its structure excludes t h i s . "
I don't know whether this verdict is true, but there are good
204 Mantras

reasons to doubt it. I knew at least one mathematician who could

only do mathematics when he was drunk, not on account of the
auspicious inebriation {sumada) of Soma, but on account of the
evil intoxication (durmada) of alcohol. It is therefore not unrea-
sonable to suppose that the mantras indra jusas va . . . might
have been composed under the influence of Soma, even though
they consist of svaräj meters—relatively uncommon meters
consisting of 34 syllables each.
These verses are the point of departure for modifications that
turn them into mantras, fit for ritual use. They constitute the
source from which ritualists have constructed the beginning of
the sixteenth Soma sequence called the " s i x t e e n t h " (sodasï).
There is a good arithmetical reason for the number sixteen
(Chapter 9B), but it is also a favorite, auspicious number (see,
e.g., Gonda 1965:115-130). The Sodasï sastra consists of
anustubh verses, their meter defined by a sequence of four
octosyllabic verses, each consisting of 8 syllables. An anustubh,
therefore, consists of 32 syllables, that is twice sixteen.
Since the mantras indra jusasva . . . consist of three verses in
the svaräj meter, and the first verse of a sastra recitation is
always recited thrice, we have 5 x 34 = 170 syllables at our
disposal. If we disregard the syntax and meaning of these verses,
and concentrate on the counting of syllables only, we can make
use of 160 = 5 x 32 syllables to obtain 5 anustubh verses, leaving
an excess of 10 syllables. Such a procedure is in accordance with
the stress on formal features such as meter that characterizes
Vedic mantras. In terms of syntax or meaning, the resulting
anustubh verses do not make sense, for they are arrived at by
cutting off' the last two syllables of the first verse and adding
them to the beginning of the second (which is a repetition of the
first); cutting off the last four of the second and adding them to
the beginning of the third (another repetition of the first); cutting
off the last six of the third and adding them to the beginning of
the fourth; cutting off the last eight of the fourth and adding them
to the beginning of the fifth; and cutting off the l a s t i g of the fifth
and putting them in storage, so to speak.
 Ritual Context 205

The entire procedure may be pictured as follows:

To provide a translation is hazardous, but an idea may be gained

from the following:
1. "Indra enjoy—drive on,
come hero—with your two steeds,
drink of Soma—like a sage,
loving the sweet, pleased with!
2. Inebriation, Indra enjoy,
drive on, come, hero, with
your two steeds, drink of Soma,
like a sage, loving the sweet!
3. Pleased with inebriation—Indra,
enjoy, drive on, come, hero,
with your two steeds, drink of
Soma liké^a sage, loving!
4. The sweet, pleased with inebriation, Indra,
your belly, like one to be praised, fill,
it with sweet—like heavens with,
pressed Soma, like paradise well-spoken!
5. Inebriants have gone to you, Indra,
fast conquering like a friend killing,
the demon like ascetics he split,
the cave like Bhrgu he conquers!''
The r e m a i n d e r — " H i s enemies inebriated with S o m a ! " — i s used
for the beginning of the next part of the sodasi sas tra, which I
shall not write out in full, because it results in the same kind of
meters, and the same kind of absurdities in terms of syntax and
semantics.
206 Mantras

Later in the sastra (which is very long) use is made of a

technique called viharanam, ''intertwining," or " t r a n s p o -
sition." Its first occurrence is in the construction of two
anustubh verses (consisting of 2 x 32 syllables) from intertwin-
ing a gâyatn verse (consisting of 3 x 8 syllables) with a pahkti
verse (consisting of 5 x 8 syllables):
3 x 8 = 24
5 x 8 = 40 ,

2 x 32 = 64.
The gâyatn verse is Rigveda 1.16.1 :

ä tvä vahantii harayo vrsanam somapitaye I indra tvä sûracaksasah 11

'The tawny horses take you bull to the Soma drinking,
You, Indra, with your sunny eyes!"

The pahkti verse is Rigveda 1.84.10:

svädor itthä visüvato madhvah pibanti gauryah I yä indrena sayävarlr

vrsnä madanti sobhase vasvïr anu svaräjyam 11
"The gaurï cows drink from the sweet liquid, basic to the ritual, enjoying
themselves with their companion, Indra the bull, to look beautiful;
bénéficient to his supremacy."

The intertwining of these two is as follows:

ä tvä vahantu harayas svâdor itthä visüvatah / vrsanam somapitaye

madhvah pibanti gauryo / / indra tvä süracaksaso yä indrena sayävarih /
vrsnä madanti sobhase vasvïr anu svaräjyo / /

In this construction, the portions from the underlying gâyatn

verse are in italics, and the portions from the underlying pahkti
verse are in Roman. (The -o ending is another feature of sast/a
recitation, to which I shall return.)
The meaning can only be guessed at, but the following may
convey some of its flavor:

"The tawny horses take from the sweet, basic to the ritual. You bull to
the Soma drinking, the gauri cows drink from the liquid. You, Indra, with
 Ritual Context 207

your sunny eyes—enjoying themselves with their companion, Indra the

bull, to look beautiful; bénéficient to his supremacy."

An intoxicated Sanskrit scholar might interpret this as a poetic

rendering of a Soma orgy; however, it merely results from the
metrical arithmetic of the viharanam technique. In terms of
syntax or semantics, none of these mantras make sense; their
ritual meaning, on the other hand, is straightforward and uncon-
troversial: they constitute the first portion of the sixteenth
sastra.
In the sequel there are further cases of viharanam, and also
instances where mantras, though recited in regular sequence,
are re-analyzed into anustubh meters by counting the syllables
of their original meters differently. The reader interested in these
exercises can find them in A G N I , 1:661-663, and can listen to
them on the accompanying cassette. The examples given should
be sufficient to illustrate the ritual use and meaning of such
mantras.

(6) hä bu hä bu hä bu . . .
These mantras are chanted by the Udgätä priest of the
Sämaveda after the Adhvaryu has placed a small image of a
golden man (hiran-mayapurusa) above the lotus leaf that was
earlier deposited and buried at the center of the Agni field; later
the large bird-shaped altar of the Agnicayana will be constructed
there. These chants (see A G N I , 1:414-417 and the accompany-
ing cassette), which continue through some of the following
rites, consist of four parts, and the mantras we are considering
constitute the last chant of the third part. In this third part, many
chants are similar in structure. They start with: ha bit hä bu hä
bu . . . .followed by a triple repetition of six syllables, five of
them identical, and the first a variation, e.g.:
phut phut phut phut phut phut
hä hu hau hau hau hau hau
kä hvä hvä hvä hvä hvä.
This is followed, in each case, by a verse, generally from the
Rigveda, set to music in accordance with a melody (säman),
208 Mantras

after which there is another round of meaningless syllables and

finally a "coda" (nidhana), also meaningless.
Such meaningless syllables from the Sämaveda are called
stobha. If Vedic mantras are called bits and pieces, it is the
stobhas that are the bits. Stobhas are very similar to the
6()'tf-mantras of later Tantrism, to which we shall return (Chapter
20).
The six mantras we have reviewed suggest that the ritual use
of mantras has little or nothing to do with their meaning. This is
independent of the fact whether or not these mantras had
meaning to begin with. What is important for their ritual use is
the formal characteristics they possess. The stobhas of the
Sämaveda are particularly rich in repetitions, inversions, inter-
polations, and other specific structures. This is the topic of the
next Chapter.
 19

T h e Syntax of Stobhas

The stobha chants of the Sâmaveda may be analysed from

different points of view. I shall not be concerned with either
textual or musical analysis, but with the structure and distribu-
tion of some of these chants. The material, based upon the 1975
performance described in A G N I , is in several respects different
from the Srauta texts, as a comparison of the following notes
with Parpola 1983 shows.
When referring to the unpublished chants of the Jaiminïya
Sâmaveda, I adopt the system of reference used in the manu-
scripts put at the disposal of Asko Parpola by Itti Ravi Nambu-
diri, t h e foreriiost sämaga of Kerala. In these manuscripts—
written down in the Malayalam script, without sound notation,
and largely from memory (that of Itti Ravi, his elders, and his
pupils)—the Jaiminïya Ärcika is divided into 112, the
Grâmageyagâna into 59, and the Aranyegeyagäna into 25 öttus
or " s o n g s " . I shall chiefly refer to the chants of the
Aranyegeyagäna, which the Nambudiris call candrasämäni,
" m o o n c h a n t s " . A reference such as AG 25.7 would thus denote
the seventh sâman of the 25th öttu of the Jaiminïya
Aranyegeyagäna . For the terminology, see below, page 297.
The first Agnicayana chants (AGNI 1:410-411) are sung im-
mediately after the Adhvaryu has placed a lotus leaf at the centre
of the Field of Agni (agniksetra) over which the bird-shaped altar
will subsequently be constructed. The Udgätä enters, and takes
up his position to the west of what will be the tail of the bird,
against the northern post of the eastern door of what will later
become the Havirdhâna shed. From this position, he sings most
of the Jaiminïya chants that characterize the Nambudiri Agnica-
210 Mantras-

yana. The first chant is based upon a cryptic mantra of the

Taittirïyasamhitâ (4.2.8.2d), which also occurs in the Atharva-
samhitä (4.1.1), but not in the Rksamhitä. The Adhvaryu recites
it at the same time, while he places the golden breastplate
(rukma) which the Yajamäna wore at his consecration to the
north of the lotus leaf:
brahma jajnänam prathamam purastâd
17 sîmatah suruco vena âvah
sa budhniyä upamä asya visthäh
satas ca yonim asatas ca vivah
kk
Born as brahman first in the east,
Vena has shone out of the glimmering horizon.
He has revealed its highest and lowest positions,
the womb of being and non-being."

This verse (above, page 200) is turned into a chant consisting

of the five customary parts ( 1 : prastäva; 2: udgïtha; 3: pratihära;
4: upadnivtr, and 5: nidhana\ see page 299) by prefixing and
affixing stobha elements that will be referred to with the help of
capital letters, in the following manner:

A; have hâ yi
B: he saya
C: au ho va
D: e rtam amrtam.

I shall refer to the four lines of the verse of TS 4.2.8.2d with the
help of lower case letters: a, b, c, and d, respectively. Then the
chant can be represented as follows:

prastäva: A A B a /
udgïtha: b /
pratihâra: c / (4)
upadrava: d A A B C /
nidhana: D D D /

We need to adopt one more convention to interpret this

correctly: whenever there is a triple occurrence of a stobha, viz.,
 The Syntax of Stobhas 211

an expression of the form X X X , the final syllable of the third

occurrence is lengthened. For example, in D D D, the third
occurrence ends in amrtäm, and not in amrtam.
Written out in full, the above expressions represent the
following chant, which is Jaiminïya Grämageyagäna 33.9.2:

prastäva: have hä yï have hä yi hesäyälbrahma jäjhänäm

präthämäm purästätl
udgîtha: vi slmatäs suruco vena ä vätl
pratihära: sa budhnyä upamä asya väyisthähl
upadrava: satas ca yonim äsätäs ca vïvah have hä yï have hä
yï hesäyä au ho väl
nidhana: e rtam amrtam e rtam amrtam e rtam amrtäm/

The only feature that is not represented in the formula (4) is the
lengthening of certain vowels within the lines a, b , c, and d of the
mantra. Of course, further abbreviations of this representation
are possible. For example, the sequence c t A A B " may be
replaced by t c W ' \ In that case, the chant becomes:

1. Wa7
2. b/
3. c / (5)
4. d WC/
5. DDD/

The advantage of these representations is that they picture the

structure of the chant clearly, and enable us to compare the
structures of different chants with each other. Such representa-
tions also enable us to express in a simple form differences
between different traditions and schools. For example, the
corresponding Kauthuma-Rânâyanïya chant differs from the
above Jaiminïya variety only in that two of the stobha elements
have different forms: A has to be replaced by:
A*: have hàl
and B has to be replaced by:
212 Mantras

B*: hi sä yä.
If these substitutions are made in (5), the result is Kauthuma-
Rânâyanïya Grâmageyagâna 321.2 (in the edition of R. Näräya-
nasvâmï Dïksita).
F r o m now on, I shall not write out the texts in full, but only
represent them by symbolic representations such as (4) or (5).
The second chant of the Udgätä that accompanies the Adh-
v a r y u ' s rite with the golden breastplate is a musical composition
on a single word: satya,"truth". The stobhas may be referred to
by:

E: ho yi
F: hä ä vu vä
G: e suvar jyotih

The chant may then be written as:

AG 25.24: satyom I satya E satya E satya F / G. (6)
H o w much more abbreviation or simplification should be re-
sorted to, in a case like this, depends entirely on the occurrence
or nonoccurrence of other chants of a similar form: if there are
no others, there is no point in abbreviating any further, b u t if
there are, it depends on the degree of similarity between them to
what extent further abstraction may be helpful in expressing the
structure.
After these relatively modest beginnings, the Udgätä bursts
into a much longer sequence of songs. These accompany the
deposition by the Adhvaryu of the golden man (hiranmayapu-
rusa) upon the lotus leaf, and continue through several subse-
quent rites. This sequence consists of four parts (AGNI 1:414—
417). The first is called the Great Chant (mahäsäman: AG 25.7),
and the second consists of seven songs (AG 9. 1-7), based upon
verses of the Purusa hymn of the Rigveda (10.90), with changes
in the text and in the order of these verses. I shall not analyze
these two parts here, because it is not easy to extract or abstract
a general structure from them.
 The Syntax of Stobhas 213

The third part begins to exhibit marked regularities, partly

obscured by irregularities. It is quite possible that the latter have
crept in over the centuries, for these chants have been sung for
almost three millenia. This third part consists of nine Moon
Chants, AG 12. 1-9. Four of these, AG 12. 3-6, consist entirely
of stobhas and are relatively short. Of the remaining five, three
(AG 12. 7-9) exhibit the same structure, and two (AG 12. 1-2) a
very similar pattern. I shall confine myself here to the structure
that is the most obvious, and that can be represented in a simple
manner with the help of our notation if we adopt one further
convention, viz., express repetition of elements by superscripts.
For example, instead of writing " R R R R R " for a five-fold
repetition of the element " R " , I shall write: " R 5 " .
The structure of each of AG 12. 3-6 may now be represented
by:

P 3 (QR 5 ) 3 P 3

P3(QR5)3P2P* (7)

The use of parentheses is self-explanatory: everything within

parentheses'should be repeated as many times as is indicated by
the superscript following the closing parenthesis. T h u s ,
44
(QR 5 ) 3 " stands for: 44 QR 5 QR 5 Q R 5 i \ or: 4 4 QRRRRR
Q R R R R R Q R R R R R " . " X " represents an underlying mantra,
different for each of the three songs, and 4 4 Y" represents the
nidhana which consists of the final portion of this mantra and/or
a stobha. The stobhas which exhibit the invariant structure are
44
P " , 4 4 Q " , and 4 4 R". Of these, 4 4 P " is the same in the three
songs:

P: hâ bu.

" P * " is a modification of 4 4 P " which is used in the final round

when 4 t P " is repeated only once and its third occurrence (like
the amrtamiamrtarn we considered before) is replaced by:

P*: hâ vu vä.
214 Mantras

While the structure of the three chants is the same, the

remaining stobhas, 4 i Q " and " R ' \ are different, in the following
manner:

AG 12.7 has Q: phät

R: phät.
AG 12.8 has Q: hä bu
R: hau
AG 12.9 has Q: bhâ
R: bham.

Written out as far as its stobhas are concerned, the last chant,
for example, becomes:

hä bu (3 x ) bhâ bham bham bham bham bham (3 x ) hâ bu (3 x )

X
hä bu (3 x)-bhâ bham bham bham bham bham (3 x) hä bu
(2 x) hä vu vä
Y

In this chant, X happens to be the same mantra TS 4.2.8.2d we

have met with before. The structure of AG 12.1-2 deviates to
some extent from this pattern (7), but it also possesses the
characteristic feature " ( Q R 5 ) 3 ' \ in the following manner:

AG 12. 1 has Q: u
R: ha
AG 12. 2 has Q: •kâ
R: hvâ

The fourth and last part of this sequence consists of a single

chant, similar to the chant for the golden breastplate (6), but with
purusa as the main stobha:
AG 25.25: purusom I purusa E purusa E purusa F / G (8)
After the agniksetra has been prepared, the ritual continues
with the piling up of the five layers of the altar. The bricks are
 The Syntax of Stobhas 215

consecrated by the Adhvaryu on behalf of the Yajamäna, and the

Udgätä contributes songs to some of these rites. I shall here
consider the sequence of chants that is sung when the " S p a c e
Filler Bricks 1 1 are consecrated. Most of the bricks are conse-
crated in a specific order, and are therefore numbered, at least
conceptually (cf. Staal 1982, Lecture III). The only exceptions
are certain bricks, occurring especially in the intermediate layers
(i.e. the 2nd, 3rd, and 4th), that are consecrated without an
individual mantra and in any order. These bricks are not
consecrated without mantras, but the mantras are the same for
each brick. There are three: the first two are called tayädevatä
and südadohasa. These are used for the consecration of every
brick of the altar. The third is the specific " S p a c e Filler"
(lokamprna) mantra (Taittirïya Samhitâ 4.2.4.4n):

lokam prna chidram prnä 'tho sida sivâ tvam

indrâgnï tvâ brhaspaiir asmin yonâv cisisadan

"Fill the space! Fill the hole!

/ Then sit down in a friendly manner.
Indra, Agni, and Brhaspati
have placed you in this womb."

While the Adhvaryu recites these mantras over the Space

Filler Bricks, the Udgätä intones eight Space Filler Chants: AG
24.5-6 and AG 25.32-37. The latter six are of the same form as
(6) and (8), but other stobhas are substituted in the place of satya
or purusa:

§3. AG 25.32: agna for Agni

§4. AG 25.33: väya for Väyu, " w i n d "
§5. AG 25.34: sürya, " s u n "
§6. AG 25.35: candra, " m o o n "
§7. AG 25.36: näka, " v a u l t "
§8. AG 25.37: sukra " g l o w " or " V e n u s . "

The nidhana portions are not always the same. At this point it
has become obvious that it would be helpful to express the
216 Mantras

structure of these chants by representing them by means of a

general functional expression, e.g., V (X), defined as follows:
* ( X ) = " X - o m I X ho yi X ho yi X M à vu va / "
In this expression, u X - o m " is obtained from " X " by replacing
the final " - Ö " of X by " - o m " . An example is:
^ (satya) = "satyom I satya ho yi satya ho yi satya hä ä vu
väl"
The different nidhana portions may now be substituted, and
all the chants of this form that we have so far considered may be
represented as follows:

AG 25.24 W(satya) G
AG 25.25 ^{purusa) G
AG 25.32 ^(agna) e jyotih
AG 25.33 ^{vâya) e räjä
AG 25.34 V(sürya) e bhräjä
AG 25.35 ty(candra) e ä bhräjä
AG 25.36 ^(näka) e prstham
AG 25.37 W(sukra) e bhrälä bhräjä.

Other chants of this form are sung by the Udgätä on the three
occasions (on the first, third, and fifth layers of the altar) when
the "perforated p e b b l e s " (svayamätrnnä) are deposited at the
centre by the Adhvaryu in collaboration with the "Ignorant
B r a h m i n " (AGNI 1:419, 461, 505; cf. Staal 1978a and 1982a:42-
53). Using our notation, these three chants may be represented
as follows:

on the first layer: AG 25.21 V(bhûr)G (for bhü, " e a r t h " )

on the second layer: AG 25.22 ^(bhuva)G (for " s k y " )
on the third layer: AG 25.23 ty(suva)G (for " h e a v e n " )

In each of these three cases, " G " represents again: e suvar

jyotih.
The last sequence of songs I shall consider is chanted after the
 The Syntax of Stobhas 217

bird altar has been completed and fully consecrated. It is now

vibrating with power, ferocious (krûra) and dreadful (ghora),
and has to be pacified and brought under control. To this end the
Adhvaryu, assisted by the Pratiprasthäta, pours a continuous
libation of goat milk over the furthest western brick of the
northern wing. This brick is chosen because it is eccentric, i.e.,
far from the centre of power, and also because it can be easily
approached from different sides (AGNI 1:509 sq.). While per-
forming this oblation, the Adhvaryu recites the famous Satani-
driya or Rudram (Taittirïya Samhitä 4.5), which derives its
popularity partly from the fact that it was later interpreted within
the perspective of Saiva theism (Gonda 1979). During this
oblation and recitation, the Udgätä chants a sequence of 57
sämans, together called Flow of Milk (ksîradhârâ). These chants
last very long and continue after the Rudra ceremonies have
been completed. Their complete structural analysis would take
up more space than is available here, but I wish to draw attention
to two of their most striking features.
I shall first take up the one that appears last. The final
seventeen of these 57 Flow of Milk chants have the structure of
(6) and (8) we have just considered, and incorporate again the
chants we have already mentioned. The others can be repre-
sented in a straight-forward manner with the help of our notation
in terms of ^ and G:

§41. AG 25.21 (see above page 216)

§42. AG 25.22 (see above page 216)
§43. AG 25.23 (see above page 216)
§44. AG 25.24 (see above pages 212, 216)
§45. AG 25.25 (see above pages 214, 216)
§46. AG 25.26: ¥ (gaur) G
§47. AG 25.27: ¥ (loka) G
§48. AG 25.28: ¥ (agner hrdayd) G
§49. AG 25.29: ¥ (dyaur) G
§50. AG 25.30: ¥ (antariksa) G
§51. AG 25.31: V {prthivî) G
§52. AG 25.32 (see above pages 215, 216)
218 Mantras

§53. AG 25.33 (see above pages 215, 216)

§54. AG 25.34 (see above pages 215, 216)
§55. AG 25.35 (see above pages 215, 216)
§56. AG 25.36 (see above pages 215, 216)
§57. AG 25.37 (see above pages 215, 216)

In this list, I have only incorporated the representation of the

" n e w " sämans, viz., sämans we have not yet met with. The
other representations have already been provided. Thus far, the
survey of these structures conveys an idea of the distribution of
a specific chant structure or melody throughout many sections of
the Agnicayana ritual. This structure is like a musical theme that
appears and reappears, with variations, at many important
junctures of the ceremony.
The second structural feature I wish to discuss occurs earlier
in the Flow of Milk chants: in the ten chants §18-§27(AG 1 1 . 1 -
10). I shall write out the first of these in full, to clearly exhibit its
structure:

§18. hä bu (3 x ) aham annam (3,x) aham annâdo (3 x ) aham

vidhärayo (2 x ) aham vidhärayah I hä bu (3 x ) y ad
varco hiranyasya I yad vä varco gavâm uta I satyasya
brahmano varcah I tenamâsam srjâmasâ yi I hä bu
(3 x ) aham annam (3 x ) . . . vidhärayah (as at the
beginning) / hä bu hä bu hä vu vä I e aham annam aham
annädo aham vidhärayah (3 x ) e aham suvar jyotih I

This chant incorporates a mantra, yad varco . . . , from the

jaiminïya Àrcika (107.34) which also occurs in the Kauthuma-
Rânâyanïya tradition, but is neither found in the Rigveda nor in
the Yajurveda. The structure of the chant exhibits a special
feature that may be represented in a simple fashion if we make
use of indexed lower case letters to express elements, as follows:

a,: aham annam

a 2 : aham annâdah
a 3 : aham vidhärayah.
 The Syntax of Stobhas 219

The special feature of these chants is that the mantra y ad varco

. . . , which I shall refer to as " Y " , is preceded by the structure:

and followed by the structure:

(a, a2 a3)3.
This feature occurs in all the ten sämans, but the number of
elements need not always be three. Using the following abbre-
viations:

P: hä bu
P*: hä vu vä
T: e aham suvarjyotih,

the general structure of the ten sämans is expressed by:

P 3 a, 3 . . . a 3 P 3 Y P 3 a 3 . . . a 3 P P P* e(a, . . . a s ) 3 T.
We are now in a position to specify the number of elements
C ' i " ) , and the elements themselves, for each of the ten sämans,
as follows: •' . •

§18. i = 3 a, aham annam

a2 aham annädah
a3 aham vidhärayah
§19. i = 3 a, aham sahah
a 2 aham säsahih
a
3 aham säsahänah
§20. i = 1 a, aham varcah
§21. i = 1 a
i aham tejah
§22. i = 4 a, manojait
a2 hrdayamajait
a 3 indrojait
a 4 . aham ajaisam
§23. i = 4 ai . disanduhe
a2 : disauduhe
220 Mantras

a3: disoduhe
a4: sarväduhe
§24. i = 1 a,: vayo vayo vayah

This could alternatively be expressed as:

i = 3 a,: vayah
a 2: vayah
a3: vayah

§25. Same as §24, but with rüpam instead of vayah

§26. i = 4 a,: udapaptam
ürddhonabhäm syakrsi
vyadyauksam
atatanam

An irregularity here is that P is: hi hi yâ au.

§27. i = 2 a,: prat he

a2: pratyasthâm

This chapter has dealt with only a few of the structures of the
stobhas of the Sämaveda. There are many more types and none
of them have been studied. I have three concluding remarks. The
first is the most obvious: these chants exhibit structures that are
sometimes similar to musical structures, but almost always
totally different from anything found in the syntax of natural
languages.
My second remark relates to the psychology of the chanters.
All these chants are transmitted orally and learned by heart,
together with their order, distribution, interrelationships, and
ritual applications and uses. Such an astonishing feat of memo-
rization can only be accounted for by assuming that such
abstract structures as we have postulated and expressed by
symbolic formulas are actually represented, in some form or
other, in the minds or brains of the chanters. This reflects the
 The Syntax of Stobhas 221

obvious fact that it is possible to remember such vast amounts of

material only because of implicit, underlying regularities.
My third concluding remark relates to the significance of these
chants. We have witnessed, even in this relatively small sample,
many strange forms, strange from a linguistic point of view, and
also strange for anyone who is looking for meaning, especially
"religious m e a n i n g / ' It should be obvious that language or
religion are not proper categories within which to evaluate the
significance of these ritual chants. Rather, their significance lies
in the structure and composition of the resulting edifice, and the
abstract structural qualities that we have represented by formu-
las. If there are anywhere structures similar to these ritual
features, it is in the realm of music. This is not so merely because
the Sämaveda may be described as " m a n t r a s set to m u s i c . "
What is more significant is that the structure of these chants,
both internally and in relation to each other, corresponds to
musical structure. Close parallels to these structures are found,
for example, in the complex expressions of polyphonic music in
Europe during the eighteenth century. The ritual chants of the
Agnicayana resemble in this respect the arias of Bach's orato-
rios, and are similar in character: their language is uninteresting,
their poetry mediocre, and their meaning trite; but the sounds,
with their themes and variations, inversions, interpolations, and
counterpoint, and the particular distributions of their elements is
what makes them remarkable. To those who have grown up in
such a tradition, and who have learned to perceive and appre-
ciate it in its traditional perspective, it is the structure of these
chants that reveals to a large extent what is felt to be their
beauty.
 20

Vedic and Tantric Mantras

Having gained some idea of Vedic mantras, we are now in a

position to discuss other types of mantras. The first that require
attention are the mantras of Tantrism: for although mantras are
important throughout the history of the Indian tradition, with
Tantric ritual they move once again to the center stage. I shall
primarily discuss Tantric mantras from Hinduism, and occasion-
ally refer to their Buddhist counterparts. As for Jaina mantras,
the most important is a set of seven that developed from ritual
salutations and refuge seeking formulas (see Jaini 1979:296-297).
Several scholars have attempted to characterize the difference
between Vedic and Tantric mantras. One difficulty is that the
latter have not been studied as well as the former and are only
now beginning to receive scholarly attention. However, some of
these comparisons have failed because even Vedic mantras have
been inappropriately described. Wheelock (1988:118), for exam-
ple, has written that " t h e Vedic mantra truthfully describes and
thereby actualizes a bandhu between ritual object and cosmic
e n t i t y , " and that the Vedic mantra " s t a n d s as a means to the
ends of the sacrifice. The Tantric mantra, on the other hand,
as the essence of the ritual procedure, is an object of value in
itself.
It is clear that these expressions are not applicable to any of
the mantras we have considered. Wheelock's terms are obvi-
ously inspired by the Brähmana literature and not products of
his own fancy. However, that does not make them any more
relevant. Brähmana interpretations are more fanciful than any-
thing contemporary scholars have yet come up with. Most
mantras, for example, do not describe, nor d o . t h e y refer to
224 Mantras

cosmic entities. The further we proceed along the six entries of

our list, the more obvious it becomes that mantras are ends in
themselves. The Udgätä continues to chant long after the golden
man has been laid down. There are no specific ritual acts with
which any of these mantras are individually associated, and
which could explain their occurrence—just as there are no
events in the life of Christ that explain any bars or themes in the
C Major Aria "GeduldV- for Tenor and Cello from Bach's St.
Matthew Passion. Vedic and Tantric mantras, therefore, are not
different in terms of the characteristics attributed to them by
Wheelock.
According to Padoux (1963: 296), Saivite mantras are different
from Vedic mantras because a Vedic mantra is essentially a
verse or a group of verses: " u n verset ou un groupe de v e r s e t s . "
However, as we have seen, this is applicable only to the textual
sources of some Vedic mantras. It does not apply to prose
mantras, to s tob has, or to any of the numerous sounds and
noises that pervade the other ritual uses of the Vedas. Moreover,
even if a Vedic mantra seems to be a verse, in its ritual use it is
not treated like a verse at all. It is treated in the same manner as
other sound sequences that never were verses to begin with. The
counting of syllables that features in the ritual use of indra
jusasva . . . (Chapters 17 and 18) is not similar to the counting of
syllables that we find in true versification; it is similar to the
counting of syllables that is applied to stobhas, and is typical of
their ritual use. Even if stobhas are interpreted, as in Chändogya
Upanisad 1.13.1-4, the interpretations should not be takem
symbolically (as was done by the philosopher Sarikara in his
commentary on this passage), but should themselves be ex-
plained in terms of syllable counting (see Faddegon 1927, and
Gren-Eklund 1978-1979). In other words, in all these mantras,
language, whether versified or not, is not treated in the manner
ordinary language ever is. Vedic and Tantric mantras can
therefore not be different on account of the fact that Vedic
mantras are "in v e r s e . "
A functional difference between Vedic and Tantric mantras
may seem to be that the latter are used not only in ritual, but also
 Vedic and Tant rie Mantras 225

in meditation. But meditation is not so different from ritual as is

sometimes assumed. It is also alluded to in the Vedas (see, e.g.,
Staal 1975:79). Moreover, a characteristic of meditation,
namely, that it is silent, is also applicable to the ritual use of
mantras. Both Padoux and Wheelock have emphasized the silent
use of mantras in Tantric ritual. But silence plays a very
important role in Vedic mantras also. Many Vedic mantras are
anirukta, " n o t e n u n c i a t e d , " upämsu, " i n a u d i b l e , " or are re-
cited manasâ, " m e n t a l l y . " The brahman priest is in principle
always silent. Though all the deities " l o v e what is out of sight"
{paroksapriyä deväh), Prajäpati is the one who has a special
preference for silent mantras and silence (perhaps because he
was not an Aryan deity, and most mantras are Aryan imports).
True, Rigveda 10.95.1 says (see Findly 1988:26):
kk
If these mantras of ours remain unspoken
they will bring no joy, even on the most distant day."

But in their ritual use, Vedic mantras are often silent, that is,
objects of meditation, just as they are in Tantrism.
In terms of the characterizations so far mentioned, it is not
easy to make a clear distinction between Vedic and Tantric
mantras. Actually, some of the latter are Vedic, and it is likely
that, with further study, more Vedic sources for Tantric mantras
will be found. Varenne (1962) discovered that in the Tantric rite
of kälanyäsa, "touching the body (of the deity or the officiant
who has become the d e i t y ) , " Vedic mantras from a relatively
late Vedic text, the Taittirïya Àranyaka, are used. The Vedic
mantras are:

ïsanah sarvavidyàncim ïsvanih sarvablultânâm brahmädhipatir

brahmano'dhipatirbnihrnasivomeastusadusivah
"Ruler of all knowledges, lord of all beings, brahman master, master of
brahman, Brahma, be favorable to me, the Always-Favorable!"

In the Tantric rite, various elements have been inserted and

" m a n t r i c i z a t i o n " has increased as follows:
om horn I ïsânah sarvavidyänäm sasinyai namah I ürdhvaniürdhni I
om horn I isvarah sarvabhütänäm angadüyai namah I pürvamürdhni I
226 Mantras

om hom I brahmâdhipatir brahmano'dhipatir brahmä I istäyai namah I

daksinamürdhni I
om hom I sivo me astu marïcyai namah I uttaramurdhni I
om horn I sadäsivom I jvälinyai namah I pascimürdhni I

The resulting sequence is again difficult to translate, that is, to

make sense of in terms of its lexical meaning. In terms of
structure, however, it is clear. Obviously, the original mantra
has been cut in five pieces, each of them introduced by the same
bïja mantra or stobha: om horn. Then, five items are inserted and
saluted (. . . namah), and each piece ends with a direction which
specifies how the head is to be touched:

To Sasinï—on top of the head;

to Arigadâ (?)—on the east (front) of the head;
to Istä (?)—on the south (right) of the head;
to Marïcï—on the north (left) of the head;
to Jvâlinl—on the west (back) of the head.

It is not uncommon for "stage directions" to become parts of

mantras. An example is the mantra susravah, " r e n o w n , " which
accompanies the giving of a staflf to a brahmin boy during his
Upanayana or initiation. The sütra text specifies this:

susravah . . . iti dandam dadhyät

"with (the recitation of) 'susravah . . .'he should give the staff."

When I filmed this ritual in Madras in the early sixties, the priest
recited the entire text susravah . . . iti dandam dadhyät as if it
were a mantra. On another occasion, when the text specified
other mantras that had to accompany a rite, the priest recited the
mantras immediately followed by the "stage direction" formu-
lated by the text: iti manträbhyäm, "with these m a n t r a s . "
The subdivision of the mantra into five pieces that accompany
the touching of five parts of the body is reminiscent of the five,
layers of the Agnicayana altar which Ikari (1975) has shown to be
equivalent to the five kos as or " s h e a t h s " of the Taittiriya
Upanisad and which recur in other quintuples. Touching parts of
the body is not uncommon in Vedic ritual: parts of the body of
 Vedic and Tantric Mantras 227

the boy who receives the Upanayana initiation just mentioned

are also touched by the priest.
One type of Tantric mantras is best known in the West or at
least its counterculture: \\\t-bîja or " s e e d " mantras which often
consist of single syllables, frequently vowels or diphthongs
followed by a nasal. We have already met with om and horn, two
syllables that answer this description. There are striking simi-
larities between these seed mantras and the Vedic stobhas
exemplified in the sixth example of Chapter 17 and in Chapter
19. Whether these similarities exist throughout the Tantric
domain is difficult to determine because so few Tantric mantras
have been studied. On the Vedic side the documentation is good.
Apart from Maurice Bloomfield's monumental Vedic Concor-
dance of 1906 (reprinted in India in 1964), we have in the domain
of stobhas a collection called Stobhânusamhâra, published by
Satyavrata Sämäsramin in the Bibliotheca Indica (Volume II,
1874, pages 519-542) and made accessible by van der Hoogt
(1929). This collection contains such stobhas as:

ä (e)re hâ-u is phat

as hâ hm it pnya
auhovâ hahas ho-i kähvau urn
bhä hai hum kit up
dada hâ-i hup mrs vava
(e)br ham hvau nam vo-i
(e)râ has ihi om

The stobha dada inspired Faddegon to coin the felicitous expres-

sion "Ritualistic D a d a i s m " (Faddegon 1927; cf. Gren-Eklund
1978-1979).
In Tantric seed mantras we find similar or identical forms.
They often occur in groups of five, a number we meet every-
where: the Danish missionary Ziegenbalg recorded them in South
India in the early eighteenth century (Caland 1926:108-109) and the
"Five Syllable Mantra" (pancâksara) Namas Sivâya is found
throughout Asia (for Bali, for example, see Goris 1926:62-63).
228 Mantras

These seed mantras often reflect phonological discoveries of the

Indian grammarians. The following quintuple illustrates the five
semi-vowels of Sanskrit phonology disguised as mantras and
incorporated in a mystic diagram, the Sri Cakra (see Michael 1986:
134):

ham
yam
va m
ram
lam

Elsewhere, we find the vowels and diphthongs of Sanskrit

developed into a system of mantras, e.g. (Brunner 1986a: 102):

ham ham
him him
hum hum
hem haim
horn haum

Such " p h o n o l o g i c a l " systems of mantras are neither rare nor

confined to India. Similar mantras occur in the 1986 volume on
Mandalas and Mantras, edited by Padoux, in which the Michael
article just quoted is published. The entire Sanskrit alphabet
becomes a system of mantras and is found on representations of
deities such as Hanumän, whose limbs they cover. Outside
India, similar systems occur throughout Asia wherever Tantric
influences have gone: Central Asia, China, Korea and Japan,
and the countries of Southeast Asia. Numerous examples are
found, for example, in H o o y k a a s ' monographs on Balinese ritual
and in Payne's dissertation on the Japanese goma ceremony
(1986).
In Tantrism, mantras are placed in diagrams and circles,
mandalas or yantras, like those studied in the 1986 collection
edited by Padoux, in Understanding Mantras (Alper 1988), and
in earlier publications such as Zimmer 1926 and Pott 1966. I have
 Vedic and Tantric Mantras 229

referred to them elsewhere in a cosmological context (see Staal

1988d). For the distinction between mandala and yantra in the
Saiva tradition, see Brunner 1986a. For illustrations of yantras in
Bali in the form of simple drawings of the human figure with the
mantras added, see Hooykaas 1983:541-548.
A characteristically Tantric doctrine is that mantras and
deities are identical. This is referred to in passing by Goudriaan
(1981:130), but emphasized by Hélène Brunner (1986b: 129) who
has advanced the Hypothesis that the point of departure for this
idea lies in the ähävana or invocation rite. The argument runs, in
rough approximation, thus. The Tantric ritualists generally as-
sume that the deity is present as soon as the mantra that invokes
it has been recited; the two thus being co-extensive, the notion
arose that the deity was already present in the mantra. These
ideas are not confined to Tantrism, except the strict identity of
mantras and deities. The notion that deities are called down by
mantras, songs or invocations, and that they heed those calls, is
a common interpretation of Vedic rites and is frequently ex-
pressed in the Vedas themselves. For example, Rigveda 1.6.9:
atah parijmann à gahi divo va rocanâd adhi
sam asminn rnjate girah
"Come down, you who are always going around, from there or from the
divine realm of light! The songs propitiate you around!"

or Rigveda 8.1.18:
adha jmo adha va divo brhato rocanâd adhi
ay à vardhasva îanvà g ira mama 'jätä sukrato prna
"Come here from earth or from the large divine realm of light!
Grow your body by these songs of mine, complete your descendents,
you skillful one!"

or Sämaveda 1.1.1:
agna a yàhi vît ay e grnüno havyadâtaye
"Come here, Agni, to the feast, after being extolled, come to the gift of
offerings!"

The connection between these invocations and the making of

oblations or offerings (expressed by two identical verbal roots
230 Mantras

hu-), led Thieme to the hypothesis that Vedic ritual originated in

a Gastmahl, a feast offered to a guest. I shall return to this theory
which has been defended with ingenuity but does not account for
the ritualization that characterizes ritual (page 438).
The Vedas come tantalizingly close to the further step^ that
the invocatory mantras are deities. The Mîmâmsâ philosophers
assert that the deities are entities postulated in order to give a
semblance of existence to what is referred to in a mantra by a
name. According to Mïmâmsâ Sütra 9.1.9, ritual is the principal
factor ipradhâna) and the deity a subordinate factor (guna). The
commentary by Sabara on this passage shows, however, that the
idea of deity in Vedic ritual is in some respects different from
what it became in Tantrism. Ganganatha Jha, who translated this
commentary (Jha 1936,111:1432-1437), has summarized it in the
following terms (Jha 1942:336-337):

The deity and the offering material are both accomplished entities, while •
the act of ritual is what is to be accomplished. The deity, therefore,
cannot be the prompter . . . The view of the opponent makes it necessary
to admit of deities having material bodies and actually eating and
drinking the substances offered; and this idea is utterly repugnant to the
Veda, which does not lend support to any such idea regarding deities.
The text quoted by the opponent regarding the right hand of Indra ("we
have taken hold of what is Indra's right hand") [Cf. the thighs of Indra,
the arms of the Asvins, the hands of Püsan in examples (2) and (4) of
Chapter 17] does not mean that Indra possesses a right hand. Moreover,
even if he did, it would be impossible for any human being to "take
hold" of it. . . . The texts that speak of the "arms" of Indra being hairy
or his "eyes" as tawny—all these are purely eulogistic. Nor is there
actual feeding or eating, at rituals; in fact, the deity never eats. . . .

Hélène Brunner made a special study of the limbs of Siva that

are touched in the Tantric kâlanyâsa rite. Touching the deity is
equivalent to touching o n e ' s body in accordance with a principle
frequently invoked in Tantric texts:

yathâ deve tathci dehe cintayet tu vicaksanah

"the intelligent person must realize in his own body {dehe) what he
realizes in the divine (image)ideveY"
 \ e die and Tantric Mantras 231

(Brünner 1986a: 100). There also exists a close correspondence

between the limbs of the divine body arid the mCdamantra or
" r o o t m a n t r a ' ' w h i c h invokes the deity. This mantra is divided
into parts that are called limbs (anga). Brunner has referred to a
passage of a late Vedic text, attached to the Äsvaläyana Grhya
Sütra, which refers to nyâsa and invokes the famous Gäyatri
mantra, taught at Upanayana and recited daily by almost every
brahman:

tat savitur varenycun

bhargo devasya dhimahi
dhiyo yo nah pracodayät
"May we receive this desirable
light of the god Savitr,
who shall impell our thoughts"

(Rigveda 3.62.10). The Äsvaläyana text subdivides the mantra

into pieces called anga, each of four syllables and connected
with parts öf the body. For other subdivisions and insertions see
Varenne 1962:186, who refers to Brhad Äranyaka Upanisad
6.3.6.
Brunner has quoted several Tantric texts that support the
conclusion that the limbs of the deity are set up following the
division of the mantra and not the other way round. This
suggests that the Tantric deities themselves originated at least in
part from mantras. Such a hypothesis would explain the notion,
often met with in popular books, that there are millions of gods
in Hinduism. For enormous quantities of mantras are imagined
to exist in superior worlds (70.000.000, for example: Brunner
1986a:91 and note 9). That gods come from language is a variant
of the Wittgensteinian doctrine adumbrated by Max Müller, that
mythology and philosophy are diseases of language. We have
watered down Western versions in personified allegories, for
example, Victory, Melancholy or the Muses, sometimes solidi-
fied as in the Statue of Liberty, an idol as welcome in Tantrism
as it would be unwelcome in Islam.
The Tantric rite of nyäsa, ''imposing a m a n t r a " (on the body:
Padoux 1963, Index) expresses primarily a system of mantras so
232 Mantras

that it is misleading to characterize it as a "ritual projection of

divinities into various parts of the b o d y " (Eliade 1958:210-211).
The idea is connected with the more spectacular aspects of
Tantrism dealt with in popular literature, especially its emphasis
on sex, which is not absent from the Vedas either (see, e.g.,
Brhad Äranyaka Upanisad 6.4).
In Vedic, meaningless syllables such as the Tantric " s e e d "
mantras are not confined to the Sämaveda but occur also
elsewhere. In the sastra recitations of the Rigveda, for example,
there are lengthened " o " 's (see Chapter 9B: Episode I, Ele-
ment 3). They often end in a nasal which results in the famous
mantra OM.There are also insertions of somsävo which means
" L e t us both recite!" but is treated as a similarly rnçaningless
element with various forms (e.g., sosomsâvo.) The Adhvaryu
responds with othämo daiva, äthämo daiva, othämo daiva
made, modâmo daivotho, and other "bizarres contortions litur-
g i q u e s , " as Caland and Henry (1906, 1:232, note 8) called them.
In the Àsvalâyana tradition of the Rigveda, the Hotä priest
murmurs before the beginning of his first sastra:
su mat pad vag de
(Caland-Henry, 1:231). Each sastra recitation moreover, has its
own syntactic structures which have nothing to do with seman-
tics. During the noctural rounds in the more advanced Soma
rituals, for example, the first quarter verse is repeated in the first
round, the second in the second, and the third in the third (see
A G N I , 1:663-680, 11:750-752, and above page 182).
The Sämaveda is different only in the sense that the choreog-
raphy of mantras becomes richer and more varied. Most of the
ritual chants consist of elaborate structures, generally preceded
by: ö him, and certain sequences by: him. The patterns become
so complex, that the priests keep track of them by constructing
figures, called vistuti, with the help of sticks on a piece of cloth
(for illustrations see A G N I , I:Figures 48-51). In many melodies
(called gciyatra) the udgltha or second portion of the chant, sung
by the Udgätä, is:
ö vä ö vä ö va him bhâ ö vä.
 Vedic and Tantric Mantras 233

In musical chants, the occurrence of such sounds is of course

e surprising. Their function is simply to fill out the melody
when there is no text. The same holds for insertions such as o or
somsävo. This is found all over the world. The Kosi of Camer-
oun insert whistling in their chants (Dammann 1963:117), but this
is exceptional. In general, melodic insertions are induced by the
phonological structure of the language in which they are in-
serted. For example, heisa hopsasa would not fit in a Vedic or
Sanskrit context, but fits in German when sung by Papageno in
Mozart's Zauberflöte'. ,
Der Vogelfänger bin ich ja,
stets lustig, heisa hopsasa!

Heisa hopsasa is reminiscent of the kind of sounds one would

use, in German, when addressing a horse or a pack animal. It
would be helpful to know what sounds the Vedic Indians used in
everyday life in similar circumstances. Such information would
not assist in explaining the ritual meaning or use of mantras, but
it would throw light on their origins and on the associations they
may have evoked in ancient India.
Before closing this section I should make brief references to
non-Tantric forms of Hinduism and to Buddhism. It is often
assumed that Tantric mantras are different from the other
mantras of medieval Hinduism. However, there are similarities.
The so-called Purânic mantras, or mantras prescribed in the
Puränas, are a case in point. Whereas they are literally mean-
ingful, unlike the Tantric /?//a-mantras, they are treated as if they
were devoid of meaning. This is shown by the fact that the
mantras
namah siväya, ''homage to Siva"
om namah siväya, "OM! Homage to Siva"
om namo nârâyanâya,klQM\ Homage to Näräyana"
om namo bhagayatëvasuaevâya, "OM! Homage to Lord Väsudeva"
srirâmajayarâmajayajay anima, "(long) live Sri Räma, live Râma, Rama
live!"
are not distinguished from each other (as scholars are likely to
assume) by the different deities to which they refer, or by their
234 Mantras

meanings, but by the fact that these mantras are, respectively,

five-syllabic (pancàksara), six-syllabic (sadaksara), eight-syl-
labic, twelve-syllabic, thirteen-syllabic, etc. (Kane V, 1958,
1962, notes 219 and 1775). Just like Vedic and Tantric mantras,
these Puränic mantras are treated not like utterance of language
but as if their main characteristic were the number of their
syllables. This is both characteristically Indian and characteris-
tically " ' m a n t r a . "
In Buddhism, the term most frequently used for mantra is
dhâranï. The theoretical distinctions made by the Yogäcära
philosophers are especially reminiscent of Vedic notions. They
distinguish artha-dhâranï, "meaning(ful)-memorizations,"
which consist of words and phonemes, not yet formulated or
even expressed mentally, from mantra-dhâranï, which are sim-
ilar but more effective: Bodhisattvas use the latter to alleviate
the afflictions of beings. This distinction is related to that
between dhâranï and samcidhi or " c o n c e n t r a t i o n " : whereas the
latter is always associated with thinking {cittasamprayiikta), the
former, according to these theorists, may be associated with
thinking or dissociated from it (cittaviprayukta). In other words,
some dhâranï are meaningful, others meaningless; but all are
treated similarly and belong to the same category (Lamotte IV,
1976:1857-1859). This speculation is compatible with the Vedic
and Tantric uses of mantras which is also characterized by the
fact that they are independent of the distinction between mean-
ingful and meaningless.
The idea that mantras are meaningless goes back to the Vedic
ritualist Kautsa. Both the Nirukta (1.15), an early work on
etymology from the Vedic period, and the Mïmâmsa Sütra
(1.2.31-39), refer to his doctrine that " m a n t r a s are meaningless"
(anarthakä mantrâh: see below page 373 and Staal 1967: 24-26,
45-47). This view has always remained the view of a minority,
for most Indian commentators and philosophers have tried to
provide mantras with meaning even if it meant invoking the
improbable or the impossible. In the Brâhmanas we have found
ad hoc interpretations, contradictions and rationalizations. In
the later literatures of Hinduism and Buddhism such rationaliza-
 V e die and Jantric Mantras 235

tions develop, and tend to become more systematic. In Buddhist

philosophy, a distinction is made between ''explicit meaning"
{nitärtha, Tibetan: hes don) and 'implicit meaning" {neyârtha,
Tibetan: drah don; see, e.g., Murti 1954:254; Ruegg 1969:56;
1973:58). In Buddhist Tantrism this developed into full-fledged
systems of " h e r m e n e u t i c s . "
Such systems and concepts derive from metaphysics and are
not directly concerned with the practice of mantras. Steinkellner
(1978) studied one such system, due perhaps to Candrakïrti,
which distinguishes one literal and three " T a n t r i c " meanings.
This system formed the basis for the Guhyasamâja school and
was adopted by Indian and Tibetan exegetes from the eighth
century onward. It should be noted that, as in the case of the
Brähmanas, nothing is authoritative about such interpretations.
They are the predictable professional views of philosophers,
theologians, and exegetes all over the world. They need not be
taken seriously as possible explanations because they them-
selves stand in need of an explanation. They do not throw light
on the nature of mantras.
There are more important kinds of evidence that have to be
taken into account before we can conceptualize, explain or even
picture the history of the Indian mantra from Veda to Tantra,
Hindu as well as Buddhist. Foremost among these are tech-
niques of chanting and recitation in the context of which many
mantras developed. The relevance of such evidence is clear in
the case of the Sämavedic stobhas, which can only be under-
stood within the context of the chants and melodies (säman) of
the S â m a v e d a ( s e ê , e.g., Staal 1961, Chapter VIII). For Buddhist
chants, Paul Demiéville has collected the relevant facts in two
articles, published with an interval of half a century between
them (Demiéville 1930 and 1980). The evidence from chant and
recitation (or " h y m n o l o g y " , in the words of Demiéville) is far
too rich and varied to be taken into account in the present
context; but it demonstrates, among other things, the impor-
tance of musical categories for explaining some of the charac-
teristics that distinguish mantras from language. The close
relationship between mantras and music reflects the general
236 Mantras

relationship between ritual and music (see Chapter 16). All we

can do in the present context is emphasize that mantras cannot
be understood unless their musical character is taken into
account. This constitutes one reason why mantras cannot be
explained in tefms of language.
It remains a curious fact that monosyllabic mantras of the
siobha type reemerged in Tantrism after apparently lying dor-
mant for more than a millenium. Their popularity stands in need
of an explanation, not their occurrence somewhere on the
subcontinent. For traditions of Sâmaveda chanting have been
handed down without interruption from Vedic times, and con-
tinue to the present day. Knowledgeable Sämavedins have
always been rare, secluded, orthodox, and reluctant to divulge
their art; but we need only assume that one became a Tantric or
Buddhist and chanted stobhas for the edification or entertain-
ment of his fellow sädhakas or monks. Though controversial,
this would not be unheard of. The Buddha himself had on several
occasions asked a young novice with a beautiful voice to come
to his cell at night and chant (Demiéville 1930:93). An opportu-
nity for transmission in centers of Vedic and Buddhist culture,
such as Banaras or Kanchipuram, was always available. That
such mantras found their way into meditation is not surprising,
especially in Buddhist monasteries. Their subsequent diffusion
is known as the spread of Tantrism in India and over large parts
of Asia.
 21

Performatives, Pragmatics and

Performance

In Sanskrit, the performative force of an entity is called its

siddhi. This is one of those formations that exemplify the
productivity of Sanskrit morphology. The root sidh-, the weak
form of sâdh-, means "establish, a c c o m p l i s h . " Therefore,
siddha means " e s t a b l i s h e d " and siddhänta the finally estab-
lished view in philosophy and in the traditional sciences in
general. Sädhana is the means by which a result is established:
for example, ritual practice interpreted as a means. Sädhaka is
the practitioner or follower, sometimes wrongly translated into
English (because of the similarity in sound?) as " S e e k e r " :
erroneous, because the Indian sâdhaka does not seek but follow.
Another related term is sädhya, the thing-to-be-established, that
is,Jn logic, the conclusion. Finally, siddhi becomes the supreme
performance: the Siddhas are the,,happy few who have obtained
it and become Members of the Spiritual Establishment.

21 A, Performatives Again

What I propose to do in this chapter has nothing to do with

Siddhas but deals with the Western categories of "performa-
t i v e s " and " p e r f o r m a n c e , " and the discipline of " p r a g m a t i c s "
which has a similar intermediate status. We have seen (Chapter
3) that Austin introduced performative utterances in an attempt
to go beyond the customary logical categorization of utterances
as statements that are either true or false. This is not quite
correct, as Quine pointed out, but these utterances of language
238 Muniras

are in any case utterances that in saying something also establish

something.
Tambiah made the most important contribution to the theory
of ritual in the seventies by giving grounds for abandoning the
semantic approach and advocating an analysis in terms of rules
(Chapter 15). Tambiah looked upon rites as things that do not
mean but do something. This is different from Austin in two
respects: Tambiah did not claim that rites do something by
meaning something; and he did not state that for rites to do
something is not surprising unless it is by saying or meaning
something. Rites, after all, are acts. In India, they have always
been called karmäni, which means " a c t s . " The ancient Greeks
also called them ôpwjjieva {dromend), " a c t i v i t i e s , " which corre-
sponds to Latin act a. Negatively, the view that Tambiah de-
fended was salutary as a corrective to the over-emphasis on
symbols and meaning that had begun to invade the study of
ritual, in anthropology as well as the study of religion. Posi-
tively, it constituted a return to earlier views and common sense
on the one hand, and made an important innovation on the other.
The scholars who defended the do ut des doctrine, for example,
had always interpreted ritual as consisting of acts and not of
utterances. But Tambiah was the first to establish that the
exclusive emphasis on meaning had become an obstacle to an
adequate understanding of ritual, and that ritual acts had to be
understood in terms of rules.
In dealing with mantras, Wheelock (1982, 1987) defended
views that are in some respects similar to Tambiah's because he
also took his inspiration from the performatives explored by
Austin and further analysed by Searle. The view he espoused,
that mantras are speech acts, was more radical than the view
that rites are performative acts. For although it was generally
agreed—albeit tacitly—that mantras belong to the realm of
speech, it had not been claimed before that they could be looked
upon as acts. It is to be noted, however, that this view is not
supported by the Indian tradition which had made a basic and
clear distinction between acts and mantras, as we have seen in
the Introductory Note to Part III. In Sanskrit, a mantra is never
 Performativesy Pragmatics and Performance 239

called an act (i.e., karman or kriyci). The two are distinguished in

order that a one-one-correspondence between them can be
established.
Wheelock began his discussion with Austin's distinction be-
tween locutionary, illocutionary and perlocutionary acts, and
concentrated on Searle's taxonomy of illocutionary acts as
" p e r h a p s the most significant advance over Austin's primitive
classification''(Wheelock 1982:54). In order to clarify this we
shall slightly modify Austin's original formulation into saying
that speech acts have three kinds of force, the locutionary, the
illocutionary, and the perlocutionary. The illocutionary force of
a speech act is Concerned with the effect the speaker intends to
produce in the hearer. This was developed by Paul Grice—
subsequently by S. Y. Kuroda—and led to Searle's classifica-
tions of speech acts based on the assumption that all speech acts
are concerned with such effects, viz., with intention.
Adopting Searle's classification, Wheelock pointed out that
there are several basic differences between "ritual speech a c t s "
and "ordinary speech a c t s . " For example, " t h e very basic
requirement that an ordinary speech event involves a speaker
and a hearer is one that is often lacking in ritual speech a c t s "
(Wheelock 1982:58). Also, " t h e most essential distinguishing
feature of ritual utterances is that they are speech acts that
convey little or no information" (ibid.). Wheelock referred, with
apparent approval, to Tambiah's view that " i n ritual, language
appears to be used in ways that violate the communication
function" (page 57).
Wheelock and Tambiah 1968 are undoubtedly right that man-
tras do not always require a speaker and a hearer, are not always
communicative and do not necessarily convey information. In
fact, they rarely or never do any of these things. However, that
mantras cannot be speech acts follows directly from Searle's
view, because on his view, all speech acts involve intention and
it is obvious from the six representative mantras illustrated in
Chapter 17, that most mantras do not involve intention. The
theory according to which they are performatives depends,
moreover, on the assumption that mantras are linguistic utter-
240 Mantras

ances. We have already seen that rites are not performatives

because they are not linguistic utterances. We shall see that
mantras are not linguistic utterances either and hence are neither
speech acts nor performatives.
The confusion to which Tambiah and Wheelock fell prey may
be summed up in the following terms: both adopted a philosoph-
ical distinction which was applicable to the area which logicians
and philosophers had long been discussing, that is, the area of
language and linguistic utterances. But since it had not been
established that rites and mantras are utterances of language, the
distinction failed to apply to these categories. Applying it to rites
did not make sense since it was already agreed that rites were
acts; and applying it to mantras was unpersuasive because the
Indian tradition consistently distinguished between mantras and
acts throughout the long period of its development.
We should be grateful that Indologists, anthropologists and
scholars of religion look beyond the boundaries of their disci-
plines and attempt to make use of concepts developed by
philosophers. It has not been my intention to criticize Tambiah
or Wheelock for misunderstanding or misapplying such con-
cepts, but merely to disentangle some of the confusions that
easily result. It is not uncommon for philosophers themselves to
misunderstand these concepts. Searle criticized Foucault for
claiming that statements were not speech acts, and Foucault
replied in a letter to Searle of May 15, 1979: " A s to the analysis
of speech acts, I am in complete agreement with your remarks.
I was wrong in saying that statements were not speech acts, but
in doing so I wanted to underline the fact that I saw them under
a different angle than y o u r s " (Dreyfus and Rabinow 1983:46,
note 1).
Though mantras are not speech acts, Austin's ideas may
throw light on mantras in another respect. Austin was originally
interested in performatives, which he contrasted with constative
utterances. Later he arrived at the conclusion that all speech
acts exhibit both features or forces. Performatives are speech
acts that perform acts in saying something (e.g., promising or
baptizing). They cannot be false, but they can go wrong, or be
 Performativesy Pragmatics and Performance 241

fc
'unhappy." Austin formulated six conditions for the felicity of
performatives. The first four are:

(A.I) There must exist an accepted conventional procedure

having a certain conventional effect, that procedure
to include the uttering of certain words by certain
persons in certain circumstances, and further,
(A.2) the particular persons and circumstances in a given
case must be appropriate for the invocation of the
particular procedure invoked.
(B.I) The procedure must be executed by all participants
both correctly and
(B.2) completely (Austin 1962:14-15).

It is clear from what has been said before that mantras are not
performatives: they do not perform acts, and need not say
anything. However, their use is governed by conditions that are
similar to Austin's four conditions. The chief differences are that
mantras need not have an effect, or a visible effect (the Mimâmsâ
thinkers devoted much discussion to their adrsta, ''invisible",
effects); what is uttered need not be words; and there need not
be more than one person uttering a mantra. It is certainly a
necessary condition for the use of mantras, on the other hand,
that only the appropriate person can properly use them (e.g., the
Adhvaryu priest: see above page 72). In general, only brahmins
can utter or hear Vedic mantras. Within a given ritual perfor-
mance, only the appropriate priest can use the prescribed
mantras at the proper place and time. In order to be able to
discharge this priestly function, a person has to be eligible, and
elected beforehand. The election of priests constitutes a special
ceremony (rtvigvarana) that takes place at the beginning of a
ritual performance (AGNI, 1:313-316).
While Austin emphasized, in his illustrations, the appropriate-
ness of the speaker (e.g., a bridegroom saying " Y e s , I will," or
a person naming a ship), Indian theorists have been equally
concerned about persons hearing or receiving mantras as about
those who recite or give them. The restrictions in Veda and
242 Mantras

Tantra are similar. N o mantras may be learned from books.

They can only be learned, at the appropriate time, by eligible
students from eligible teachers. Members of low castes, or
people beyond the pale of caste (such as outcastes or foreigners)
may be punished (e.g., by having molten lead poured into their
ears) for hearing Vedic mantras even inadvertently. Among
Vedic brahmins, other restrictions obtain. The few Sämavedins
of Kerala, for example, will not teach their mantras to Rigve-
dins, thereby further endangering the continued existence of
their own Veda. In Tantrism (as in Maharishfs ' T r a n s c e n d e n t a l
Meditation"), a person is given his own mantra, and is not
supposed to divulge it at any time.
All such conditions are similar to those formulated by Aus-
tin—only they go further. Mantras should be pronounced cor-
rectly and completely; but they should in addition be recited
with the correct degree of loudness, at the correct pitch, and at
the correct pace (Äpastamba Srautasütra 24.1.8-15: below,
page 355). Moreover they, or their specifically prescribed por-
tions (e.g., bhakti in the Sämaveda), should be recited in a single
breath (see A G N I , 1:311, 602, 622). All such requirements that
govern the use of mantras resemble the conditions formulated by
Austin, but they are more extensive and more stringent than
anything that applies to the normal use of a natural language
such as English or Sanskrit.

21B. Pragmatics

Performatives are a category of linguistic utterances that are

acts that perform something. Their analysis is therefore part of
semantics, the discipline that studies the relations between the
expressions of a language and their meanings—which may
include sense, reference, and whatever other category of mean-
ing philosophers, logicians, linguists or others have convincingly
postulated. Performatives do more than connect with " r e a l i t y "
through referring; they make an impact upon it. In so doing,
performatives may be said to go beyond semantics. There is
another area of studies that is also closely related to semantics
 Performatives, Pragmatics and Performance 243

without it being quite clear whether it should, or should not be

regarded as part of it: pragmatics. Although this field of studies
has never turned into a broadly recognized discipline, it ad-
dresses real problems. In such sentences as: " Y o u were here
y e s t e r d a y , " the truth of the sentence depends on who is 4 4 you,",
where is " h e r e , " and when was " y e s t e r d a y . " When these
variables vary, the truth value of the sentence may change. For
example, if I say " Y o u were here yesterday"*to Peter it may be
true while if I say the same tö Kazuo it may be false. Saying it
in different places at different times increases this variability.
Variables such as' " y o u , " " h e r e " or " y e s t e r d a y " have been
called "indexical t e r m s , " "egocentric particulars," or " t o k e n -
reflexives" (in the words of Peirce, Bertrand Russell, and Hans
Reichenbach, respectively). Carnap initially considered prag-
matics an empirical field of investigation (Carnap 1936-1937) but
subsequently dealt with its theoretical foundations (Carnap
1947), following which the field was developed as a formal
system by logicians such as H. K a m p , D. Lewis, R. M. Martin,
D. Scott, R. Stalnaker and especially Richard Montague (1968,
1970: reprinted in Thomason 1974). I related pragmatics to
Austin's performatives and attempted a linguistic analysis of one
of its characteristic features (Staal 1970a).
Pragmatic notions were current in classical antiquity as well as
in ancient India; they are undoubtedly here to stay. At the same
time, the status of pragmatics as a science has so far remained
undetermined; in fact, it may be argued that, outside the domain
of formal logic, pragmatics died in Jerusalem (Staal 1971).
Pragmatics is generally defined, following Morris and Monta-
gue, as the discipline that is not only concerned with the
relations between expressions (as is syntax), or between expres-
sions and their meanings (as is semantics), but with the relations
between expressions, their meanings, and their users or contexts
of use. Pragmatics may be relevant to the study of ritual and
mantras provided we extend the notion of " e x p r e s s i o n s " be-
yond language so as to include rites as well as mantras. This is
what Tambiah and Wheelbck did unsuccessfully, for mantras or
rites are not like linguistic expressions in possessing meaning or
244 Mantras

reference. It is appropriate in the present context, however,

because they are similar to linguistic expressions in that they are
being used. They involve, therefore, the specific features that
are the subject matter of pragmatics, namely, users and contexts
of use.
In the Vedic area pragmatic relationships are straightforward
since the users, executors or performers of rites and mantras are
invariably the same in almost all ritual contexts. Like the ritual
acts themselves, most ritual recitations are executed by the
Adhvaryu on behalf of the Yajamäna, although on specific
occasions other priests recite particular pieces. In Soma cere-
monies, the sastra recitations are always executed by the Hotâ,
Maiträvaruna, Brähmanäcchamsin or Acchäväka priests in ac-
cordance with specific rules. The ritual songs are always chanted
by the Udgätä, with a few exceptions, well-known from the
tradition (ç.g., the Subrahmanyâhvâna: AGNI 1:369, 386, 596),
while the stuti or stotra chants of the Soma ritual are always
executed in the same fashion by the trio of Prastotä, Udgätä and
Pratihartä. Were we to add a list of exceptions and special cases,
the information contained in the present paragraph, suitably
formulated, would therefore take care of pragmatic problems as
far as the users are concerned.
As for " c o n t e x t s of u s e , " the ritual context determines
throughout what is to be recited or chanted, as well as when and
on what occasion. The knowledge of these contexts of use is an
important part of what constitutes a knowledge of the ritual—the
knowledge that distinguishes a ritual specialist from a nonritual-
ist. But such " p r a g m a t i c " information depends on the structure
of the ritual itself, and cannot replace the latter.
A pragmatic analysis of the function of the users of the
mantras, i.e., of the priestly functions, may throw light on their
position in society outside the ritual enclosure. Such an analysis
pertains to historical reconstruction and is therefore a diachronic
enterprise, unlike the synchronie analysis that seeks to under-
stand the ritual as a system. One contemporary scholar who has
studied these historical problems fruitfully is the German San-
skritist Klaus Mylius. Mylius has, for example, analysed the
 Performatives, Pragmatics and Performance 245

offices of two Rigvedic priests who take part in the Soma

ceremonies, the Potä and the Acchäväka (Mylius 1977; 1982;
1986). His analysis led him to conclude that both these priests
had a lower status than the majority of the other officiants, but
that the Acchäväka occupied the lower status of the t w o ; he is in
fact related to the vaisya class. The Potä is referred to in the
Rigveda, where he is assigned menial j o b s ; the Acchäväka is of
later origin and he is not " e l e c t e d " - b u t " c a l l e d . " Yet his office
cannot have been created long after the Rigveda, as was, for
example, the office of the Grävastut.
It is clear from these examples that a pragmatic analysis may
be relevant to historical reconstruction. However, pragmatics
does not have the same function here as it has in linguistics or
philosophy, because it does not apply to individuals who deter-
mine the truth value of a statement, but to priestly functions
which are stereotypical roles.

21C. Performance and Competence

Turning now to " p e r f o r m a n c e " we are faced with another

concept taken from the confines of contemporary linguistics and
philosophy. "Performative S t u d i e s " have been applied to the
study of ritual and are also applicable to mantras. So far, this has
not shed much light, as we have seen (Chapters 15 and 21A).
Applying pragmatics makes more sense and yet, the results do
not contribute to the study of ritual systems (2IB). Applying
performance studies (which has nothing to do with "performa-
tives") has not shed much light either but this is not to say that
the concept of " p e r f o r m a n c e " is not applicable to ritual. Rituals,
after all, are performed, and it is their manner of performance in
which students of performance are interested. The lack of
success of these studies is due to the fact that they have adopted
a behavioristic or phenomenological perspective and confined
their attention to a feature of " a p p e a r a n c e " that cannot be
understood without paying attention to its underlying reasons,
causes, or " r e a l i t y . " For " p e r f o r m a n c e " is.based upon " c o m -
246 Mantras

p e t e n c e , " and may be described, but cannot be understood or

explained, without taking competence into account.
Without using these neologisms and in only slightly misleading
but simpler terms, we say that what people do is based upon
what they know. This knowledge need not be conscious, let
alone cohsciousJy formulated knowledge, but there must be an
internal structure that explains overt behavior, unless it is a
product of chance. It has accordingly become clear, especially in
linguistics, that the investigation of performance will proceed
only so far as the understanding of underlying competence
permits (see, e.g., Chomsky 1965:10-15). According to
Chomsky, any investigation of an activity such as language or
ritual

necessarily deals with performance, with what someone does under

specific circumstances . . . To the extent that we have an explicit theory
of competence, we can attempt to devise performance models to show
how this knowledge is put to use. If we knew only that language consists
of words, our performance models would necessarily be very primitive
and of restricted interest; we could study the sequence of linguistic signs
and their formal and semantic properties but nothing else. With L richer
theory of competence that incorporates structures of greater depth and
intricacy, we can proceed to more interesting performance models.
Study of performance relies essentially on advances in understanding of
competence. But since a competence theory must be incorporated in a
performance model, evidence about the actual organization of behavior
may prove crucial to advancing the theory of underlying competence
(Chomsky 1980:225-226).

The study of generative grammar as universal grammar has

proved so fruitful precisely because of this:

We saw that the study of generative grammar shifted the focus of

attention from actual or potential behavior and the products of behavior
to the system of knowledge that underlies the use and understanding of
language, and more deeply, to the innate endowment that makes it
possible for humans to attain such knowledge. The shift was from the
study of language regarded as an externalized object to the study of the
system of knowledge of language attained and internally represented in
the mind/brain. A generative grammar is not a set of statements about
externalized objects constructed in some manner. Rather, it purports to
 Performatives, Pragmatics and Performance 247

depict exactly what one knows when one knows a language: that is, what
has been learned, as supplemented by innate principles. Universal
grammar is a characterization of these innate, biologically determined
principles, which constitute one component of the human mind—the
language faculty (Chomsky 1986:24).

Performance Theory, at the time of writing an American favor-

ite, exhibits all the faults of the empiricist caricature of scientific
method we have studied in Chapter 2. At present it is another,
not interestingly different, form of behaviorism. It is applicable
to the study of ritual but only insofar as the understanding of
ritual competence permits.
In linguistics, or psycholinguistics, performance studies have
reached results that are nontrivial and even surprising. This is
due to two facts: (1) they have been undertaken in the context of
studies of competence; and (2) performance, which is by nature
finite, exhibits, in linguistics as in the study of ritual, a capacity
that is infinite. This explains why the inherent limitations of a
performative approach are especially clear in linguistics. A
simple example may serve as an illustration.
Imagine we wish to study the verbal forms of English from a
performance that exhibits English usage. Assume we have
collected a small corpus of data which includes the following six
forms:

he goes he went
he sees
he goes to see her he went to see her
he used to see her.

The pattern in which these forms have been printed exhibits a

structure which suggests that there should be two more forms to
complete it:

he saw
(*?) he uses to see her.
248 Mantras

Considerations of symmetry, however, are not decisive, for no

one has demonstrated that language must be in some sense,
" s y m m e t r i c a l . " The question arises: is the form that is marked
(*?) grammatical?—We may find the answer only by going
beyond this finite corpus of data, study other finite corpora of
data, and then go beyond all of them and postulate rules that
account for the formation of the past tense. It is likely that there
will be limiting conditions under which these rules operate, for
the rules of language are of a restrictive sort. This can not be
done, however, by restricting oneself to a theory of perfor-
mance. For such a theory adopts a behavioristic methodology
which refuses to go beyond the immediate data or admit that
performances may be different from each other and yet reflect
the same competence that is in people's minds.
We learn from data beyond our small corpus that " h e s a w " is
English; but we never corne across " h e uses to see h e r . " The
rule system must exhibit patterns of structure that explain this
and relate it to other structures. They may come under syntax,
semantics, phonology, or some other domain. It is conceivable
that we cannot find any explanations among these theories of
competence, and that the theory of performance offers an
explanation. Chomsky and Miller have found restraints of a
performative nature that prevent sentences that are perfectly
grammatical according to the model of linguistic competence
from being actually produced. Miller (1964:36) has discussed
three examples:

(R) Remarkable is the rapidity of the motion of the wing of

the hummingbird.
(L) The hummingbird's wing's motion's rapidity is remark-
able.
(E) The rapidity that the motion that the wing that the
hummingbird has has has is remarkable.

Miller comments:

When you parse these sentences you find that the phrase structure of (R)
dangles off to the right; each prepositional phrase hangs to the noun in
 Performatives, Pragmatics and Performance 249

the prepositional phrase preceding it. In (R), therefore, we see a type of

recurring construction that has been called right-branching. Sentence
(L), on the other hand, is left-branching; each possessive modifies the
possessive immediately following. Finally, (E) is an onion; it grows by
embedding sentences within sentences. Inside 'The rapidity is remark-
able" we first insert "the motion is rapid" by a syntactic transformation
that permits us to construct relative clauses, and so we obtain 'The
rapidity that the motion has is remarkable." Then we repeat the
transformation, this time inserting "the wing has motion" to obtain 'The
rapidity that the motion that the wing has has is remarkable." Repeating
the transformation once more gives (E).
It is intuitively obvious that, of these three type of recursive opera-
tions, self-embedding (E) is psychologically the most difficult. Although
they seem grammatical by any reasonable standard of grammar, such
sentences never occur in ordinary usage because they exceed our
cognitive capacities (Miller 1964:36).

Generative grammar explains that and why (R), (L) and (E) are
all grammatical because it provides a theory of competence.
Psychology has shown that the measure of complexity of (E)
exceeds our cognitive capacities. As a theory of performance
based upon a theory of competence, the psychology of language
can therefore explain what linguistics cannot, namely, that (E) is
unintelligible and never uttered.
Miller reports on an experiment he and Stephen Isard carried
out with Harvard undergraduates. They were asked to memorize
sentences that differed in their degree of self-embedding. They
found that everybody can manage one embedding, some people
two, but everybody has trouble with three or more (Miller 1964:
37).
When we study mantras or ritual, we are primarily interested
in competence—in what the ritualists know and not only in what
they do. That is why we include information that the Srauta
Sütras and other sources provide—for example, that certain
mantras are not uttered or "uttered m e n t a l l y / ' Neither a tape-
recorder nor a behaviorist can accept this; but then neither is in
a position to arrive at a general and explanatory model or theory.
The same holds for a performance theorist. He can describe and
enumerate, but not account for anything. Of course, perfor-
250 Mantras

mance cannot be ignored. In the present book, we are primarily

concerned with a theory of competence but such a theory cannot
be evaluated without the kind of evidence that was presented in
A G N I , and that consists largely of performance. The account of
the 1975 Agnicayana that was the subject of A G N I provided
both the "thick description" of performance and the "grand
t h e o r y " of competence. Some features of its methodology were
explained in the book itself in the following terms:

The 1975 Agnicayana was a traditional Nambudiri performance, though

we were partly responsible for its occurrence and indirectly contributed
to the nonsacrifice of goats, which was in some sense a deviation from
the tradition. In the social sciences there is much interest in moderniza-
tion and other changes that influence people's behavior and societies as
a whole. The traditional philologist confines himself to texts and to what
they can teach us about the past. In the humanities we are not only
interested in people's behavior and in their past, but also in what is in
their minds—for example, in what they know. Unfortunately, minds and
knowledge are invisible, and therefore have to be made manifest,
something that can be done in many ways, including outside prompting.
This would cause no surprise to the scientist working in his laboratory,
where he studies reactions he has himself induced. Naturally, we need
not be surprised that, in any situation, the observer may influence what
he observes. The philosopher muses on the boundaries of objectivity, the
romantic may wish to limit such influence, and the moralist to direct it.
The scholar or scientist is content to describe, analyze, and explain its
occurrence and extent (AGNI 11:475).

Patanjali was perhaps the first to state that the enumeration of

forms of performance lead nowhere unless they are accounted
for by a system of rules that exhibits competence (page 40).
A G N I posed the question: what insight can be gained from a
single performance of a known tradition about ritual perfor-
mance as well as competence? Schechner (see Schechner 1986,
1987; Staal 1987 and above page 93) noted that the 1975
performance of the Agnicayana must have been different from
other, earlier performances. That fact was taken into account in
A G N I ; it is a dead issue. Schechner presented his case with flair.
As the Sanskrit says, " e v e n a crow assumes the demeanour of
 Performatives, Pragmatics and Performance 251

an eagle when it comes upon a dead lizard" (mrtam dundubham

äsädya käko'pi garudäyate).
A sociologist or political scientist may explain why in the
seventies and eighties, many American social scientists have
been fascinated by the study of performance. It may not be
accidental that these scholars belong to the society that re-
elected a President noted for performance, not competence.
 22

Mantras and Language

One assumption underlies all accounts of mantras that I am

familiar with: the assumption that mantras are a kind of lan-
guage. Since many mantras are expressions of Vedic or Sanskrit,
they obviously often consist of language. However, it does not
follow that they are language. Nor does this follow from the fact
that linguistic methods appear to be fruitful in their study.
Since language has been studied well, and mantras have been
studied only rarely, one way to determine their relationship is by
trying to find out to what extent linguistic methods are applicable
to the study of mantras. We have seen that the syntactic
approach is promising as distinct from both the semantic, which
has run into problems, and the pragmatic, which has limited
significance. What about a phonological approach?
Mantras that consist of sentences of a language are always
constructed in accordance with the phonology of that language.
We should therefore look at those cases where mantras do not
consist of language, e.g., at stobhas and bïja mantras. But here
we reach the same conclusion: even if they are not words of
Sanskrit, these mantras are constructed in accordance with the
phonological rules of Sanskrit. I have found two apparent
exceptions to this rule, one in the Stobhänusamhära (referred to
on page 227), and one discussed by Padoux. The first is pnya. I
do not believe thatp/ia- occurs in Sanskrit in initial position, and
neither does pnya-. In middle position both are available, e.g.,
svapna, " s l e e p , d r e a m , " and svapnya, " a vision in a d r e a m "
(the latter occurs in the Atharvaveda, and is rare). Perhaps pnya
was constructed by a Sämavedin who heard svapnya and
mistakenly made the assumption that this form consisted of the
familiar reflexive pronoun s va- and a hypothetical -pnya.
254 Mantras

The unphonological mantra studied by Padoux is certainly not

pronouncable: rkhksem. However, its analysis (Padoux 1963:
356-358) is both pronouncable and clear in Tantric terms:
ra-kha-ksem. I therefore believe that we are entitled to retain the
general conclusion that Indian mantras are constructed in accor-
dance with the phonological rules of Sanskrit.
This conclusion is further supported by the fact that, when
mantras are exported from India to other countries, their pho-
nological structure generally adapts itself to their new environ-
ment. In Tamil, for example, the term mantra itself becomes
mantira both as a word of the language and inside mantras in
which it occurs (Diehl 1956:290). For Chinese, numerous illus-
trations of a similar nature may be found in van Gulik 1980. For
Japanese, Payne (1986:400-412) has listed ninety-three mantras
used in the Japanese Buddhist fire ritual called goma (from
Sanskrit homo) in their Sanskrit forms and Japanese equivalents
that are not translations but transliterations. I shall quote a few,
adding a translation if it makes sense:

1. Sanskrit: om sarva-tathägata-päda-vandanam karomi,

4
OM. I salute all the T a t h ä g a t a s "
Japanese: on saraba tatagyata hana mana nay kyaromi
2. Sanskrit: om tathägatodbhaväya svähä, " O M . For the
production of Tathägata—Svähä!"
Japanese: on tatagyata dohanbaya sowaka
3. Sanskrit: om padmodbhaväya svâhâ, " O M . For the pro-
duction of the L o t u s — S v ä h ä ! "
Japanese: on handobo do hanbaya sowaka
4. Sanskrit: om vajra karma kam, " O M . Thunderbolt rite—
Kami-
Japanese: on bazar a kyarama ken
5. Sanskrit: om bhüh kham,"OM. Earth—Space!"
Japanese: on boku ken
6. Sanskrit: om agnaye gaccha gaccha bhüh, " O M . Go, go
to Agni!"
Japanese: on agyanau ei gessya gessya boku.
 Mantras and Language 255

Anthropologists have paid much attention to ritual, and to a

lesser extent, mantras. Scholars of religion have studied both
topics. All have assumed that mantras are some kind of linguistic
utterances. This assumption on the part of anthropologists is in
line with the high regard some have had for linguistics as an
example or paradigm of scientific methodology. Such an expla-
nation hardly applies to scholars of religion w h o , in general,
have stayed away from scientific methodology (an exception is
Hans Penner to whom we will return). Anthropology has been
repeatedly inspired by developments in linguistics, and these
waves of inspiration have flowed along channels cut by the
subdisciplines of phonology, syntax, and semantics. Thus Lévi-
Strauss has been inspired by the phonology of Jakobson, Singer
(e.g., Singer 1984) by Peirce's semiotics which is a kind of
semantics, and Tambiah by Austin's performatives, as we have
seen. Oddly enough, what appears to be missing is syntax—the
most basic of these disciplines, at least according to many
linguists, logicians and mathematicians.
Syntax is not really missing in anthropology. We find it in the
famous 1899 article by Hubert and Mauss already referred to:
" E s s a i sur la nature et la fonction du sacrifice/' Hubert and
Mauss offered a " s c h e m e abstrait du sacrifice" which is primar-
ily syntactic in nature: for example, they made, for the first time,
the elementary but fundamental syntactic observation that rites
have a beginning, a middle and an end. Their analysis is to a
large extent derived from the syntactic analysis of Vedic ritual
found in the Srauta Sütras. Mauss had been introduced to these
texts by Sylvain Lévi who taught during 1896-1897 at the
Collège de France, specifically for Mauss (Mauss 1969, 111:538)
a course on Vedic ritual which was subsequently revised and
published in 1898 as " L a doctrine du sacrifice dans les
B r ä h m a n a s . " Hubert and Mauss also referred in their article to
published sources based upon the Vedic Srauta Sûtras, such as
S c h w a b ' s 1886 monograph on the Vedic animal sacrifice entitled
" D a s altindische Thieropfer." Durkheim (1915:386) concluded
from Hubert and Mauss, as well as from other anthropological
data, that a rite can serve different e n d s ; but he did not draw the
256 Mantras

further conclusion that ritual is therefore independent of the

ends it is supposed to serve. Accordingly, he did not pay much
attention to syntax, which is a pity for anthropology (cf. above,
pages 125-130).
Why are we justified in characterizing the analysis of Vedic
ritual given in the Srauta Sütras as syntactic? T o provide an
adequate answer to this question would necessitate a lengthy
excursus into Vedic literature, but in the present context it
should suffice to state that the syntactic analysis of ritual in the
Srauta Sûtras supplements and is to some extent supplemented
by its semantic interpretation in the Brähmanas. Just as the
Brähmanas stand at the source of much of Indian mythology, the
syntactic analysis of the Srauta Sütras is closely related to the
origins of grammar. We shall see in Chapter 26 that (1) the
ancient Indians possessed a science of ritual which used primar-
ily syntactic methods, and that (2) linguistics originated in India
in close association with this syntactic analysis of ritual.
European phonology has been profoundly inspired by the
Indian grammarians, and J a k o b s o n ' s concept of "distinctive
f e a t u r e s " still bears the stamp of their influence. The study of
syntax entered Western linguistics much later, with Zellig Harris
and more definitively with N o a m Chomsky (Chapter 6A). T h u s ,
if we had to picture the influence exerted by linguistics (and by
logic and philosophy via linguistics) on anthropology in tabular
form, and in terms of the three subdisciplines to which we have
referred, adding " p e r f o r m a t i v e s " as a fourth, we should prop-
erly include these ancient Indian forerunners, as follows:

Linguistics Anthropology
phonology Prätisäkhya, Pänini, Lévi-Strauss
Jakobson
syntax Pänini, Chomsky Srauta-sütras,
Hubert & Mauss
semantics Frege, Peirce, Saussure Singer
performatives Austin, Grice, Searle Tambiah
 Mantras and Language 257

In most of these relationships, the influence has been in the

direction from linguistics (including logic and philosophy) to
anthropology; the only exception is the ritual science of the
Srauta Sûtras, which I have listed in the table under w 'anthro-
pology" for lack of a better heading and which influenced Pânini
and the other Indian grammarians.
What, then, is the relation between these various disciplines
and the science of religion? Before we address this question we
must make several observations. First of all, it is not really
anthropology as a whole that has undergone the influence of
syntax and pragmatics; it is rather the anthropological study of
ritual. In the case of phonology, the case is different, and the
consequences have been baffling: for Lévi-Strauss introduced
into anthropology a form of analysis that stressed binary oppo-
sitions and that was based upon distinctive features, though all
linguists know that distinctive features had been postulated
successfully only in phonology, and play no role in syntax or
semantics. Lévi-Strauss, in other words, introduced into anthro-
pology a linguistic method he regarded as a universal panacea,
while it is applicable only to a subdiscipline of linguistics. One
might even go further and reason that it is a priori unlikely that
a method that can be used in phonology but not in syntax or
semantics could be useful in anthropology. Moreover, why
should two suffice for man when so many other natural numbers
exist and are found elsewhere in nature? In his 1957 article fcin
Defense of the Number T w o , " Halle has shown that the system
of distinctive features due to Jakobson can make do with much
less conceptual machinery than is usually employed by assuming
that the structure of features is binary. In other words, specific
empirical restrictions and conceptual sophistication combine to
show that, in phonology, two is the number we need. But
Lévi-Strauss has not given a similar demonstration for anthro-
pology. If Lévi-Strauss had studied linguistics later, or Chomsky
developed syntax earlier, Lévi-Strauss, instead of paying atten-
tion to binary opposition, might have introduced the syntactic
methods that had been adumbrated by Hubert and Mauss in their
258 Mantras

study of ritual, but that were not fully developed in linguistics

until Chomsky.
Since ritual is one of the main areas of research common to
anthropology and the science of religion, one might expect that
the latter science would have been inspired by linguistics or logic
as well. In fact, the scene for such an influence had long been
set. As early as 1867, Max Müller predicted: " I t was supposed
at one time that a comparative analysis of the languages of
mankind must transcend the powers of man: and yet, by the
combined and well directed efforts of many scholars, great
results have been obtained, and the principles that must guide
the student of the Science of Language are now firmly estab-
lished. It will be the same with the Science of Religion" (in:
Waardenburg 1973, 1:86). What happened, in fact, was that the
comparative study of religions developed as Müller had pre-
dicted; but when the "Science of L a n g u a g e " made a method-
ological transition from comparative and diachronic philology to
a synchronie science of linguistics, the "Science of Religion"
lagged behind. Instead of developing a syntactic method of
analysis appropriate to its object, it fell prey to phenomenology,
existentialism, and hermeneutics. It was left to anthropologists
such as Mauss tö study religious phenomena like ritual in the
manner in which de Saussure had studied language, as a
" s y s t è m e où tout se tient." Thus the science of religion failed to
make the transition which would have turned it, like linguistics,
from a respectable branch of scholarship into a contemporary
scientific discipline as well.
Belatedly, scholars of religion have expressed an awareness of
the need for such a transition. In a reaction to my article " T h e
Meaninglessness of Ritual" (Staal 1979b), Hans H. Penner has
stated that " t h e study of linguistics is the necessary foundation
for explanations in religion" and that "language as we all know
is composed of signs, and all linguistic signs have phonological,
syntactic and semantic c o m p o n e n t s " (Penner 1985). Yet, in the
same article, Penner has asserted that my syntactic analysis of
ritual is " i r r e l e v a n t . " He has supported this by emphasizing
 Mantras and Language 259

that, by my own admission, these pieces of analysis fctdo not

correspond to any existing ritual" or " t o any actual ritual."
Disentangling some of the misconceptions that are at the root
of these assertions should help us to understand more precisely
the relationships between the methodology of science and the
science of ritual—or any empirical science, for that matter. It
would be correct to say that anthropology has been influenced
by linguistics; but this does not mean that specific pieces of
linguistic analysis have been of great use to anthropology. For
example, some of Lévi-Strauss' work serves to warn us that an
uncritical acceptance of distinctive features is artificial and
unproductive within anthropology. In my study of Vedic ritual I
found something different, viz., that some ritual structures can
be generated by rules that are similar to the rules that linguists
call transformations, while others are unlike any structures
found in natural languages or with which linguists are familiar
(Chapters 11, 19 and 23). From a methodological standpoint, the
most basic issue is not whether ritual or anthropological struc-
tures are similar to linguistic structures. Of importance is that all
such structures are postulated and never correspond exactly to
actual facts such as rites. They correspond, if they are adequate,
to features that are abstracted from actually existing rites (cf.
page 102).
Lévi-Strauss may have been wrong with regard to the specific
binary oppositions, or other specific phonological or syntactic
structures, that he postulated; but he is right with respect to
method. What is basic to his method, and to all scientific
method, is that structures are postulated entities that are not
empirically given—that are, in fact^ invisible—but that enable
us, provided they are adequate, to account for what is empiri-
cally given and visible. Lévi-Strauss has in fact told us where he
came upon this idea, which was novel and startling to him, a
French philosopher by training, and which remained the main
insight that separated him from existentialists such as Sartre (see
Sartre 1960 and Lévi-Strauss 1962, Chapter 9): he took from
geology, from Marx and from Freud the idea that is basic to all
science, viz., that reality is different from what it appears to be
260 Maniras

(Lévi-Strauss 1968:61-^62). This is obvious to scientists, as

indeed it is to most ordinary people; the only persons who have
failed to understand it are certain types of philosophers bien
étonnés de se trouver ensemble, such as behaviorists, phenome-
nologists, practitioners of hermeneutics and ordinary language
philosophers such as Wittgenstein (Chapter 2). What anthropol-
ogy has taken from linguistics is therefore, in the final analysis,
nothing peculiarly linguistic but rather an insight into the char-
acteristic features of science.
These general considerations at the same time help to explain
the importance of logic and mathematics. These latter disci-
plines study structures in general, including the most general
structures that can be imagined. The abstract, underlying struc-
tures that are postulated in order to account for empirical facts
and events can generally be expressed in a mathematical or
logical form. This is the common insight and methodology
shared by anthropologists such as Singer, Tambiah and Lévi-
Strauss (whatever their differences), but that empiricists such as
Edmund Leach (e.g., in his 1985 review of Singer 1984), and
many scholars of religion, fail to appreciate and understand.
We are now in a position to see that the influence of linguistics
on anthropology and on the science of religion does not neces-
sarily * imply that the objects of the latter two sciences are
languages or even " s y s t e m s of meanings and s y m b o l s . " What I
have attempted to demonstrate is that ritual is not a language,
but is, like language, a rule-governed activity. Since it is an
activity that is governed by rules, it becomes important to
discover what actual rules, and what kind of rules, govern it. In
pursuing such questions, I found that ritual structures can be
analysed in syntactic terms not by methods specific to linguis-
tics, but by mathematical and logical methods that have also
influenced linguistics. I cannot say whether other traditional
object areas of anthropological or religious study are also
rule-governed systems; but mythology, for example, does not
appear to be such a system, at least insofar as Vedic mythology
is concerned. Vedic ritual and Vedic mythology are in fact
remarkably separate developments, that rarely correspond with
 Mantras and Language 261

each other and that are accordingly distinguished from each

other by the relevant experts (e.g., Tsuji 1952:187; Renou 1953:
16—above page 65—29; Dandekar 1982:77). This continues to
apply to later Indian rituals: Hindu or Tantric rites do not
attempt to reproduce any story, legend or myth (Renou 1953:59).
We are now in a position to extend the hypothesis on ritual
and grammar to the origin of language from syntax (above, page
112). It seems likely that human language is not as old as was
once believed: perhaps around one hundred thousand years
(Burkert 1987:152 suggests about forty thousand). Ritual is much
older; Neanderthaler man had elaborate rituals (Burkert, ibid.).
This is supported by the facts of animal ritualization which are
similar to human rituals especially with regard to their structure
(see, e.g., Barlow 1977). It is likely that the same holds for
mantras, for mantras occupy a domain that is situated between
ritual and language. To mantras or mantra-like sound structures
among animals we shall turn in Chapter 23.
An important fact that supports the idea that mantras are older
than language, is the following. Mantras are generally transmit-
ted within linguistic contexts, but when the language of the
context is translated, the mantra is not: it remains invariant.
Thus we find the same or similar mantras in India, southeast
Asia, China, Korea and Japan. The only difference between such
mantras is, as we have just seen, that their form is adapted to
their new phonological environment, just as Sanskrit homa or
dhyäna becomes Japanese goma or zen. Such transformations
are akin not to translation but to transliteration. This applies
equally to apparently " m e a n i n g l e s s " mantras such as om or hum
phat and to apparently "meaningful" mantras such as namo
buddhäya or sivâya 'namah. The evidence for the untranslata-
bility of mantras is plentiful all over Asia (see e.g., H o o y k a a s ,
Skorupski and Strickmann in AGNI II). Among the " B a u d d h a
B r a h m a n s " of Bali, for example, namo buddhäya or sivâya
namah are not translated as "Hail to B u d d h a ! " or "Hail to
Siva!<" but remain in their original form (see Hooykaas 1973,
passim). We meet not only with such clear-cut cases but also
with mixtures of translated and untranslated (because untrans-
262 Mantras

latable) forms. This is not confined to " G r e a t e r I n d i a , " but

occurs on the subcontinent itself as in the following example
" t r a n s l a t e d " from the Tamil: " K a n Kan Kal Kal Cunatin Pin Pin
Kalai Kalai Separate Separate Tuva Tuva T ü m a Mantirakäli
Lord of the Smoke S v ä h ä ! " (Diehl 1956:290).
There are additional reasons which support the idea that
mantras are older than language. Mantras may or may not
consist of language; but they are treated, in either case, in a
manner which is totally different from the way we use language.
The situation is similar to a common occurrence in biology when
a part of the body that is used for a specific purpose is also used
for something quite different; for example, when legs, which are
used for walking or running, and perhaps kicking, are also used
for swimming. Legs are used, in that case, as if they were fins.
If we assume that language was preceded by mantras in the
course of biological evolution, but that the " o r i g i n a l " mantras
are no longer there, it follows that all we have access tc at
present is mantras-couched-in-language. These vestigal mantras
are not used for the expression of meaning, like the other forms
of language, but they are used differently, like the original
mantras. In other words, we have an analogy: language is to
mantras what legs are to fins.
We may formulate this conclusion in slightly different terms
by distinguishing between apparently meaningless mantras such
as om and apparently meaningful mantras such as siväya namah.
Let us assume that om was an "original" mantra; that is, we
assume it existed before language was born. Suppose that
Pithecanthropus, for example, was humming om but could not
talk. Let us further assume that this om re-appeared in Sanskrit
where it is now used in the manner mantras are used, e.g.,
preceding, following, or interspersed between chants and reci-
tations, accompanying rites, meditation, etc. This om is not used
in the manner in which other words are used: it is not declined
like a noun, for example, but it may be quoted like other words
(e.g., with ///). N o w contrast this with siväya namah, apparently
a Sanskrit expression: this can be construed as a sentence
consisting of two nouns with an implicit verb, there is a käraka
 Mantras and Language 263

relation between the two nouns which is expressed through case

endings, etc. This expression, then, is a mantra, like om, but it
is couched in language: it belongs to language in the sense that it
is governed by the rules of language; but it is also used in the
context of ritual recitation, meditation, etc. Though an expres-
sion of language, it may be used in the manner of om, just as
legs, though ordinary human limbs and not piscine accretions,
may be used in the manner of fins.
The hypothesis that mantras are in some respects unlike
language, and are remnants of something that preceded lan-
guage, may explain certain curious and hitherto baffling facts
about mantras: foF example that they are repeated endlessly, but
also that they have a proclivity to reduce to nothing (via japa
" m u t t e r i n g s " and upämsu "murmurs'*'to manasä "in t h o u g h t " :
Renou 1949; Renou and Silburn 1954); that they consist of
"poussière v é d i q u e , " as Renou called it; or that they result from
" l e découpage des vieux hymnes en formules ou même en
fragments devenus des corps inertes dans la trame liturgique"
(Renou 1960:76; quoted in Malamoud 1983:33).
While form, in a natural language, is at most as important as
meaning, the form of mantras is more important than their
meaning. In this respect, mantras are a typical product of Indian
civilisation-—a civilisation where form is all-important, where, as
in Malamoud's summary of Renou's formulations, " c ' e s t l'ar-
rangement des formes et la spéculation sur les formes qui
révèlent le plus clairement le contenu, qui sont, en fin de
compte, le c o n t e n u " (Renou 1978:3). Mantras are not confined
to India, but this insistence on form may explain why India has
produced more mantras than any other civilization.
That mantras are different from language is supported by their
being cut and broken into pieces, and subjected to strange
insertions and musical or other sound transformations. But this
hypothesis does not seem to derive support from the fact that
some mantras consist of language: for such mantras cannot be
older than language at least in their present form. In order to
address this problem in appropriate terms we should take
account of an important difference: that between historical
264 Mantras

origin and current function. We have already met with a special

case of that difference: the difference between etymology and
usage. The general distinction is deeper and wider, especially in
a biological perspective.
Illustrations of this distinction are discussed in a recent review
on the relations between culture and biology and the origin of the
human mind by S. J. Gould (1983). In the course of evolution,
our ancestors first began to walk upright, and subsequently
developed a large brain. The humanoid creatures that lived in
Africa some three or four million years ago did already walk
upright, but their brain remained at an ape's characteristic size.
It took a long time before the brain began to develop to its
present dimensions. Whatever the complex reasons for these
developments, the outcome enabled men to perform all manner
of operations bearing no direct relation to their original impetus
or motivation. Walking upright, for example, enabled them to
develop gestures (e. g., mudra) and to make fire, but may also be
related to the origin of speech, since hands were now free to do
certain things that previously had to be done by mouth (e. g.,
carrying food). Similarly, the brain came to be used to perform
many tasks that are unlikely to be related to the original reasons
for its biological growth. Such a state of affairs is not confined to
biology: t4 I may put a computer in my factory only to issue
paychecks and keep a c c o u n t s , " writes Gould, " b u t the device
can (as a consequence of its structure) also calculate pi to 10,000
places."
N o w let us return to mantras. These have, by and large, been
studied by Sanskrit scholars and therefore primarily within a
diachronic perspective. But once we adopt a synchronie point of
view, it is apparent that the Sanskrit that occurs in mantras is
often used in an unintelligible fashion and not in the manner in
which any natural language is normally used. Even those man-
tras that say something or have meaning are not used like
linguistic utterances when they are ritually used. From the point
of view of their ritual use, there is no difference in treatment
between mantras we would regard as meaningful and mantras we
would regard as meaningless. In the context of a natural Ian-
 Mantras and Language 265

guage such a state of affairs is inexplicable—nay, unthinkable;

the distinction between meaningful and meaningless is funda-
mental to human language in all its normal uses. Though specific
mantras may derive frpm sentences of a specific language, their
use indicates a different background which has nothing to do
with language. The synchronie study of this usage has therefore
diachronic implications: it points to a more distant origin,
preceding the succession of stages or periods in the development
of any particular language.
We have already met with many illustrations of uses of
mantras that are different from any use of language. Here is
another example.- The original meters of many mantras, that
were accorded great importance during the period of Rigveda,
are lost in the ritual application (prayoga). In the transmission of
mantras {adhyäya), the names of the metres are sometimes
preserved, but in actual recitation the characteristics of the
metres have disappeared and are replaced by special renderings,
often musical, of the Vedic accents (udätta, anudätta and
svarita). This treatment is characteristic of Veda recitation and
is given not only to poetry, but also to prose (provided the
originals were accented), and not only to meaningful but also to
meaningless mantras. The entire development has nothing to do
with the normal preservation, continuation, development or
change of language, but is rather a musicalization of language.
Let me summarize. Many mantras are constructed from
language, but in their ritual use, which is their proper use, they
do not conform to any of the normal uses of language. Though
many specific mantras are derived from language, the species of
mantra need not therefore be derived from the species of
language. On the contrary, in order to explain these various
anomalies, we must assume the opposite, viz., that language
derived from mantras in the course of evolution. In that case, the
origin of language marks an entirely new use of mantras,
unrelated to their original function—-just as the many functions
of walking upright or of a large brain mark new uses that are
unrelated to their original uses.
Can we set up a more detailed hypothesis about the develop-
266 Mantras

ment from mantras to language? We may guess, provided we do

it in such a way that our speculation can be discussed rationally.
M y present guess is that meaning was introduced last into an
evolution that consists of three stages:

I. Earliest Stage M A N T R A S of T Y P E 1
These are sounds subject to phonolog-
ical constraints, e.g., bija mantras such
as him, or stobhas such as bham.
II. Intermediate Stage M A N T R A S of T Y P E 2
These are sequences, two-dimensional
arrangements, or elaborate construc-
tions of mantras of type 1, sometimes
subject to syntactic constraints, e.g.:
hä bu hä bu hä bu bhä bham bham
bham . . .
or:
huvä yi väcam I väcam huvâ yi I . . .
{Jaiminiya Aranyageyagâna 1.2:
A G N I , 1:525).
III. Final Stage LANGUAGE
These are mantras of type 2 subject to
semantic, further syntactic and differ-
ent syntactic constraints, e.g.: väcam
yaccha, ''Control your s p e e c h ! "

I must leave it to specialists to provide chronological estimates

for the duration of the first two stages in this scheme of
evolution. The earliest stage represents features that are found
among vertebrates, and are certainly pre-human. (The term
"phonological''' in this context refers to any rules that put
constraints on the combinations of sounds.) The intermediate
stage may be anthropoid or characteristic of early man, but is
probably much older. This is at any rate suggested by the bird
songs discussed in the next Chapter. The final stage corresponds
to the last fifty to hundred thousand years of the development of
homo sapiens.
 Mantras and Language 267

I regard the evidence in support of the hypothesis that mantras

are older than language as strong enough to be subject to rational
discussion. The evidence for the priority of monosyllabic man-
tras over polysyllabic mantras, that is, for the priority of Stage I
to Stage II, is less compelling and stands in greater need of
empirical data on bird song and on growling, miauling, singing,
barking and chirping by many other animal species, from whales
to grasshoppers. To think that monosyllabics are earlier than
polysyllabics may be an instance of what may be called the
fallacy of "atomistic reductionism." On the other hand, there
may be serious grounds for such a priority. Apart from the
evidence from babbling babies, there is one kind of aphasia, in
which the patient i s in a; position to produce and recognize
phonemes, but not words; in another kind, he can produce and
recognized words, but not sentences (see Jakobson in: Jakobson
and Halle 1960, 1964). Such facts suggest the priority of Stage I
to Stage II. That language itself developed from monosyllabic
forms is an old speculation. Von Humboldt discussed it in the
last chapter of his famous monograph (Humboldt 1836:373-414).
Let us take another look at the form of our hypothesis. Since
mantras are used in a manner that is different from the manner
language is used, we postulate that their use reflects the function
of something else that preceded them and that is more properly
" m a n t r a " . In other words, we postulate " o r i g i n ä r ' mantras,
now lost, and " l a t e r " mantras which survive and sometimes
resemble language, but language used in a pre-language fashion.
In other w o r d s , I assume that there has been a development of
B (human language) from A (mantras), followed by the occa-
sional emergence of functions in B that are more easily explained
in terms of its predecessor A than in terms of its successor, B.
Such a situation, we have seen, is not rare in biology. The
earliest vertebrates were fish, and the wings of birds, as well as
the limbs of reptiles and mammals, developed from fins. The
present functions of these body parts are clear: fins are for
swimming, wings for flying, and legs for running. Of course,
these limbs did not originate for the sake of performing these
functions. They arose* by chance and were selected in accord-
268 Mantras

anee with the demands of the environment. The variation we find

is, in fact, extraordinary. Crocodiles no longer have fins, but use
their legs for swimming. The earliest crocodiles, such as Pela-
gosaurus, lived in the open seas. Since their legs and tail did not
enable them to swim well, they began to live in and around
rivers. So here we have a case of the development of B
(crocodiles' legs) from A (fishes' fins), followed by the emer-
gence of functions in B (swimming) that are more easily ex-
plained in terms of A (fins) than in terms of their successors B
(legs). .
Another interesting case is penguins. These are birds, but they
cannot fly. Their wings have developed into flippers which
enable them to swim extremely well (20 miles per hour). Walking
is difficult fpr penguins, but they can glide on their bellies on ice
over long distances with the help of their flippers. So here we
have a development from fins (A) into wings (B), but the wings
are mostly used in the manner in which fins are used, and to
some extent in the manner of ski poles. This becomes more like
humans using language in the manner of mantras. Men use their
arms and legs as they use their language: the former are
generally used for walking, running, grasping, catching, gestic-
ulating, etc., and sometimes, archaically, for swimming; the
latter is generally used for speaking or thinking and sometimes,
archaically, in the manner of mantras. Numerous parallel devel-
opments in other animals, and çpuntless more distantly related
cases, therefore support the hypothesis that human language
developed from mantras, and still preserves rudiments of this
mantric background.
V There are cases outside religion where man uses language
entirely or almost entirely in the manner in which mantras are
used. This resembles the penguins' use of wings as if they were
fins, but in the case of man, it either is considered regressive and
pathological, or is confined to babies. Leopold von Schroeder
observed in 1887 that there are striking similarities between
mantras and the utterances of megjal patients. Such similarities
have been noted and commented on by Eggeling, Keith, and
others, but mostly in rhetorical fashion. Von Schroeder (1887:
 Mantras and Language 269

112-114) was more straightforward and serious. He began his

discussion with an illustration of mantras, quoting those that are
recited by the Adhvaryu priest when the ukhä pot, chief vessel
of the Agnicayana, is manufactured. Von Schroeder translated
from Maiträyani Santhitä 2.7.6, but I shall provide here the
parallel passages from Taittiriya Samhitä 4.1.5 1-q and 6 a-d (see
A G N I , 1:297-299 and cf. Ikari in: A G N I , 11:168-177):

1. You are the head of Makha

m. You are the two feet of the ritual.
n. May the Vasus prepare you
with the gâyatrï meter
in the fashion of the Angirases!
You are the earth.
May the Rudras prepare you
with the tristubh meter
in the fashion of the Angirases!
You are the sky.
May the Adityas prepare you
with the jagatï meter
in the fashion of the Angirases!
You are heaven.
May the Visvedevas, common to all men
prepare you with the anustubh meter
in the fashion of the Angirases!
You are the directions. .
You are the unchanging direction.
Make unchanging in me children,
abundance of wealth,
abundance of cattle, heroism,
and similar things for the Yajamäna.
o. You are the waist band of Aditi.
r
p. May Aditi grasp your hole
with the pankti meter
in the fashion of the Angirases!
270 Mantras

q. Having fashioned the great ukhä

made of clay as a womb for Agni,
Aditi gave it to her sons saying, " F i r e it!"

à. May the Väyus smoke you with t h e g â y a t r ï meter

in the fashion of the Angirases!
May the Rudras make you smoke with the jagatï
1
meter
in the fashion of the Angirases!
May the Visvedevas, common to all men,
make you smoke with the anustubh meter
in the fashion of the Angirases!
May Indra smoke you in the fashion of the Angirases!
May Visnu smoke you in the fashion of the Angirases!
May Varuna smoke you in the fashion of the Angirases!

b . May Aditi, the goddess,

in union with the All-gods,
dig you, trench, in the realm of the earth
in the fashion of the Angirases !

c. May the wives of the gods, the goddesses,

united with the Visvedevas,
put you, ukhä, in the realm of earth
in the fashion of the Angirases!

d. May the Dhïsanâs, the goddesses,

united with the Visvedevas,
fire you, ukhä, in the realm of earth
in the fashion of the Angirases!
May the wives, the goddesses,
united with the Visvedevas,
fire you, ukhä, in the realm of the earth
in the fashion of the Angirases!

Von Schroeder compared these mantras with the following

piece written by a patient and quoted from Th. Güntz (1861; I
translate from the German):
 Mantras and Language 271

First Prayer: .'*..•

Schiller save his soul and consciousness
Jesus save his soul and consciousness
My mother save her soul and consciousness
van der Velde save his soul and consciousness
Trommlitz save his soul and consciousness
Gerstäcker save his soul and consciousness
Voss save his soul and consciousness
Seume save his soul and consciousness
Körner save his soul and consciousness
Arndt save his soul and consciousness
and save the soul anid ponsciousness of all poets of the book of songs.
Second Prayer:
for all the names that are in Schiller's work.
Third Prayer:
for the soul of my family.
Fourth Prayer:
to destroy my consciousness and my ego.

Von Schroeder also quoted a prose passage from a patient at the

hospital Rothenberg near Riga (I translate from the German):

With humility and affection walk the streets, the indicated, with full
knowledge go the streets, which favor going the road with humility, and
with deep devotion go the streets, which favor to build the church and
keep the peace, which indicated the way which is necessary and
desirable for that, build the road with God's desire, buy the peace, and
then with good spirit build the church, which is favored, and with good
intention gain the stage of learning, which could be desirable for that,
with devoted endeavor give roses to the institution, build God's church
and show his submission with much humility, with much submission and
humility try to reach that goal, with much submission try to gain that,
and with humility walk the way which is required, make use of God's
love, with good intentions lead a good life, with right decision take the
road which is required, with good intention go the road which is
required, use God's love, with progress go the way, of God's love, build
the church, God's love, build the church, God's love, build the church
and with good intention, God's love, build the church and with good
intent, God's love, build the church (the last two phrases are repeated
about 80 times, and it goes on like that for several pages).
272 Mantras

When the psychiatrist asked why he wrote the same thing all the
time, the patient answered that he did not know anything else.
Though these writings are pervaded by religious notions, no
one would regard them as religiously inspired writing. It is likely
that we have here a case of regression to an earlier stage of
development: language is used here in the manner of m a n t r a s -
type II mantras, to be precise, for it is mainly semantics that is
affected. Stobha-like mantras are used by other kinds of pa-
tients, and in cases of aphasia, to which I shall return.
Mantra-like uses of language are also found among babies, and
here the recapitulation of phylogeny by ontogeny provides even
more striking support for the thesis that language has developed
from mantras. Nancy Budwig drew my attention to Ruth Weir's
study on the babblings and pre-sleep monologues of a two-and-
a-half year old child, alone in his crib, talking to himself. Here is
an example of what he uttered a few minutes before the onset of
sleep:
like like
one like
two like
three four like
monkey's like
up up
light light
turn the light
light
all gone all gone
it's all gone it's all gone
it's not all gone
it's not all
' stop it stop it
there (squealing)
yayaya wau wau gigouboubou gigouboubou
now it's all gone
all gone (falsetto)
go go go go
all gone all gone all gone all gone
good luck
that's one
two
 Mantras and Language 273

go go go go (falsetto)
close the door
gee gee gee gee gee gee

(Weir 1970:128). The following sequence immediately preceeded

sleep, and contains more stobha-like elements (I have replaced
the phonetic transcriptions by approximate spellings):

yiii (squealing)
III
did
gigigi'gi
the baby the baby the baby
(Baby is crying in the adjoining room)
baby the baby baby (6 times)
iii : • • • ' ••. ' ' ••••"•, ;-: ••-••' "'•'-•
baby baby baby
bay /
baby
bay
happy baby
that's the baby
"•• • • bay : ;. •. . •-•••• • . • •• . . •, ••••;•
:
: •.• •;•.;••..., • baby. • •'•••••.,.. . .. • . . . - . • • . •..
• ,. .. that's the baby
baby
(
yaa
aa (squealing)
SLEEP

(Weir 1970:197).
Mental patients and children often display features that are
reminiscent of earlier stages of evolution, and that may be
referred to as archaic. Religion is generally conservative and
characterized by archaic features. It is probable that there are
other features of religion that may be interpreted as regressive.
Glossolalia or speaking-in-tongues is a related form of regression
(see May 1956). Mantras are always archaic. They are often
attributed to ancestors or primeval sages (such as the Vedic
rsis), or are regarded as eternal or as having originated in a
golden age (krtayuga or satyayuga). In Sri Lanka, where de-
274 Mantras

mons are apparently primeval, mantras are referred to as the

"language of the d e m o n s " (yaksâ bäsäva: Tambiah 1968:177).
The archaic nature of mantras is related to the fact that many
mystical phenomena are archaic (cf. Staal 1975). The mystical
state is a state of awareness that can be reached or produced
with the aid of mantras, a state of consciousness that is " b e y o n d
language" or "ineffable." Mantras give access to this ineffable
state. To say with Renou, Padoux, and Wheelock (in Under-
standing Mantras'.Alpzr 1988) that mantras are beyond the
boundary of language, at the highest level of speech "situated
beyond language and eventually right to the zone of silence," or
to say that mantras "point backwards to the source of language,
which is the source of all creation itself ' (pages 98, 120) is not
merely a matter of phenomenological, religious or spiritual
metaphor, also found in the Indian texts; such expressions may
be taken literally as asserting that mantras are the predecessor of
language in the process of human evolution.
All natural languages share certain phonological properties
that are regarded as universal (see, e.g., Chomsky and Halle
1968, Part IV). Are there also universal mantras? It may seem
premature to ask such a question since mantras, outside the
Vedic realm, have been studied so haphazardly. Moreover, we
should exclude historical influences, borrowing and exports: for
example, mantras exported from Sanskrit into Chinese, Korean,
Japanese, or Tibetan, and adapted to the phonological structure
of the recipient languages. All of these illustrate, incidentally,
T. R. V. Murti's statement that " B u d d h i s m is Hinduism for
e x p o r t . " However, Vedic and Sanskrit have no monopoly in the
export of mantras. There are purely Chinese mantras in Taoism
and, according to Parpola, the famous mantra om may have been
imported into Vedic and Sanskrit from the Dravidian (Parpolß.
1981).
Actually, as far as om is concerned, we are on firmer ground.
According to Jakobson (1962:541), " t h e most natural order of
sqund production is an opening of the mouth followed by its
c l o s u r e . " This is a very apt description of the mantra om. As for
ontogeny, Jakobson informs us that the child first passes through
 Mantras and Language 275

a babbling stage in which precisely such sounds are produced,

and then arrives at the "first acquisition of conventional s p e e c h "
in which it clings to the model " c o n s o n a n t plus v o w e l . " Then
repetitiveness comes into operation—-the most basic element of
ritual syntax—resulting in what in linguistics is called reduplica-
tion. Consonants formed by a complete oral closure predomi-
nate, and this leads first to mama, which does not, however,
refer to the ïtiothër but is a general expression of affection.
Language comes into being only when purely referential mech-
anisms begin to operate, when there is, in J a k o b s o n ' s words,
" t h e transition from affective expression to designative
language." Thus papa arrives on the s c e n e — " t h e first distant,
merely deictic, rudimentary cognitive attitude in the child's
verbal b e h a v i o r " (page 543).
The importance of om and its priority to language is inherent
in this scenario, which depicts how om comes before mama, and
mama before p a p a who introduces language. Variations of om,
with repetition, survive in Western Asia: am-en. Another uni-
versal mantra answers the general description of "opening the
mouth followed by its c l o s u r e " equally well: the mantra him
with its variant hum, both common in Vedic and Tantric
contexts. Intoned at the beginning of many chants, it is not
confined to India or even Asia. In Mozart's Zauberflöte Papa-
geno sings:

Hm hm hm hm - - - -
. _*{• . •• .

Hum is no longer confined to the old world since the American

composer Ruth Crawford-Seeger composed in 1930, as Paul
Attinello informs me, Chant 1930 which begins:

•.••„•.• Hurr\ Hum Hum.

The occurrence of " h " in these mantras may be due to their

onomatopoeic representation of breathing. This is much clearer
than in the case of ong and ang, about which the Danish
missionary Bartholomäus Ziegenbalg wrote around 1711: " T h e
276 Mantras

heathens say that breath, when it is inhaled, produces the sound

ong, and when exhaled, the sound ang" (Caland 1926:109). Om,
itself is also related to breathing: it figures prominently in
pränäyäma recitations (see A G N I 1:283, 380, Plate 62). Or
perhaps, the author of the Taittiriya Samhita explained it
correctly when referring to the wind:

väyur himkartä
"The maker of the sound HIM is Vâyu"
{Taiîtinya Samhitâ 3.3.2.1 a).

Other candidates for universal mantrahood are hi and ha.

Compare for example the German jingle:

Unter einen Apfelbaum

hi ha Apfelbaum
hatt'ich einen schönen Traum
hi ha schönen Traum. ,

Hi and hay are common in Peyote songs, which consist in

general of meaningless syllables, especially among the Arapaho
(see Nettl 1953). Ha is also found on Tierra de Fuego: when
Waldon and Drayton landed there in 1838 from H, M. S. Beagle,
" a group of natives took their arms and jumped with them in
time to the following song:

'wHa ma la ha ma la ha ma la ha ma la
O la la la la la la la la" (Bowra 1966:388).

Throughout our investigation, I have mainly used data from

Asia. Similar phenomena occur all over the world under different
names, and should be studied within a similar perspective.
Anthropologists are familiar with Peyote songs and related
phenomena (e.g., McAUester 1949 and Nettl 1953, just quoted).
Classical scholars are familiar with the Greek magical papyri,
mostly from Egypt, that abound in meaningless; sounds and
syllables. Edited first by Preisendanz, they have now been
revised and republished by Henrichs (Leipzig 1973-1974). Bib-
lical scholars are familiar with the description in the New
 Mantras and Language 211

Testament {Acts 2) of the Holy Ghost descending on J e s u s '

disciples, taking the form of tongues of fire and making them
" s p e a k with other t o n g u e s . " Glossolalia, already referred to, is
especially common in contemporary America as a recent survey
depicts:

Tongue speaking, or glossolalia, almost vanished after the apostolic age

except for a brief revival by Montanus in the second century. Saint
Augustine set the pattern for the Catholic Church and the Reformers by
asserting that God withdrew the gift after it served its purpose. This was
also the view of Thomas Aquinas. There is no evidence that Luther or
Calvin spoke in tongues, but early Methodists revived the gift, and it was
soon flourishing among the Quakers, Shakers, Irvingites, Mormons, and
other fringe sects. After 1900 a variety of churches sprang up in the
United States, on fire with tongues and faith healing, to become the
denominations now called Pentecostal. Today there are about thirty-five
of them. They are the fastest growing segment of Christianity, not only
here but throughout the world.
Charismatics, sometimes called neo-pentecostals, are evangelicals,
not necessarily fundamentalist, who accept the gifts of faith healing and
tongues. The term applies of course to the old or "classical1' Pentecos-
tals, but also to Catholics, Episcopalians, and members of mainline
Protestant churches where there has been since 1960 an astonishing
inrush of Pentecostal fervor.
Non-Christian glossolalia is a problem for charismatics. Ancient
soothsayers, and devotees of Greek and Roman mystery cults, often
gurgled meaningless sentences. In the Aeneid (Book 6) Virgil describes
tongue speaking by a Roman sybil, Some Moslem sects and primitive
cultures practise glossolalia. In The Great Dictator, Charlie Chaplin's
Hitler gives a rousing speech in German double talk. None of this,
charismatics maintain, is the real thing. Non-Pentecostal fundamentalists
think the babbling of their Pentecostal brothers isn't the real thing either,
and may even be inspired by Satan (Gardner 1987:18).
 23

Mantras and Bird Songs

If mantras are older than language in the course of biological

evolution it would seem likely that similar sounds exist and are
used in a similar manner by other animal species. This expecta-
tion can only be properly explored by biologists, but even a
cursory look at bird songs shows that it is fulfilled at least to
some extent. Let me begin with a specific example.
When the " e x t r a " pebbles of the Agnicayana we have already
discussed (above page 194) are being consecrated by the Adh-
varyu on behalf of his patron the Yajamäna, the Udgätä sings his
pebble songs. The first of these is Jaiminïya Grämageyagäna
45.2.1:

yo no hä bu I idam idam purä hä bu I pra vä pra

vasya yä hä yi I ninä ninäya tam u vo hä bu I stüsä yi
sakhäya yä hä yi I dramütäyäyi I ö yi là 11

We can represent the structure of this chant by introducing

capital letters to stand for its units, which are either bits of
mantras or stobhas, as fallows:

"A" for y o no
" B " for hä bu
l4
C" for idam idam purä
"D" for pra vä pra vasya
" E " for yä hä yi
" F " for ninä ninäya tam u vo
k
'G" for stüsä yi sakhäya
4t
H''fordramütäyäyi
k
'J" for ö yi lä.

The structure of JGG 45.2.1 may now be represented by:

280 Mantras

AB/CB/DE/FB/GÈ/H/J// (9)
H o w do we arrive at an analysis of this type? How arbitrary
are the units we have selected to be represented by capitals? The
answer to these questions is that we can only arrive at a
nonarbitrary analysis provided we analyse a number of similar
forms. First of all we make use of the subdivision of the sâmans
into smaller units, called bhakti, that are preserved by the
tradition (because a new breath has to be taken at the beginning
of each bhaktî); and that are represented in printed texts, and in
the version given here, by slanted bars ('7."). In addition we
have to take into account that hâ bu, yä hä y/, and ö yi lä are
stobhas that function as units in many other songs. The correct-
ness of (9) therefore depends on the correctness of other forms.
When the same sort of analysis is applied to other recitations it
yields the following structures:

A A B/ . . . / A A B C / D D D / / (10)

(the first chant of the Udgätä, sung during the foundation

ceremonies of the altar: A G N I 1:410);

A A A B B B C C C / . . / A A A B B B C C C' /D D' D ' / /

(11)
(the first space-filler chant during the consecration of the first
layer: A G N I 1:443; note that we use the convention that C and
C' are similar, or have identical beginnings, and the same holds
for D and D ' ) ;

A A A / . . . / A A A' / / . (12)

(agner vrata sung at the northern hip of the altar after the five
layers have been consecrated: A G N I 1:539);

A A A B / . . / A A A C B' / / (13)
{agner arka sung at the head of the altar on the same occasion);
and so forth.
Now let us look at a bird song: four stanzas sung by the Black
Flycatcher Ficeêula hypoleuca published by J. C. Röche
 Mantras and Bird Songs 281

( " L ' o i s e a u musicien" no. 6) and analysed by François-Bernard

Mâche ( 1983:93) in the following form:
A B C B C B D E /
B' C B G B C B F E ' /
A B C B C D/
B'CBCBCBFE'// (14)
Here the capital letters stand for small phrases, motifs or
figures that each consist of a few tones, and are therefore
comparable to the units of mantras that each consist of a few
syllables. I have used the slanted bar ' 7 " . where Mâche begins
a new line or " s t r o p h e . " Here too the selection of units on which
the analysis is based derives from a study of other songs that
consist of similar units, for example:
A B B C / A A A A B / A B C B / A A A B C B / A
A B C/Ä A A A G B / A A AC B/A A B/A A B
C B/A A A B/A A B / A A B B / A A// (15)
(song of Acrocephalus schoenobaenus: Mâche 77);
A A A B/ A AC7 A AC B / A AC B /A A AC
BV A A A C / A B 7 A A C B / A A C / / (16)
(song of Acrocephalus dumetorum: Mâche 78);
A B A B C A ' B B ' A" B ' B (17)
(song of Sylvia communis, Mâche 97; note that A' and A" are
again similar, hence A and A" are similar, but less similar than
either A and A' or A ' and A") And so forth.
If we compare these structures of mantras and bird songs with
each other we find a certain similarity (e.g., the doublets,
triplets, and refrains discussed in Chapter 16 occur in all) but no
strict identity. However, all we are interested in is structural
similarity. If the structures happened to be strictly identical, it
might prove that JGG 45.2.1 was caused, influenced or inspired
282 Mantras

by the songs of Ficedula hypoleuca, perhaps because that bird

occurred in a part of India where the composers of the säman
might have heard it. Mâche (page 82) has argued that a certain
structural pattern in Stravinsky which also is found in a bird song
was copied " p l u s ou moins c o n s c i e m m e n t " because the bird is
common on the banks of the river Bug in Ustilug where
Stravinsky used to work. Such a causal explanation would be
interesting if we could establish it in the realm of the Sämaveda,
and might demonstrate that there is a similarity in structure
which makes such causal links possible. Actually, Acrocephalus
dumetorum Blyth or " B l y t h ' s Reed Warbler" of example (16),
which breeds in Baluchistan, is a winter visitor to the entire
subcontinent, and Sylvia communis Latham, the " W h i t e t h r o a t "
of example (17) is an autumn migrant to Pakistan and northwest
India (Ripley 1982:424). However, similarities in the realm of
biological evolution only rarely express specific causal links;
what they do express is similar responses to the pressure of
selection-—a feature of biological theory to which I shall return.
So far we have chiefly looked at structural patterns. The
structural similarity between mantras and bird songs is certainly
striking, but it would only be significant if it were not also to
occur in language, the domain to which mantras have been
generally—albeit unwittingly^—assigned by the authors of the
Brâhmanas as well as by contemporary scholars. What we find
in this respect is unexpected, and therefore intriguing. Repeti-
tions such as AA, AAA, AAAA, A B B B , etc., and refrain-like
structures that seem to be common to both mantras and bird
songs, are entirely absent from the syntax of ordinary language.
In fact, any linguist who is familiar with syntactic structures
cannot fail to be struck by the absence in almost all such
structures of the typical repetitive features of both mantras and
bird songs. There is an area of overlap: the domain of poetry.
However, poetry seems in several respects to constitute an
intermediary area between mantras and ordinary language.
Excluding poetry (which would deserve separate treatment) and
confining ourselves to mantras and bird songs, their similarity in
structure and dissimilarity fr^m the structures of language are
 Mantras and Bird Songs 283

curious and suggestive. But are we not barking up the wrong tree
because we are confining ourselves to structures at the neglect of
meaning and function, categories in terms of which mantras and
bird songs seem to be vastly different?
I have argued in Chapter 13 that many mantras and rites do not
possess a clear meaning or function. Many mantras, as well as
stobhas or bïja mantras, consist of meaningless sound; but even
those that are sentences of Vedic or Sanskrit are often unrelated
to the rites they accompany. As for rites, the rites of the srauta
ritual are especially devoid of meaning or function. Rites of the
grhya or " d o m e s t i c " ritual, on the other hand, seem often quite
meaningful and functional. But how can there be such a differ-
ence in meaning or function if the actual forms of srauta and
grhya rites, together with the accompanying mantras, are so
often similar or even identical?
We shall approach this problem by first taking a closer look at
the meaning and function of the grhya rites, or, more generally,
of the general category of non-srauta rites. This would be too
vast a task to undertake within the compass of the present
chapter had not Gonda fortunately come to our aid by publishing
a manual in the Handbuch der Orientalistik, which assembles
precisely the data we need under the title: " V e d i c Ritual: the
non-solemn r i t e s " (Gonda 1980). Here we witness rites per-
formed not only in connexion with the traditional samskära
ceremonies which correspond in part to the "rites de p a s s a g e "
familiar to anthropologists following Van Gennep, but also on a
great number of other occasions, .e.g.: making rain, warding off
dangerous or annoying animafs, averting dangers such as fires
and robbers, curing illness and preserving health, going on and
returning from a journey, entering a house, conquering enemies,
inheriting property, a variety of occupations connected with
cattle, anniversaries, attaining great age, austerities, prayers,
vows, magic, witchcraft, exorcism, divination, purity and impu-
rity, auspiciousness and inauspiciousness, learning and concen-
tration, not to mention seasonal ceremonies of many kinds.
" T h u s occasions were f e w , " concludes Gonda, " w h e n Vedic
man did not seek the aid of rites and mantras to promote in some
284 Mantras

way or other his welfare or to avert the ills, real or imaginary,

which might invest h i m " (page 410).
In Tantrism, which develops the domestic rites in several
respects, mantras are similarly ubiquitous. A tenth or eleventh
century Tantric text, the Prapancasära declares (8.15):

Therefore this famous string of letters, which attaches itself to the world,
with devotion assign to your body, mutter with soft voice, sacrifice and
worship,
in order to obtain incomparable poetic power, longevity, fame, loveli-
ness and fortune; for the sake of the destruction of all calamities and
sorrows, and for release from existence j >
(translation hy Goudriaan 1981:134).

In all these cases, rites and mantras seem to have a meaning or

function. Our startling earlier conclusion that rites have no
meaning seems to turn into the opposite which is not less
startling: there is no meaning or function rites can not possess.
But this has always been a problem with regard to rites and
rituals. Jack Goody drew attention to this problem as we saw at
the onset of our discussion of ritual (Introductory Note of Part
II). Kristofer Schipper introduced the expression accretion to
capture another dimension of the same phenomenon.
We shall revert to this problem but only after having paid
attention to the corresponding problem in the realm of birds and
bird songs, which seem to have a much clearer function. Or do
they? To begin with, ornithologists were satisfied with a general
characterization, stressing first that bird song ""probably always
serves in one way or another to distinguish the s e x e s " (because
bird songs are mostly found in males), and second that " t h e r e
must be a need to communicate. The need may be to show the
signaler's location (contact notes), his awareness of danger
(alarm calls), or his readiness to mate (courtship songs) or
defend a breeding or foraging territory (territorial s o n g ) " (sum-
marized in Hartshorne 1973:15).
Another account of the functions of bird song runs as follows:
' T h e singer, more usually a male than a female, has two primary
reasons for which to sing, the one to defend its territory against
 Mantras and Bird Songs 285

other males, and the other, to attract females. The same song or
call (which is shorter and simpler than a song) may have different
meanings, depending on its context—the call that in the mating
season both defends territory and attracts females functions,
after the mating season is over, to defend territory alone. As
against such general signals, there may be specific ones, and
these specific ones may be graded in structure, length, and other
characteristics, by which I mean that different intensities may
call forth different intensities of response. The occasions that
have been said to elicit distinctive songs or song-variants are
many. They include the search for a nest-site; the building of the
nest; the incubation of the eggs; the spelling of the female in her
incubation duties; the leaving of the nest by the young; migra-
tion; and autumn and winter. Calls, too, have many varieties and
functions. One classification includes those for pleasure, terri-
torial defense, flight, feeding, nesting, flocking, aggression,
general alarm and specialized alarm (to distinguish flying from
non-flying p r e d a t o r s ) " (Scharfstein, forthcoming).
That the same song or call may have different meanings
depending on the context accords with our earlier observation
that rites have all kinds of functions. A similar observation was
made by van Gennep in 1911 ("the same rite . . . can change its
meaning depending on the position it is given in a c e r e m o n y " )
and by Durkheim in 1915 ("according to the circumstances, one
and the same ceremony serves two distinct functions"): see
above, page 127. Later in his essay Scharf stein specifies this
context dependence in the case of a blackbird song as follows:
' T o w a r d the end of the season, when the song has lost its overt
sexual function, it reaches its most highly organized, in the
human sense: most musical, f o r m . "
In more specialized studies, ornithologists have felt the need
to use the ethologists' concept of ritualization. This concept is
invoked when the original function of a behavioral pattern is no
longer visible or known. Let us start with an example not
involving song. In a general study of the behavior and songs of
a species of finch {Estrildida), Güttinger (1970) has tried to
'explain a pattern of inflexion ( " V e r b e u g u n g " ) during which the
286 Mantras

male bird bends its head either to the right or to the left,
independent of the position of the female even if she is close by.
Güttinger says (1970:1054): " A u s dem Bewegungselement der
bisher besprochenen Arten kann infolge des hohen Grades der
Ritualisierung die ursprüngliche Bedeutung der Verhaltensweise
nicht mehr ermittelt w e r d e n . "
German ornithologists use "ritualization" in contrast to
"Ableitungsmöglichkeit," or "possibility to d e d u c e " : the latter
term is used when it is possible to explain a pattern of behavior,
including a song, in functional terms. Applying the distinction to
the realm of Vedic ritual we could say that there exists an
"Ableitungsmöglichkeit" for the grhya rites, but not for the
srauta rites. As in the Vedic realm, such a distinction requires a
great deal of analysis in specific cases. Some young finches of
the species Spermestes, for example, when in the nest and about
to be fed by the parents, raise their wings in a special manner.
Immelmann (1962) explained this as a kind of contest between
siblings; moreover, the wings can be raised in such a way that
the neighbor is partly hidden, so that the young bird who
performs the act is more likely to be fed first. But the theory does
not hold, as Güttinger (1970:1063) has pointed out, because the
parents take great care to feed all their young equally, including
not only those that do not raise their wings, but also those that
are sick. In such cases, there seemed at first to be an "Ablei-
tungsmöglichkeit," which on further analysis turned out to be
spurious.
Detailed examples of the ritualization of songs have been
studied by Thielcke (1964) in the European tree creeper {Certhia
brachydactyla Brehm and Certhia familiaris L.). The sounds
from which these songs are constructed are primarily three,
which Thielcke, using German phonology, represents as tür, tut
and srih. Though the functions of some of the songs are clear,
the functions of others are not, and ritualization is accordingly
postulated. Tree-creepers exhibit all the characteristics of ritua-
lization that Lorenz (1951) and Morris (1957) have enumerated:
the primary sounds are rhythmically repeated; the sequences are
changed in such a way that " e i n e urspünglich weitgehend
 Mantras and Bird Songs 287

veränderliche und aus verschiedenen Gliedern bestehende Folge

von Verhaltensweisen zu einem einzigen, in sich starren Bewe-
gungsablauf zusammengeschweisst w i r d " (Thielcke 1964:405,
after Lorenz 1951); and there are "typical intensities." The
latter concept, introduced by Desmond Morris, indicates, for
example, that a bird in the course of a courtship dance may
repeatedly twist its body four times (Morris 1957:4-5). Since the
order of elements, the sequence of pitches, the intervals between
the elements, and the number of elements are much more
constant in songs than in sounds, Thielcke has argued that the
phylogenetic development of the song of the European tree-
creeper has taken place in the direction from sounds to song.
I am not an ornithologist and I am not in a position to provide
a detailed and exhaustive discussion of the vast amount of
information on bird song that is nowadays available. However,
a general trend in these studies seems to be the following.
Whereas originally, most ornithologists stressed functional ex-
planations, and were convinced that there were good pragmatic
or utilitarian reasons for the execution of songs by birds, it has
increasingly been found that there are no such reasons, and that
birds often seem to sing simply because they like to. This is
expressed in the more specific context of the " o n t o g e n y " of bird
song, for example by Nottebohm (1970:951) when he writes:
" O n e is forced to conclude that the achievement of a relatively
stable and predictable song does not imply the existence of a
preconceived or acquired g o a l . " Or it is expressed in more
general and theoretical terms, for example by Lorenz (1967:75):
" t h e r e are mechanisms in existence which reinforce economical
perfection in motor skills independently of the attainment of the
ultimate biological goal in whose pursuit the learned movement
is d e v e l o p e d . " Lorenz mentions as examples the swimming of a
shark, the gallop of an antelope, and the gymnastics of a gibbon.
Here the singing of birds may be plausibly added: in fact, we
have already seen that " p l e a s u r e " is listed as one of its possible
reasons (Scharfstein quoted above). Lorenz adds: " T h e most
convincing argument in favor of my speculative assumption lies
in the fact that acquired motor skills of this type, more than any
288 Mantras

other types of movements, are forever being performed for their

own saké in the obvious absence of any other motivating or
reinforcing factors. Indeed, the very concept of play is based on
this fact to a large e x t e n t . " All of this does of course apply to the
human animal: " w e know that when we skate, ski, or dance, we
perform these activities for their own sake, for the enjoyment
they give u s " ( i b i d ) . To add music, mantras and ritual is but a
natural extension of this list.
Two of the books and one of the articles from which I have
quoted are concerned with a similar problem: the relation
between bird songs and music. The first book, Charles Hart-
shorne's Born to Sing (1973), is by an ornithologist who is also
a philosopher; the second, François-Bernard M â c h e ' s Musique,
Mythe, Nature (1983) is by an ornithologist who is also a
musicologist and composer; the essay by Scharfstein, a chapter
from a forthcoming book on the nature of art, is by a philosopher
and esthetician interested in birds. All are concerned to show
that it is impossible to define " h u m a n " music in such a way that
it will not also encompass bird song. All stress that birds must
accordingly be assumed to possess something like an aesthetic
sense. (Scharfstein has in fact cast doubt on the functionalist
bias of some of the professional ornithologists: " T h e y tend to
assume that every nuance of song or display has survival value
or disvalue, and they try, often with conspicuous cleverness, to
turn each apparent exception into further evidence for the truth
of their t h e s e s . " ) In the present context I need only the least
controversial feature of the conclusion of these three authors: I
don't need the hypothesis that birds have an aesthetic sense
(though it may be correct) but accept that these authors have
demonstrated beyond reasonable doubt that bird songs cannot
be explained in functionalist, utilitarian, or pragmatic terms.
We may now return to mantras, only to find that we are in a
position to make an important observation. The idea that birds
can talk, and that bird songs are a kind of language, belongs to
the realm of myth and has never been taken seriously by
ornithologists. But the idea that mantras express meaning, and
are a kind of language, has generally been taken for granted by
 Mantras and Bird Songs 289

Sanskritists. Of course, there are good reasons for this silent

assumption. It is due not merely to the circumstance that
Sanskritists tend to confine themselves to Sanskrit, but to the
much more important fact that man, the animal that speaks (ÇÔJOV
kàyov e'xov), is obsessed by language so that we tend to assume
that any articulate sound is " l a n g u a g e " in its usual sense. This
is exactly what happened in human history and prehistory. It has
generally been fancied that birds could really talk even though
the evidence has never been telling. N o wonder that mantras,
which are often indistinguishable from it, continue to be re-
garded as language.
Conceptual issues involving cross-overs between nature and
culture tend to be complex. Functionalist explanations in the
realm of mantras are always closely connected with problems of
semantics, whereas functionalist explanations in the realm of
bird song are not connected with semantics. However, in both
areas, emphasizing syntax combines in a natural manner with
de-emphasizing semantics. Thus there is a close association
between Chapters 10, 11, 12 and 13. Similarly, when ornitholo-
gists found that it was difficult or impossible to provide function-
alist explanations, their interest in syntax increased. Thus we
find, since roughly 1965, syntax mentioned in the studies of bird
songs by Thielcke, Güttinger, Baptista, Nicolai, Payne, Beer,
Nottebohm, Johnson, and Bertranv The work of these scholars
is discussed in an unpublished paper by Luis Baptista, " O n
Parallels between Bird Song and Speech D e v e l o p m e n t . " How-
ever, the first systematic use of syntactic methods and represen-
tations on a large scale is found in Mâche's book of 1983. The
author is aware of the novelty of his approach, "abandonnant le
point de vue phonétique qui est celui des zoologistes pour un
point de vue syntaxique" (Mâche 1983:64).
The contrast between functionalist and non-functionalist ex-
planations in biology is closely related to the continuing debates
on the theoretical foundations of Darwinian natural selection—a
problem that can not be adequately discussed in the present
context, but to which a brief reference should be made. The
notion of "survival of the fittest," in particular, can be easily
290 Mantras

misinterpreted in such a manner that it becomes tautologous and

empty. Eldredge exposes such views in the following terms:
" O n l y (God, natural selection) could have fashioned such a
marvellous organic machine! There is a difference, of course:
God, as a supernatural being, does not belong in science,
whereas natural selection patently does. But used in this inap-
propriate fashion, natural selection becomes a mere substitute
for the Creator. It tells us nothing, really, about tribolite eyes or
anything specific or meaningful about how they came into exist-
e n c e " (Eldredge 1982: 508-509; quoted by Koster 1983: 4).
Such interpretations were actually given by Tom Bethell
(1976), according to whom "survival of the fittest" involves no
more than "differential reproductive s u c c e s s , " viz., the produc-
tion of more surviving offspring than other competing members
of the population. This formulation, however, defines fitness in
terms of survival only and is therefore empty because survival of
what survives is a tautology. Such interpretations are similar to
the functionalist bias of some ornithologists who seem to be
"straining to accomodate awkward f a c t s " (Scharfstein).
Gould (1977:40-45), from whom the quote from Bethell has
been taken, has convincingly shown that Darwin did not make
such a basic logical mistake. Biologists define the fittest before-
hand, not only by their subsequent survival. Natural selection is
a theory of local adaptation tç> immediate but changing environ-
ments; it does not preach general progress in the cosmic sense as
inherent in the workings of evolution, which explains the Vic-
torian unpopularity of the idea. In fact, the theory of natural
selection did not triumph until the 1940s. Voltaire's Dr. Pangloss
asserted that the nose was fashioned to bear spectacles, but we
can now see clearly that our world is not, as Candide thought,
" t h e best of all possible w o r l d s . " We know, as François Jacob
formulated it in Le jeu des possibles (1981:33), that " l e monde
vivant aujourd 'hyi, tel que nous le voyons autour de nous, n'est
q u ' u n parmis de nombreux possibles."
How then are mantras and bird songs related within the
perspective of biological evolution? Their similarity is not what
biologists call " h o m o l o g o u s , " but what they call " a n a l o g o u s . "
 Mantras and Bird Songs 291

The difference is explained by Gould (1977:254): "Similar features

due to common genetic ancestry are 'homologous'; similarities due
to common function, but with different evolutionary histories, are
'analogous': the wings of birds and insects, for example—the
common ancestor of both groups lacked wings." To apply this
definition, we have to broaden it in a curious fashion: for the
similarity between mantras and bird songs is due not to common
function, but to common non-functionality. Mantras and bird songs
share not only certain structural properties, but also lack of an
inherent or absolute purpose. It is precisely these features that
express the common characteristic of both as essentially satisfying,
pleasurable and playful—features that, in the case of mantras, have
remained even though language has intervened.
I shall briefly deal with a final objection to the thesis that mantras
and bird songs are analogous: are not mantras and language
learned, whereas bird songs are instinctive and inborn? The first
reply to any such objection should be that the difference between
these two modes of acquisition may be large from the point of view
of the individual, but is less important in terms of biological
evolution, and need not prevent certain phenomena from being
analogous to each other. However, this first reply is not relevant in
the present case, for the distinction between learned and innate
does not obtain in any straightforward fashion. Language, first of
all, unlike languages, is partly inborn. One's first or mother
language is learnt in a realtively short period and through a process
that is different from the learning of any subsequent language
because "universals of language" are characteristic structures of
the human mind. This much, if anything, has been established
conclusively by Noam Chomsky. It is entirely likely that the same
holds true of rites and mantras: though specific cases have to be
learnt, their general underlying structures are innate.
Bird songs, on the other hand, are partly learnt. Kroodsma
(1978) provides a comprehensive recent survey on ' 'Aspects of
Learning in the Ontogeny of Bird Song: Where, From Whom,
When, H o w Many, Which, and H o w A c c u r a t e l y ? " Kroodsma
(1974) had already shown that juvenile males of the species
Bewick's wren (Thryomanes bewickiï) " a r e probably capable of
292 Mantras

learning from their father . . . but a premium is placed on

matching the song types of neighboring males with whom
interactions will occur throughout life; thus, song types of the
father may be modified or new ones learned in order to match
more closely the songs of neighboring m a l e s . " Many birds learn
songs only within a "critical p e r i o d , " for example, within the
first 10 months of a potential lifespan of 5 or more years. "After
the end of this sensitive period further acoustic stimuli do not
alter the bird's r e p e r t o i r e " (Nottebohm 1970:952). However, in
the indigo bird (Vidua chalybeata), "flexibility to change song
types in the repertoire of the individual is apparently maintained
throughout life" (Kroodsma 1978:216). Since so many songs are
learnt, "variation in song repertoire sizes among different pop-
ulations of a given species may be striking. Long-billed marsh
wrens near Seattle, Washington, may have 150 song types, while
an Illinois male may have as few as 3 0 " (Kroodsma 1978:221).
That mantras are similar to bird songs may seem far-fetched to
Western Sanskritists, but it would not cause surprise to the
säman chanters, other composers of mantras, or Indians gener-
ally. Following Genesis, the Western tradition has generally
stressed the difference between man and the other animals (see
Staal 1979c:2-3). But in India, the names of many sämans are in
fact inspired by birds. The kraunca melody evokes a kind of
curlew. We also have sämans called trßyakrauhca and even
vähnidhanakraunca, a kraunca melody "ending in s p e e c h . "
Other sämans are called plava after an aquatic bird, bhäsa after
a bird of prey, and väsa after a croaking bird. The well known
chant syena is called after a vulture (see A G N I 1:88-90), and
sauparna derives from suparna which is not only a lotus leaf, but
also a large bird of prey. Vedic schools have been called after
birds, e.g., the Kausika after the owl and the Taittiriya after
tittiri, the partridge (a totemic bird according to Kosambi (1950)
who concluded that the Taittiriya was a totemic clan—a hypoth-
esis criticized by Brough (1953)). N o wonder that väc was used,
already in the Rigveda, not only of men, but also of other
animals and especially of birds.
Geese informed Janasruti Pautrâyana about the great master
 Mantras and Bird Songs 293

Raikva-with-the-chariot (Chändogya Upanisad 4.1-3). A goose

and a diving bird taught Satyakäma Jäbäla each a quarter of
brahman (Chändogya Upanisad 4.7-8). Indian sages have al-
ways been eager to gain knowledge and insight from birds. So
why shouldn't we?
The similarity between mantras and bird songs, which is
greater than that between mantras and ordinary language, may
have to be studied in the wider perspective of "animal
language," a topic that goes far beyond the confines of the
present investigation. Let me only note that many claims made
ealier on behalf of language-like communication among chim-
panzees, gorillas and even dolphins, have been largely retracted.
It is now widely recognized that there is a difference between
communication and human language. It is obvious that animals
communicate with each other; but these communications, even
if they are systematic, do not resemble the system of human
language. This is not to say that human language is something
incomparably grand: we have seen something different, namely,
that it is illogical and roundabout. These puzzling features are
precisely the features that call for an explanation and that have
led to confusion (see Chapters 28-30).
To suggest explanations for human data within a biological
perspective implies acceptance of a form of Darwinism. Al-
though many features of biological evolution in terms of Dar-
win's original theory have been abandoned or at least questioned
during recent decades (e.g., the gradual nature of evolution), one
tenet remains central: the importance played by chance. Varia-
tions arise by chance and some are selected subsequently in
accordance with the demands of the environment. A theory that
explains human data in terms of " m e a n i n g " tries to make sense
of everything and minimizes chance or rejects it altogether. It
becomes important, therefore, to inquire whether chance plays
any role in the domain of mantras. We have seen that mantras
and acts are generally associated with each other but that the
association is unlike that between linguistic utterances and their
meanings. How then is a specific mantra selected to accompany
a specific rite or vice versa? This is the topic of our next chapter.
 24

Mantras by Chance

The verse that is our point of departure is the first of the ninth
book, the only book of the Rigveda that deals in its entirety with
Soma and its ritual purification and preparation. It is the first
verse of a hymn (RV 9.1) that belongs to a group of hymns all
written in the same meter, called gäyatri. The text of this
gâyaîrï-verse is:

svädisthayä madisthayä
pâvasva soma dhärayä (18)
indräya pätave sutâh (RV 9.1.1)

The translation is straightforward and uncontroversial:

"With most delicious and inebriating flow.

Soma, purify yourself,
Pressed for Indra to drink."

The gâyatrï meter consists of three octosyllabic meters which

have the scheme:

x x x x U — U — (19)

where x may be occupied by a syllable of any kind, U by a

verse-medial short syllable and — by a long syllable or by a
verse-final short syllable. In the case of RV 9.1.1, the distribu-
tion is as follows:
296 Mantras

U— U — U' —

U — U— U — U — (20)

— — U — U — U —

The a c c e n t " ' ' " which has been printed over certain syllables of
the text in (18) is called udätta, " r a i s e d , " and it is possible that
it was originally spoken with raised pitch (but see Gray 1959a
and b for a dissenting view). Whatever its original pronunciation,
it was later marked by reciting the previous syllable at a lower
pitch, and the following at a higher pitch. A few more rules have
to be taken into account before we can properly pronounce this
verse, but we shall not trouble ourselves with them. The lower
accent is called anudätta, " n o t raised," and is written in the
manuscripts (that are all of later date) with a horizontal bar
below the syllable; the higher, svarita, " ( r e ) s o u n d i n g , " is writ-
ten with a vertical bar above it. The result is as follows:
i i
svadisthayä madisthayä
i i
pavas va soma dhärayä (21)
indrâya pätave sutah

This mode of recitation, which is called svädhyäya, only serves

the purpose of transmission, from teacher to pupil or from father
to son. Mantras are never recited in this manner in the ritual
itself: to be fit for ritual use, the traditional recitations have to
undergo certain modifications. In many cases, verse have to be
turned into songs which are then incorporated in the Sämaveda.
In the above case, the first step o f t h a t process of incorporation
is extremely simple. The udätta, svarita and anudätta are
chanted at three different pitches, each at an interval of about a
second from each other, the first {udätta) the highest, the last
{anudätta) the lowest, and the svarita at an intermediate pitch.
The manuscripts refer to these tones with the help of numerals:
 Mantras by Chance 297

" 1 " for udâîta

" 2 " for svarita (22)
" 3 " for anudäita.

These numerals are written above the syllables, as follows: .

12 3 12 3
svädisthayä madisthayâ
12 3 12
pavasva soma dhäravä (23)
1 2 3 12 32
indrâya pâtave sutah
(23) is d o s e r to (18) 'than it is to (21), which shows that the
Sämaveda is earlier and closer to the original Rigveda than the
Rigveda system of horizontal and vertical strokes used in the
manuscripts. This is in accordance with the chronology pro-
posed by Kiparsky (1982, Lecture II) for the development of the
Vedic accent system, provided we insert the Sämaveda after the
Rigveda as it was known to Pânini, but before the Rigveda
recension with which we are familiar.
Before we can sing (23), we have to know at which pitch the
syllables not marked with a numeral have to be sung. The rule is
again simple: these are sung at the pitch of the preceding
syllable. The basic chants of the Sämaveda, thus provided with
a pattern of song that derives from the accentuation of the
original, are listed in the first part of the Sämaveda, which is
referred to as Samhitä—as in the case of the other Vedas—and
also, more appropriately, as.arcikä, "list of v e r s e s , " from rk
" v e r s e " from which the appellation rgveda, " V e d a of V e r s e , " is
also derived.
The next portion of the Sämaveda lists the melodies to which
these verses are sung. It consists of the more common
grämageyagäna, " s o n g s to be sung in the village," and the more
esoteric aranyageyagäna (or aranyegeyagäna), " s o n g s to be
sung in the forest." Both these gäna-hooks correspond to the
svâdhyâya of the other Vedas: they mainly serve the purpose of
transmitting the tradition to the following generations; they are
sometimes ritually used in their gäna form, but more often this
form undergoes further modification for the sake of "ritual
298 Mantras

application" (pray o ga). This application is the actual purpose of

these songs and of the Sâmaveda itself.
There are numerous melodies in these gäna collections be-
cause one verse is generally sung to different melodies. But we
also meet with the opposite: different verses are sung to the same
melody. In that case, the melody needs to be listed only once;
and this has led to some confusion among students of the
Sâmaveda.
The Vedic tradition is celebrated for the care with which it has
been handed down; despite the vicissitudes of Indian history
during the last three thousand years, there are in the Rigveda no
variant forms or " r e a d i n g s " (a term based upon the Western
assumption that these compositions are " t e x t s " which are
" r e a d " ) . This applies to the place of the accents as well: it never
varies. In the case of the transmission of melodies, greater
variation is expected; and this may well account for the belief
that there were originally a thousand schools in the Sâmaveda
(Renou 1947:88) as against two in the case of the Rigveda and
Yajurveda each. When one hears contemporary traditions of the
Sâmaveda, the musical renderings vary greatly. Yet, two tradi-
tions can be clearly distinguished: the Kauthuma-Ranâyanïya,
which is still found in many parts of India (chiefly Tamilnad,
Andhra, Karnataka and Maharashtra); and the Jaiminlya which
is confined to Kerala and a few villages in Tamilnad.
The Jaiminïya is almost extinct; its manuscripts are rare and
have, with few exceptions, remained unpublished. The Kau-
thuma-Rânâyanïya tradition of two closely connected schools,
which I shall refer to as K-R, is still relatively strong, and its
texts have been published in their entirety. In these latter texts,
a simple notation for the pitch of tones of the songs has been
adopted in the manuscripts, most of which are of recent date
(i.e., no more than a few centuries old). The printed texts have
adopted this notation in which the notes are referred to with the
help of the numerals 1, . . . ,7 which designate pitches in their
descending order, again with intervals of roughly a second.
There are seven notes in the Sâmaveda according to these
 Mantras by Chance 299

manuscript notations, which is consistent with some musicolog-

ical accounts (e.g., that of the Näradasiksä).
In the grämageyagäna of the K-R school, our verse is listed as
item 468, and it is provided with nine melodies to which it is
sung. The first of these, referred to as " G G 4 6 8 . 1 " , is the
following:
1r 2 1 2 1 2r 1 2 2
svädäisthäyä/ madâisthâyâ / / pavasvaso / mädhä' 1 rä'23 yä / /
1 2 1 2 1
indräyäpä / / taväisü'23 tä'343h / o'2345 i / / da / / (24)

In order to know how this actually sounds, a few more

conventions must be explained: " r " denotes lengthening; the
numerals in the line sound the same as those written above the
syllables, each lasting one beat or time unit (mäträ)\ they also
induce lengthening of the preceding syllable. The song consists
of five portions that are separated by double bars ('7 / " ) , further
subdivided into smaller portions separated by single bars ( " / " ) .
The number of these latter, smaller subdivisions varies, but the
former subdivision into five is constant in an important class of
rituals: these five portions, called bhakti, are of fundamental
importance in the Soma rituals, where they are chanted by
different priests facing different directions, sitting in a particular
fashion and following a pattern that is always the same (see
A G N I 1:608-609). The five portions, which should each be sung
with a single breath, are called: (1) prastâva ( " p r e l u d e " ) , (2)
udgltha ( " c h a n t " ) , (3) pratihära ( " r e s p o n s e " ) , (4) upadrava
( " a c c e s s o r y " ) and (5) nidhana ("finale"). Three of the four
Sämaveda priests participate in such a chant, viz., Prastotâ,
Udgätä, and Pratihartä. The assignment of bhakti portions to
them is as follows:

1) prastâva by Prastotâ;
2) udgltha by Udgätä;
3) pratihära by Pratihartâ;
4) upadrava by Udgätä; and
5) nidhana by all three.
300 Mantras

When we compare (24) with the original verse (18), we

observe that certain syllables have been expanded, or otherwise
modified. For example, many of the short a ' s are lengthened into
long <f s, and / and e have become cd. The latter modification is
well k n o w n to linguists: it is the famous vrddhi("lengthening")
discovered by the Sanskrit grammarians that is referred to in the
first sûtra of Pänini's grammar. At the same time, the notion of
vrddhi is one of the cornerstones of Indo-European comparative
phonology, as well as the basis of our notion of " s o u n d l a w . "
This illustrates in passing that the derivation of the ritual chants
of the Sâmaveda from the Rigveda (and not only the formation of
the Padapâtha from the Rigveda, as V. N . Jha has shown) has
contributed to the origin of the Sanskrit grammatical tradition.
These phonetic or sound modifications are often determined,
or partly determined, by the formal nature of the syllables of the
original, in particular their length; the structure of the song is
therefore related to the original meter which is also based upon
the distinction between short and long syllables. These relation-
ships are dealt with in texts such as the (K-R) Puspasütra, which
makes use of grammatical technical terms such as vrddhi and a
great many others. Song (24) is also called a gâyatrî song—a
circumstance to which we shall return.
We finally observe that in (24) new syllables have been added,
in particular the termination of the upadrava and all of the
nidhana:
1 ; .
o'2345 i / / dâ //

This is a stobha of the familiar type, but properly expressed

together with its musical structure.
In order to study the ritual significance of our song we need to
be familiar with the structure of the entire system of its deriva-
tives in the corpus of the Sämaveda. Instead of pursuing this by
numerical references to the published texts of the K-R Sâmave-
da, which would be intelligible only to specialists, it will be more
interesting and worthwhile to publish songs from the cofpus of
the Jaimintya. Since these songs have never been published, I
 Mantras by Chance 301

have made use of recordings and of a copy of a manuscript in the

Malayalam script prepared by my collaborator Itti Ravi Nambu-
diri. The original (see page 209) is in the possession of Asko
Parpola (University of Helsinki). Two other manuscripts of the
Jaiminîya songs are known: they are B 497 and B 61-62 of the
India Office Library, both discovered by A. C. Burnell in the
1870's in the Tamil region and both written in the Grantha script.
I have mentioned the most important variant /'readings" (re-
ferred to as ;' "C") that occur in a handwritten copy completed by
Willem Caland in 1906, which was based upon R497 and collated
with B 61 and B 62. (For the Aranyageyagâna, Caland did not
note the stobhas or the muscial syllable notation given in B 62).
A photograph of Caland's manuscript was in the possession of
Dr. A. A. Bake at the School of Oriental and African Studies,
and was given to me by Mrs. Bake. The original is in the
University Library of Utrecht. As the reader will see, some of
the variants help to establish the boundaries between the bhakti
portions. There are still instances where these remain unclear in
the manuscript tradition, and therefore must be determined from
oral tradition. This is easy to do because of the rule that one
bhakti has to be chanted with one breath (page 280). If the
number of bhakti portions is different from five, the method of
chanting has to be further specified.
Here follow the nine songs based upon the gâyatrï verse from
the Grämageyagäna:

Jaiminîya Grämageyagäna 49.2.1-9

(1) svädäyisthäyä madäyisthäyä / pavasva somädhäräyä / indrâyâpâ /
taväyi sütäfr / oyilâ / /
(2) svädisthayä iyâ iyâ madisthäyä /pävasva so iyä iyä madhäräyä /
indräyapä iyâ iyâ /taväyi sütäfr / oyilä / /
(3) svädisthayau ho vä iyâ madisthâyâ / pavasva sau ho vä iyä
madhäräyä / indräya pau ho vä iyä /taväyi sütäh / oyi|ä / /
C: poho
(4) oyi svädi / sthayä mädisthäyä vuvovä / pavasva somä dhäräyä o indrä /
yä pä au ho vä / tave sütäji / /
302 Mantras

(5) uhuvâyi svadï/ sthayä madïsthâyâ au ho vä / pavas va soma dhârâyâ

uhuvâ indräyäpä / tävä au ho va / sütäh II
(6) svâdâyisthayâ madisthayâ pâvâsvâ somadhârayâ âyindrâyâ pâ hâ bu /
tavâ yi sutâ bu / va / /
C: madisthayâ / somadhârayâ /
(7) svâdisthayâ madisthayâ/pavasva somadhârayâ / indrâyapâ 7 tavâ ü tavâ ü
tavâ vu va au ho va / sütäh / /
C: somadhârayâ /
(8) au ho hm bhâ e hiyâ hâ hâyi svâdâyisthayâ mâdï o yi mâdï / au.ho hm bhâ
e hiyâ hâ hâyi sthayä paväsvä so o svä so / au ho hm bhâ e hiyâ hä häyi ma
dhäraya indrä o indrâ /au ho hm bhâ e hiyä hä häyi ya pä taväyi sütä oyi
sütah / au ho hm bhä e hiyä hä hä au ho vä / î / /
(9) svâdisthayâ madäyisthayä / paväsvä soma dhäräyä / ä yindräyä pätavä
hä vu vä/ sütah / /

A few observations may be made. First of all, the Jaiminïya does

not use a numerical notation to refer to the pitch of the tonal
pattern which characterizes the melodies. There exists a
Jaiminïya musical notation by means of syllables from the
Grantha script, of which we noted that Caland copied it from
B 497 but not from B 61-62. It is not marked in the Malayalam
manuscript.
Secondly we observe that the differences between the
Jaiminïya and the K-R version (above (24)) are phonetic: K-R
often has " ä i " where the Malayalam manuscript of the Jaiminïya
version has " ä y i ' \ although Caland's manuscript also has " a i . "
This is not due to differences between the Malayalam and
Grantha scripts, because both of these can express " a i " as well
as " ä y i . " These minor phonetic differences are determined not
by differences of affiliation but by geography (Tamilnad versus
Kerala). Important differences, however, are due to the Vedic
school. For example, the K-R h a s o f t e n " o " where the Jaiminïya
has " a " ; and the well known K-R stobha " o i d â " appears in the
Jaiminïya tradition always as " o i l â , " using the Dravidian sound
' T \ pronounced like r in American English, but further back
(see Renou 1947:99; Staal 1961:69). The occurrence of this sound
may indicate that the Jaiminïya tradition originated in South
 Mantras by Chance 303

India or in an area of the North at a time when a Dravidian

language, w,aç still spoken ttfcere.
The third observation was made in A G N I , from which I quote
(1:278):

When dealing with these songs and chants it should be remembered

that the Sämaveda is replete with what, from a textual point of view, are
unexpected variations and varieties. These features are characteristic of
the Sämaveda, Kauthuma-Rânâyanïya as well as Jaiminlya. Whenever
regular patterns seem to emerge, there are new deviations that break the
pattern. To treat the text as if it were corrupt would be to miss its very
raison d'être. But even if we accept its playful deviations, we find that
the rules of this game often escape us. Many forms that may seem to be
printing mistakes or mistakes of the manuscripts are therefore in fact
what they should be.

The seventh of this sequence of nine songs is used in its

Grämageyagäna form in the Agnicayana ritual. We have studied
other chants from the same ritual context (page 217). After the
bird altar has been consecrated, it has to be brought under
control and pacified. The Adhvaryu, assisted by the Pratipras-
thätä, pours a continuous libation of goat milk over the western
brick of the northern wing. This brick is chosen because it is
relatively soft and tender, far away from the center of the altar
which is a center of power, and also because it can be easily
approached from different sides (AGNI 1:509 sq.). During this
oblation, the Adhvaryu recites the famous Satarudriya or Ru-
dram (Taittinya Samhitä 4.5), which derives its popularity partly
from the fact that it was subsequently interpreted within the
perspective of Saiva theism (Gonda 1979; Arnold, forthcoming).
During this oblation and recitation, the Udgätä sings a sequence
of 57 sämans, together called Flow of Milk (kslradhärä). The
fourth of these is our GG 49.2.7.
Why was this song chosen to perform this function? It fits with
the others only to some extent: it shares with most a triple
repetition (of " t a v ä ü " ) , and with many the stobha " a u ho v ä . "
Both these features, however, are quite common in the
Sämaveda. The more complex features that distinguish a series
of later songs in the Flow of Milk sequence (Chapter 1.9) are not
304 Mantras

found in it. One is tempted to believe that there is no answer to

the question of why this particular song was selected for use at
this particular point of the ritual. It came from the collection like
a seed that falls from a blossom and is carried through the wind
until it settles down somewhere.
What about the stobha " t a v ä ü " itself? Again, we observe a
phonological, and therefore linguistic correspondence—or is it
pseudo-linguistic? " T a v ä ü " comes from " - t a v e , " which should
yield, in linguistic terms, " t a v â i " or " t a v â y i , " as indeed it does
in many other songs of the sequence GG 49.2. What we find,
therefore, is a variation of a linguistic relationship. This is quite
common in these songs, and it supports the idea that the
derivation of songs from verses is closely connected with the
origination of linguistics in India.
In Sanskrit, "tava\* also has a meaning: it means " y o u r s . "
The stobha can be taken to mean " y o u r s U , " which is almost as
good as " y o u r s t r u l y . " But this meaning correspondence is
adventitious and without significance, which does not imply,
however, that it could not evoke semantic associations among
some users. " T a v a , " moreover, occurs in other stobhas (see
Puspasütra, ed. Simon, 1909: 770 s.v. tava tyad), and " ü " is
quite famous in later Tantrism (see, for example, Padoux 1963:
203-206). Historical connexions between Vedic stobhas and
Tantric brjamantras are likely to exist (Chapter 20): a good
example is the famous Tantric phat which occurred already in
the Sämaveda (see van der Hoogt 1929:99; cf. phät phat phat
phat phat phat\ A G N I 1:416).
Next we shall consider some of the more esoteric " s o n g s to be
sung in the forest." Jaiminiya Aranyageyagâna 15.7 (corre-
sponding to K-R AG 16.1) includes the following song based
upon our verse:

ayâmâyâm ayämäyäm ayäniäyäm svädisthayä madäyisthäyä / pavasva

somadhäräyä / indräyapä taväyi sütah / ayâmâyâm ayämäyäm ayämäyä
au ho.vä / ï / /
C: ayämäyämayämäyä (2x), ayämäyämayämäyämayämäyämauhovä
 Mantras by Chance 305

Again, there is triple repetition, and k k ayam" also happens to

mean something, viz., "this o n e , " without carrying any deeper
significance. It is a well known Vedic stobha (see van der Hoogt
1929:11: ay am yah, 91: ay am ayâyam, 99: ay am vâyau, 111:
ay am) and is perhaps the source of the equally well-known
Tantric stobha " a i m " (see, for example, Bharati 1970:119).
This song is ritually used in the'construction of the Agnica-
yana altar. When the bricks of the first layer are laid down the
Udgätä chants whenever it is the turn of a ritually significant
brick to be deposited and consecrated. At the consecration of
the seventh brick, called Dürvä after the grass of that name, the
Adhvaryu recites Taittiriya Samhitä 4.2.9.2 c-d (see AGNI
1:423):

"Rising up from every stem, from every joint, dürvä, extend to us a

thousandfold, a hundredfold.
You who extend with a hundred, arise with a thousand!
To you, goddess brick, may we offer with oblation."

At the beginning of this recitation the Adhvaryu looks at the

Udgätä, to alert him, and the Udgätä intones his song, AG 15.7.
Why this song is chanted here at this time is not known. There
is no specific phonetic, phonological, syntactic, semantic or
pragmatic similarity between recitation and chant, apart from
the fact that the recitation is again in the gayatrï meter and the
chant is a gayatrï chant—both varieties that are exceedingly
common.
N o specific semantic connexion appears to exist between the
ritual applications of these two derivatives from our Rigvedic
verse that occur in the two basic gäna books of the Sämaveda.
Other ritual uses of chants that are derived from these songs
pertain to the Soma ritual and are listed in the Üha and Ühya (or
Rahasya) Gäna collections of the Sämaveda. In these two
collections, the songs are ordered in the same sequence in which
they are used in the Soma rituals, and they are given in the ac r ual
forms in which they are sung.
The ritual context of the Soma ceremonies has been treated in
Chapter 8. In the Agnistoma, there are three pressings of Soma;
306 Mantras

at the third pressing, there are two sequences of chants and

recitations; the first of these, which is the eleventh sequence of
the entire pressing day, consists of a series of seventeen songs,
of which the first and the fourth are based upon our verse. These
songs are listed in Ühagäna 3.2-6 (see A G N I 1:646-648), but
there are several difficulties, notably one remarked upon by
Caland (Caland-Henry 1907:180): the first three songs are not
listed there (though their textual form, without melody, occurs in
the corresponding passages of the Samhitä). So how should they
be sung?
The answer is that these are to be sung in the Gäyatra melody,
which is explained somewhere else, once and for all, but not in
the corpus of the Sämaveda. The same holds for other verses in
the gâyatrï meter, in particular the famous Gâyatrî verse of the
Rigveda (3.62.10: see above, page 231) that is not listed in the
Sämaveda either. It is often quoted in Vedic literature, some-
times in the context of a wish for inspiration (see, for example,
Gonda 1980a:45, 104); and is recited during many rites and on
many occasions (see Gonda 1980b, s.v. " S ä v i t r i " ) . According to
the Satapatha Brähmana (2.3.4.39), all wishes of the ritualist will
be fulfilled by this mantra because they are impelled by Savitr,
who is the " i m p e l l e r " (prasavitr) of the gods. This is a typical
comment o f t h a t Brähmana—vacuous and ad hoc.
Why was this mantra picked to play such an important and
auspicious role? There are hundreds of mantras in the Rigveda
that say something similar. The answer to such questions is
always the same: there is no answer. It fell from the heap like thé
windblown seed we have already encountered. For some arbi-
trary reason, supreme significance is attached to it. It occurs
everywhere, with one exception: it does not occur in the
Sämaveda. Or rather, it is placed at the beginning in some
collections (in some manuscripts and in most of the printed
editions), remaining outside the systematically numbered se-
quences and all classifications, while in other collections it is
simply not found.
The reason for this special treatment, once the mantra had
acquired its special function, is not difficult to find: this mantra
 Mantras by Chance 307

had too much meaning. A prayer for inspiration that is daily

recited, that defines a brahmin and distinguishes him from
others, is not the kind of mantra that provides the auspicious but
meaningless sounds that make up the melodies that accompany
ritual activity. The Sâvitrï, in fact, is so important to a brahmin
that it has spilled over to other twice-born castes, albeit in a
different form: Ksatriyas have a Sävitri in the tristubh meter, and
Vaisyas have one in the jagatï meter (see Malamoud 1977:89).
The Gâyatrï thus has meaning as well .as social significance—
unlike most mantras and other ritual chants—-although this
significance is partly theoretical, like much that concerns the
ancient varna system (the theory of four castes, set up in the law
books, as distinct from the reality of thousands of jäti units of
caste which characterizes Indian society). The Gâyatrï, there-
fore, remains significant and eo ipso isolated.
The Gâyatrï melody occupies a similar position in the cere-
monies of the Soma ritual. Many of the stuti or stotra chants,
whose sequences define a Soma ritual, are set to this melody.
They all receive the same treatment, a treatment which in fact-
obliterates-almost-all the meaningful characteristics of the orig-
inals, replacing them by stobhas and other meaningless sounds,
particularly long o ' s . Since only the prastäva retains the original
text, it is only that portion that is explicitly noted in the tradition.
The other bhakti portions are always sung in the same manner.
The udgïtha is always (see above, page 232):
ö vä ö vä ö va him bhä ö vä,

with very long ô ' s (note that South Indian scripts, unlike Nâgarî,
distinguish between short o and long o). The remaining bhakti
portions are collectively treated as follows. The Pratihartä
always sings " h i m " together with the Udgätä, breathes in, and
chants "vak"''(which comes from nowhere) while he holds his
breath—therefore almost inaudibly. The Yajamäna and some
other priests, in accordance with complex rules, should also
chant " o " (see A G N I 1:603).
We are now in a position to return to the ritual uses of our
verse, Rigveda 9.1.1, in the Soma ceremonies. We can now
308 Mantras

understand why, for the first song of the third pressing sequence
of chants, only the prastäva needs to be known. It is listed in the
Samhitä as follows:

svâdisthayâ madisthayöm (Jaiminïya Àrcika 64.1).

The fourth chant is listed in Üha Gäna 3.2. It preserves more of

the original text although the udgitha has also disappeared:

svädisthayä ma dâ yisthayâ / svâ so ö /â yindrä / ö pä tavä hâvu va /sü

tah /'/. (see AGNI 1:646).

In all the cases of ritual application of mantras we have so far

studied, nothing remains of the rich literature of the Vedas but a
collection of sounds and syllables. Entire passages that orig-
inally were pregnant with meaning are reduced to long d ' s . This
is precisely what distinguishes mantras from the original verse:
to be made into a mantra, and thus fit for ritual consumption, a
verse has to be subjected to formal transformations, operations
that apply to form but not to meaning. This is why Renou
referred to mantras as "poussière v é d i q u e / ' and why he men-
tioned ' i e découpage des vieux hymnes en formules ou même
fragments devenus des corps inertes dans la trame liturgique"
(1960:76-77 quoted by Malamoud 1983:33).
Ritual traditions have obvious social significance in that they
identify groups and distinguish them from each other. They give
people, in that hackneyed contemporary phrase, " a sense of
identity.' v That identity, however, is often due to distinctions
that rest upon meaningless phonetic variations. Thus the
Jaiminïya and Kauthuma-Rânâyanïya schools differ from each
other by such characteristics as vowel length, or because the
former uses 4 t ä " where the latter uses t k o . " Up to the present
time, the Vedic schools themselves are distinguished from each
other by such variations of sound that can be more easily
explained in grammatical than in religious terms. The Gâyatrï
mantra itself is pronounced differently by Nambudiri brahmans
 Mantras by Chance 309

belonging to the Yajurveda o r Sâmaveda and Nambudiris be-

longing to the Rigveda: the former pronounce the visarga " h " at
the end of " n a h " (p. 231) as an "f," a sound that is generally
believed not to occur in Sanskrit.
Actually, the bilabial spirant ' f ' is an optional variant pre-
scribed by a grammatical rule which also introduces another
sound, the velar " k h " , which is believed to be absent from
Sanskrit. Special symbols are needed to add these sounds to the
alphabet, as in the manuscripts or printed editions of Pänini's
rule 8.3.37: kupvoh X kÇ9pau ca, " X and Ç9 may be substituted
in the place of the visarga when a voiceless velar or a voiceless
labial, respectively, follow."
In this symbolism, " X k " r e f e r s to the velar " k h " , and " Q 9 p "
to the bilabial spirant "f". The interpretation of this rule
depends on conventions that have been established elsewhere in
the grammar. Thus we have to supply padasya, " i n the place of
a w o r d , " from 8.1.16; samhitäyäm, "in continuous pronuncia-
t i o n , " from 8.2.108; and va, " o p t i o n a l l y , " from 8.3.36. That the
rules are to be understood in this order follows from 8.2.1, and
that the optional vä means " p r e f e r a b l e " has been shown by
Kiparsky (1979). That the Yajurveda and Sâmaveda have
adopted the preferred form is in accordance with the central
place they occupy in the ritual tradition.
In the Gâyatrï, the voiceless labial which follows the visarga
" h " is the initial " p " of " p r a c o d a y â t . " Thus the variation
between Vedic schools and between the Vedas themselves is
reduced to the optionality of grammatical rules. This is not a
modern phenomenon; it goes back to the origin of the tradition.
The bilabial spirant "f" occurs not only in the grammar of
Pâpini, but in two phonetic treatises (Prätisäkhya) of the Yajur-
veda, namely, the Taittirïya of the Black and the Väjasaneyi of
the White Yajurveda. (It is lacking again in the Mâdhyandina,
another school of the White Yajurveda: Renou 1942, 1957:396).
Throughout this section we have observed that considerations
of form and formal derivations and transformations are foremost
in the minds of the ritualists. In some cases these forms appear
to be purely arbitrary, since they do not correspond to any type
310 Mantras

of formal relationship that is known or seems to make sense. In

other cases the rules of grammar, especially those that introduce
options, are used and lead to further ritual developments and
proliferation. In both cases, we can give a syntactic but not a
semantic account of the relationships between mantras and acts.
We shall formulate such an account for Rigveda 9.1.1 by means
of a syntactic structure that represents the knowledge of the
ritualist. We have to make some distinctions to begin with.
The ritualist, first of all, knows this verse. That is, he knows
how to recite it, for this is what he learned when he learned the
Rigveda by heart as a boy—a knowledge he shares with many
other brahmins who also belong (by birth) to the Rigveda but are
not ritualists. But there is no traditional transmission of the
meaning of the verse, and so he does not necessarily know its
meaning. If he does, he must be a Sanskrit scholar to whom
meaning has become a matter of personal interest. A practising
ritualist who is asked for the meaning of a recitation or song
invariably replies: go to a scholar of language or philosophy.
A brahmin who belongs to the Sämaveda knows more: he
knows how the verses of the Rigveda have been transformed
into songs. H e will probably know various chants derived from
that verse, depending on his expertise, and if he is really good,
he will know where these occur in the collections of the
Sämavedic corpus—which is large and complicated, as we have
. seen. • H e shares such knowledge, however, with other
Sâmavedins and it still does not qualify him as a ritualist.
Only the ritual expert or vaidika knows, unlike an ordinary
Rigvedin or Sämavedin (or Yajurvedin, for that matter), the
association between mantras and acts, and when and where in
the ritual these have to be inserted and executed. He knows
something like the following structure (I say "something l i k e , "
for he uses, as we shall see in Chapter 27, a system of reference
rooted in oral tradition and in some respects different from the
system used by Western scholars, which presupposes the avail-
ability of printed texts):
 Mantras by Chance 311

LANGUAGE Rig veda 9.1.1

MANTRA Grämageyagäna 49.2.7 Aranyageyagâna 15.7 Arcikä 64.1 Uhagäna 3.2*

t i
Stuti I Stuti IV

ACT Flow of Milk 4 Layer \, Brick 7 Sequence 11 Sequence 11

Such a structure makes sense only within the larger pattern of

hundreds of similar structures, which together constitute the
edifice of the ritual. Vedic ritual makes sense, but it is structural
sense. An exhaustive synchronie analysis of the ritual can be
given in terms of the structural relationships between such
forms, without reference to meaning or external function. Such
an analysis will refer to the rules that relate structures to each
other, and is therefore syntactic in nature. If and when the
system changes (which may be due to a variety of causes,
semantic as well as nonsemantic), the rules or the relations
between them change, and this can again be studied in syntactic
terms. The syntactic approach can therefore completely account
for the ritual facts. It also shows that many cases of association
between a mantra and an act are accidental and due to chance.
I shall refer to this structural feature as inner chance.
There are other developments in the domain of ritual that
illustrate the workings of chance. Many of these are connected
with the origins of ritual systems. The study of such systems has
been undertaken by anthropologists, who have paid insufficient
attention to mantras and history and whose orientation has been
predominantly semantic. We have to study such semantic ac-
counts, however, because they often provide the only informa-
tion that is available, and also in order to find out whether the
case of Vedic ritual is perhaps unique (see below, Chapter 30C).
I shall discuss three anthropological studies in the next Chapter
and return to the question of chance afterwards.
 25

Anthropology Without Mantras

The three anthropological studies we will study in this chapter

are concerned with Southeast Asia and all have links with India.
The three geographical areas are Thailand (studied by Stanley
Tambiah), Sri Lanka (studied by Gananath Obeyesekere) and
Bali (studied by Clifford Geertz). These three authors differ in
the degree of their semantic involvement: Tambiah possesses
the least (not surprising in view of his earlier studies) and Geertz
the most. Tambiah and Obeyesekere pay attention to mantras
which Geertz hardly mentions. I was tempted to characterize the
three approaches as " W e a k S e m a n t i c s / 1 " M e d i u m S e m a n t i c s / '
and " S t r o n g S e m a n t i c s / ' but this could be taken as a criticism,
especially of the weak, which is not what I had in mind. I have
therefore accepted the lead of these authors themselves and
called their approaches, respectively: "Object S e m a n t i c s / '
" P s y c h o - s e m a n t i c s / ' and "Thick S e m a n t i c s . "
In each of these case studies I shall attempt to approximate, at
least in principle, that (practically as well as philosophically)
impossible ideal demanded by Franz Boas: complete the data
before embarking upon theory. Of course, I cannot be exhaus-
tive; I am only interested in illustrations. However, in the case
of Tambiah and Obeyesekere, this procedure can to some extent
be realized because their data are presented independently of
theory, and a r e / m o r e o v e r / s i m i l a r in kind. I shall therefore start
with the data of these two authors. The theories of Tambiah and
O b e y e s e k e r e / o n the other hand, are very different, and so I shall
save them for last. In the case of Geertz it is almost impossible
to draw the line between facts and interpretations. This will
314 Mantras

require special attention and since Geertz has also written the
most, his section will be the longest.

25A. OBJECT SEMANTICS: Stanley Tambiah

Popular prejudice would expect Tambiah's book The Buddhist

Saints of the Forest and the Cult of Amulets (1984) to contain no
mantras and little on ritual. For did the Buddha not condemn
both?
Popular prejudice, of course, is not a good guide, even though,
by definition, many people follow it. Both Buddhism and Jainism
rejected Vedic ritual, but neither rejected ritual (see Heesterman
1964:29; reprinted 1985:42), We know, moreover, that
Mahäyäna Buddhism, which is not simply " l a t e r " Buddhism (as
a previous generation, reared on Rhys Davids, imagined) but
which is in many respects as close to the lost " o r i g i n a l "
Buddhism as the Theravâda, is replete with ritual (for refer-
ences, see pages 399-400, 4U-r412). The same holds for the
Theravâda itself, even in its contemporary forms. Spiro (1972;
1982:191) observes: " I n d e e d , for those who conceive of the
latter form of Buddhism as a precursor of modern humanism (or
even of analytical philosophy!), its rich ritual system comes
rather as a shock, or is viewed as a late degeneration of ritualless
purity."
Spiro's observations derive in part from his fieldwork in
Burma. Roughly a fifth of his book is in fact concerned with
Burmese Buddhism as a ritual .system. Tambiah's book is partly
based upon his fieldwork in Thailand, and although his treatment
is more specific and sophisticated than Spiro's, it confirms the
latter's impressions as formulated in the above quote. Neither
work is confined to what is labelled " B u d d h i s t , " and they
include local Burmese and Thai traditions, sometimes referred
to as " a n i m i s t . " Spiro has tried to keep these two trends apart,
but Tambiah has convincingly shown that this cannot be done.
As for the term "saints of the forest," which occurs in Tam-
biah's title, it is not a product of the author's imagination, nor is
it Thai or " a n i m i s t : " it derives from the Pali ärannaväsi or
 Anthropology Without Mantras 315

vanavâsï, "forest dweller," which contrasts with gämaväsl,

"village dweller" or nagaravâsî, " t o w n dweller" (Tambiah
1984:53). We recognize here the same distinction encountered in
the previous chapter (page 297) between grämageyagäna,
" s o n g s to be sung in the village," and aranyageyagäna, "songs
to be sung in the forest," a distinction that is quite common in
Vedic tradition (see, for example, the Äranyakas or "forest
b o o k s , " a class of works to which the classical Upanisads are
generally attached as a kind of appendix; cf. also Malamoud
1967). Buddhist tradition has preserved the terminology, al-
though nagara, " t o w n , " has been added, reflecting the progress
of history and increasing urbanization. The latter development
was more important for the growing Buddhist order than for the
successors to the Vedic tradition. Even so, though the label is
" B u d d h i s t , " we are dealing with an Indian tradition.
The distinction between " f o r e s t " and " v i l l a g e " is, of course,
related to the distinction introduced by Louis Dumont (1959,
1966) between the renouncer and the man-in-the-worldor man-
of-caste (cf. also Heesterman 1964, reprinted 1985:26-44). One
might expect that the renouncer of village, family and caste has
also renounced ritual; but this is refuted by many facts, including
two that we have encountered: the "songs to be sung in the
forest" are also ritual songs; and Buddhist monks, who are
renouncers, also perform rituals. In Buddhism, the exemplary
outcome of a renouncer's experience, we find D u m o n t ' s distinc-
tion not only obtaining between monks and laymen, but also,
within the community of monks themselves, between the
" f o r e s t " and "village/town" monks. It might be supposed that
the former operate on a level that is beyond ritual. But the forest
monks are in fact given to " p r a c t i c e " (patipatn) which includes
ritual as well as meditation, and meditation itself is closely
related to both recitation and ritual, as we shall see in a moment.
Since the "village/town" monks also perform rituals—albeit
different ones—we have come full circle.
Thai Buddhism is replete with mantras, and they have left
their traces in Tambiah's book. For example, Achar Man, the
saint whose life in the forest is graphically described by Tambiah
316 Mantras

in a long and fascinating chapter, used a system of mantra

meditation that was adopted by Acharn Fun and other leading
disciples (Tambiah 1984:139). The amulets, which play a very
important role in Tambiah's book, derive their power from the
fact that they have been consecrated, which is also done in the
time-honored manner by ritual and mantras. The consecration
ceremony, attended by the supreme patriarch of the Buddhist
sangha and by representatives of the government (the deputy
prime minister and the minister of defense, no less), begins with
offerings made to the shrine deities with the chief court brahmin
presiding. (I cannot expatiate on these Thai brahmins who are
marginal to Tambiah's subject; but they add another Indian
element to the amalgam of Buddhist and Thai traditions; cf.
Filliozat 1965 and Sarma 1972.) ' T h e r e a f t e r , the most important
sequence was staged. While 4 monks at the time, sitting on the
side, chanted paritta chants continuously, 72 senior m o n k s ,
divided into three batches of 24, sat for four hours at a time
meditating silently and transferring virtue to the pile of amulets
by means of a cord that passed through their hands and was
attached to the pile . . . T h e monks sit cross-legged, close their
eyes, practice concentration, and say words mentally and si-
lently 9 ' (pages 244-245).
Transfer of power by means of a cord predates the invention
of electrical wiring. It is a common ritual theme that we find, for
example, in the Agnicayana consecration where a string made of
hemp links the Yajamäna to the ukhä pot (AGNI 1:323—324; Plate
46). What is of interest in the present context, however, are
paritta chants and silent meditation. The former term, which
means " p r o t e c t i o n , " refers to the chanting of Buddhist scrip-
tures as charms and for the sake of exorcism. This was intro-
duced to Sri Lanka in the fourth century C . E . (Lamotte
1976:1860) and is still extremely popular. As for meditation, it
might be supposed that it is something rather different from
recitation of chanting—something more " s p i r i t u a l " than "rit-
u a l . " The answer is provided by Tambiah in the above passage:
meditation is not some sort of idling in emptiness, but "saying
words mentally and silently." Read " m a n t r a s " for " w o r d s , "
 Anthropology Without Mantras 317

and the situation becomes indistinguishable from that of Vedic

ritual where silent recitation plays a very important role (see,
e.g., Renou 1949; Renou and Silburn 1954; above, page 263).
The details of these meditations are not described by Tam-
biah, and it is unlikely that considerations of form play as
important a part in them as they do in Vedic ritual. Yet they
deserve much closer study for they are at the core of the
consecration that gives life to the amulets.
The consecration of amulets is a special case of the consecra-
tion of a Buddha statue, or Buddhäbhiseka, which Tambiah
describes on pages 248-257: ' T h e climactic sequence is the
chanting by the four monks of the gäthä buddhäbhiseka (verses
of consecration) . . . The four monks recite a particular cycle of
verses, and each recitation takes about thirty-five minutes . . . "
(Tambiah 1984:249).
I submit that these " p a r t i c u l a r " ritual recitations, which
constitute the core of the consecration, which itself is the
essential step in providing the amulets and Buddha statues with
the powers that are the subject of Tambiah's book, have a
complex structure that can be exhaustively described in syntac-
tic terms. Moreover, I would suggest that such recitations are
the heart of the matter, not only in the eyes of the participants,
but also for any scientific analysis that succeeds in making sense
of such cults.
Before we turn to theory pur sang I must refer to a contro-
versy that raged in the pages of The Times Literary Supplement
and to which my attention was drawn by Michel Strickmann.
The controversy is not only entertaining, but relevant in the
present context because it is concerned with the relationship
between philology and anthropology, and the* study of ritual
requires the assistance of both these disciplines. We often find
classical scholars and historians complaining when anthropolo-
gists and other social scientists ignore or neglect the historical
background of the society in which they are working. Tambiah
cannot be faulted for doing this; on the contrary, he has taken
pains to inform himself about the historical and doctrinal back-
ground of Buddhism, a subject that he rightly regarded as
318 Mantras

indispensable for the understanding of his Buddhist monks, but

with which he was not professionally familiar. N o w , Buddhism
is an enormously complex subject; even if one confines oneself
to the Theraväda (a confinement that is definitely unnatural), it
requires sustained and concentrated study for a goodly number
of years. We should respect Tambiah for his efforts to go beyond
his professional field of expertise, where his fieldwork appeared
to demand it; and one should not be surprised or upset when,
occasionally, he errs.
Not so Richard Gombrich, an Oxford Sanskritist who has
himself done fieldwork in Sri Lanka and written a book about
Sinhalese Buddhism (Gombrich 1971). In a savage review (Gom-
brich 1985) he pins on Tambiah a long list of errors, mostly
pertaining to spelling, languages and history. Tambiah has
replied to this charge, and turns out to be right in many cases
(though least often in matters of history where he or his
proofreader has incurred Gombrich's wrath by twice replacing
" A . D . " by " B . C . " ) . Most of the items on this laundry list of
errors and alleged errors are marginal to the chief concerns of
Tambiah's book. Some are of general interest, however, and are
also related to our present context. Among these is a discussion
on the traditional class of extraordinary powers (siddhi, iddhi,
abhijna) found in the Yoga, Hindu as well as Buddhist (cf.
above, page 237). One of the most colorful of these is lévitation,
and Tambiah has a story to tell about a pilot of the Royal Thai
Air F o r c e , w h o , on a practice flight over the Mae Pang moun-
tain:
saw a monk sitting in meditation upon a cloud, and (he) had to practice
much skill in swerving away to avoid hitting the apparition. Having
grounded the plane, the pilot with his flight map in hand went scouting on
the mountain, and there,he recognized Luang Pu Waen as the monk he
had seen in the clouds. The news of Luang Pu Waen's power of lévitation
and flying spread quickly—it was broadcast on the radio and splashed in
the newspapers. And the public's demand for his medallions soared
(Tambiah 1984:272).

Gombrich accused Tambiah of attributing to the Buddha the

view that such powers are commendable, to which Tambiah
 Anthropology Without Mantras 319

replied that he had only followed the descriptions of Buddha-

ghosa's Visuddhimagga which Gombrich himself had acclaimed
in his book as " t h e unitary standard of doctrinal orthodoxy for
all Theraväda B u d d h i s t s . " The Visuddhimagga belongs, as a
matter of fact, to the fifth century C . E . , and since the earliest
accounts of the Buddha describe him " o n l y as a man, not as a
super-human b e i n g " (Nakamura 1980:83), we don't know any-
thing about what his opinion might have been with regard to such
special powers. What w e do know is that in the Yoga system
these powers were not frowned upon. Gombrich's opinion is
therefore based upon Western prejudice—the same prejudice we
find in Mircea Eliade (1964:223,401 and 1969:338) whose perti-
nent statements were refuted by Wasson (1968:331) followed by
myself (1975:97). This is ironic, for Gombrich himself has
demonstrated that Eliade's statements on Buddhism are riddled
with errors of fact and interpretation (Gombrich 1974). What is
instructive in the present context is that prejudice is more
pernicious than ignorance—a circumstance which is strikingly
relevant to the understanding of ritual and to which we shall
return.
The last part of Tambiah's book deals with "Conceptual and
Theoretical Clarifications" (pages 291-347). Tambiah does not
make much use of the ideas on performative analysis developed
in his 1979 lecture. True, he uses the term " i n d e x i c a l , " but it
does not make much of a theoretical contribution. The idea that
is most interesting in the present context is the "objectification
of charisma in objects and fetishes" which the cult of amulets
illustrates. Though W e b e r ' s theory allowed for the "routini-
zation of c h a r i s m a " in social institutions and social positions,
Tambiah notes that " h e was blind to the symbolism of objecti-
fication of charisma in objects and fetishes" (page 339).
Given such a process, semantic analysis is clearly applicable
and appropriate. It is these features that suggested my caption
"object s e m a n t i c s . " For if the introduction of semantics is
justified by anything, it is the use of objects that refer to
something else; and if amulets are not that, nothing is. Thus we
320 Mantras

have ample justification for semantics, albeit outside the domain

of ritual.
Tambiah also uses the expression "semiotic c o d e , " but it is
not clear what it can do that plain English cannot. More striking
is the description in his last chapter (entitled: " T h e objectifica-
tion of charisma and the fetishism of objects") of the relative
scarcity and ensuing commercial value of the Thai amulets. They
become "private possessions of laymen who expect to use the
amulet's potency to manipulate, overpower, seduce, and control
their fellow men and women in an ongoing drama of social
t r a n s a c t i o n s . " Tambiah applies to his amulets M a u s s ' expres-
sion of a "magico-religious guarantee of rank and prosperity"
(page 342). Actually, if one replaces in many of these passages
" a m u l e t s " by " d o l l a r s " one obtains an interesting theory of the
origination of capitalism. In this respect, Tambiah's work sup-
plements what we now know of the development of Buddhist
monasteries on the trade routes between India, Central Asia and
China into fortified constructions that served not only as motels
but also as banks (see Staal 1988b).
Tambiah's conclusions suggest a clarification of my own
position on questions of semantics. I do not reject semantics or
the use of symbols in anthropology or the study of culture. I only
reject the idea that these notions are helpful tools in the analysis
of ritual. The home of semantics is language, and semantic
analysis applies in the first place in contexts of language; then,
via language, it may also apply to mathematics, philosophy, art,
mythology and many other domains of experience. To say that
semantics does not apply to ritual is tantamount to saying that
ritual is not a language. To claim that all of culture consists of
symbols and meanings, and is in that respect language-like,
would simply entail that ritual cannot be part of culture.

25B. PSYCHO-SEMANTICS: Gananath Obeyesekere

Gananath Obeyesekere's Cult of the Goddess Pattini (1984)

provides a detailed description of a cult that still thrives in Sri
Lanka and South India and that has close connexions with
 Anthropology Without Mantras 321

Buddhism, Jainism, Hinduism and Sinhala tradition. The textual

traditions of this cult are embodied in a compendium of thirty-
five ritual texts or songbooks. Obeyesekere has translated most
of these texts, although the number 35 is an idealization that is
artificially maintained by emendations, splitting, addition and
elimination (page 32). H e has in fact provided translations of 58
different ritual texts that together occupy more than 150 pages of
his book. Many of these texts are sung and enacted in the course
of rituals. It is obvious that we are dealing with a very rich
tradition of rituals and songs, and that Obeyesekere has done
pioneering work in making it accessible. His interests in this
tradition is clearly semantic, and this is not surprising when one
places it in the various contexts in which he himself has placed
it with consummate skill.
Obeyesekere's book consists of 629 pages packed with infor-
mation and it would be impossible to do it any justice within a
short compass. In fact, the author says in his Preface (page xvii):
4t
My Pattini research spans the whole of my anthropological
c a r e e r . " I recall how I accompanied Obeyesekere once to one of
those beautiful isolated villages where he had already done a
great deal of fieldwork. That was in 1965, when I was recording
mantras, ranging from Buddhist upasampadä ordinations to
Sinhalese charms against snakes and cattle's foot disease. A half
day walk across a hill and through scattered jungle patches led to
a lush green valley of rice-fields, cocoa-nut palms and rubber
trees, with a river flowing at the bottom—a paradise not for the
people who live there, unfortunately, and not for the inquiring
mind, but certainly for the senses of any intrepid visitor who has
not yet lost them. And yet, despite abundant meaning and social
context, the semantic analysis of the Pattini cult may not be all
there is to it. There are definite indications that another approach
could contribute something that is still missing.
. Some of the texts are never sung or enacted, and Obeyesekere
has omitted most of these. But there are also cases where a text
is omitted because it does not seem to fit semantically. For
example, when the kapurâla's white robe is consecrated, Obe-
yesekere says about the accompanying song (text 11): "it r is not
322 Mantras

an important one in content, and I shall only quote excerpts from

i t " (page 111). Instead, he concentrates on the semantically
relevant contrast between the whiteness of the kapurâla's robe,
which has to be pure because he interacts with gods who are
pure, and the patched yellow robe of the Buddhist monk, which
in theory should be made of rags and which relates to " h i s
involvement in processes of death and decay (polluting activi-
ties)."
There are other similar cases where a song is omitted because
its content does not fit its ritual context, and where Obeyesekere
makes comments such as "this text is of little significance"
(page 240). The reader wonders why, in many such cases, the
songs are nevertheless sung. There are also examples that
display some degree of formal complexity. A text that was first
recited by the chief kapuräla continues to be recited by him and
another senior priest, while an assistant dances to the beat of
drums (page 189). The rhyme structure can be quite complex, so
that Obeyesekere comments: "This kind of complexity is a
function of greater specialization, literacy and book learning and
closeness to Buddhist civilizing influences" (page 190). Else-
where the audience intersperses the song with shouts (page 191 ;
cf. A G N I 1:642, and the pratigara interjections of the Adhvaryu:
above page 232 and AGNI 1:623, 640, 650, etc.). We also come
across "high-sounding combinations of Pali or Sanskrit, or both,
and Sinhala" (page 202). In all these cases, formal consider-
ations play a role, and a syntactic analysis might be able tç>
account for structures that the semantic approach fails to
explain. ,
There are numerous instances where the content of a song is
not clear or intelligible to the participants (e.g., page 239). The
identity of particular gods is often unclear or controversial (e.g.,
page 87). There is a mystery of twelve "missing d e i t i e s , " which
the author interprets as a survival of a gradually disappearing
cult (pages 285-293). Elsewhere, songs that are meant for a
particular occasion are sung on other occasions as well (e.g.,
page 221). Some songs are " n o t intrinsic to the ritual" (page
231); others (e.g., the "head-to-foot v e r s e " ) are " s t a n d a r d " and
 Anthropology Without Mantras 323

appear in various sections of the ceremony (page 96). Many

chants incorporate intricate numerical and metrical structures
(e.g., "twenty-one short stanzas and sixteen long s t a n z a s : " page
101). All these are cases of ritualization which can probably be
explained along syntactic lines. Phonological analysis would be
in order when studying the various modes of chanting, about
which Obeyesekere does not provide much detail although he
sometimes refers to style or mentions particular songs that are
sung at a faster or slower pace. In one context, an a is deleted,
because "it would introduce an extra syllable and upset the
rhyme scheme and r h y t h m " (page 89: " h e n c e sura is used
instead of asura"—a fact that should delight Sanskritists, as the
author hints).
It is likely that the songs of the Pattini cult are often more
closely related to the acts they accompany than is the case in
Vedic ritual. Some acts are very similar to Vedic rites. The
cutting of the milla tree, for example (pages 141-143), which
provides firewood for the fire-trampling rituals, is very similar to
such Agnicayana rites as the fetching of mud for the ukhâ pots
(AGNI 1:288-296) or the cutting of the hilba tree from which the
sacrificial pole will be made (AGNI 1:590-591 and Plate 96). The
extent of such similarities would further depend on the structural
relationships that obtain in the Pattini cult between the songs and
the acts, a topic that Obeyesekere has not enlarged upon. It is
clear that there are, in this instance, semantic similarities, and
likely that there are syntactic similarities as well.
The importance of form and formal structures is obvious to
anyone who reads between the lines, and it sometimes comes to
the surface. The most telling example is a rite that asks permis-
sion from the gods and forgiveness for accidental faults, which is
necessary because, as the kapuräla says, " t h e hand may go
wrong and the mouth may go w r o n g " (page 112). Rarely do we
find a more straightforward indication of the continuing empha-
sis not on meaning, but on act and recitation as the two main
vehicles of ritual activity.
What, then, has Obeyesekere to say on meaning, and what are
the theoretical observations he makes? It would be foolhardy to
324 Mantras

deal with a topic so large in the context of a book so rich in

information that is entirely new. However, I shall try to give at
least an idea of the author's two chief contributions in this area.
We should first of all note that Obeyesekere includes much
information and speculation on the historical background of
rites, and in so doing goes beyond the synchronie analysis of the
ritual system. About this he has some remarkable things to say—
remarkable not least because they apply to ritual as well as to
myth:

Any mythological tradition comes from historically diverse sources. A

tradition of myth is a composite of preexisting beliefs, beliefs that are
newly invented, and those incorporated from other belief systems. An
analysis of the sources of a given mythic tradition may help us unravel
the processes by which a religious tradition came to be constituted.
Anthropologists have been averse to a historical study of myth ever since
Radcliffe-Brown castigated the diachronic study of oral tradition as
'pseudohistory.' However, in a literate culture like Sri Lanka, with
historical records going back about two thousand years, it may be
possible to relate the tradition of myth and ritual to historical periods or
events and thereby to 'verify' hypotheses or propositions regarding the
former. It may therefore be possible to construct a "speculative history"
of a mythic tradition, grounded on historical data (page 283).

As a matter of fact, these historical reconstructions are not so

different from what professional historians are often engaged in
doing. What is interesting is what Radcliffe-Brown did not say:
namely, that insofar as anthropology deals with the study of
ritual, such historical reconstructions are "pseudo-anthro-
p o l o g y . " For although these reconstructions may be of interest
by themselves, they may not be relevant to an understanding of
the ritual per se—jüst as the etymology of a word may not be
relevant to its meaning or use. Obeyesekere is, of course,
conscious of this. His readers should be similarly aware that
these explanations do not explain the ritual system. They only
add to the impression that, in historical terms, it is a hodge-
podge—just like the systems Tambiah studied and like the Vedic
" G r e a t Tradition" itself (Staal 1963, 1972 2 )—a topic to which we
shall return in Chapter 28A. In the meantime—and before
 Anthropology Without Mantras 325

another Gombrich comes along—one should be grateful to the

author for his historical reconstructions which do not merely
connect Sri Lanka with South India, but point at more distant
origins for certain features (in particular, the ritual drama of the
resurrection of Pattini's consort) in Western Asia and in the
Mediterranean. Although such connexions remain unproven
they do not conflict with what is presently known about early
Indian history. Moreover, although Obeyesekere is not fright-
ened of speculation, he knows when to stop—unlike S. B.
Steever, the author of a perfunctory review in the Journal of
American Oriental Studies who wishes that Obeyesekere had
gone further and included theology and existentialist metaphys-
ics (Steever 1985:187).
There is perhaps a reason for Steever's unreasonable expec-
tations: for Obeyesekere uses similarly alien kinds of semantic
interpretation by adopting the Freudian psycho-analytical reduc-
tions that he had earlier used, for example, in Medusa's Hair
(1981). This is the second area of "semantic interpretation"
offered to elucidate the cult. I am not competent to judge these
psycho-analytical attempts at explanation, but it strikes me that
they may only be helpful when we are dealing with neurotic
individuals about whose persona! life and childhood a great deal
is known. In Medusa's Hair we were provided with seven case
studies of people who Obeyesekere had not actually put on the
couch, but with whom he had lengthy personal conversations. In
the Pattini book, however, we are dealing with cults that are
much less personal. Even though the author asserts in his
Preface that he "got to know intimately" the kapurälas from a
number of villages and towns, a psycho-analytical treatment is
accordingly less convincing. I shall not belabor this point but
give one illustration of a psycho-analytical interpretation that is
simply out-of-date.
Although Obeyesekere's book is primarily devoted to the
Pattini cult, many other closely or less closely related ritual
developments are also provided with a detailed description and
analysis. More than a hundred pages are devoted to the Anke I iy a
rituals that have nothing to do with ankles but a great deal with
326 Mantras

anal intercourse. This is regarded by the participants with both

fascination and abhorrence. According to Obeyesekere, how-
ever, the homosexual theme is not as important as the "key-
e m o t i o n s " of "castration and impotence anxieties" (page 487).
(It is only on these pages (486-487) that I found two Freudian
slips: " h e t e r o s e x u a l " for " h o m o s e x u a l , " and "superordinate
m a l e " for "subordinate male.") But when he tells us that an
important part of the ceremonies, in which only adult males
participate, consists in two teams pulling ropes that are attached
to two interlocked horns (either horns of the sambar deer, or
wooden hooks or tree roots representing such horns) until one of
them snaps, and that these horns stand for penises, while in a
closely related rite a male demon shaves the body hair of another
man, one wonders. One also wonders whether these reductions
of homosexual activities to castration fears, so outmoded and
yet so tediously familiar, could not be straightened out by some
measure of Gay Liberation. As it is, they are no more than stains
on a book that in every other respect is vibrant with new facts,
new ideas, and new insights.
The historical reconstructions add much to the value of this
book and the psycho-analytical suggestions, although of limited
relevance, do not interfere with the presentation of the materials
and hence do not detract from its value. I shall revert to the
former which exhibit outer chance (pages 344-346). However,
neither type of analysis throws light on the ritual systems as
such. Given the nature and abundance of the data, a syntactic
analysis should therefore be rewarding.

25C. THICK SEMANTICS: Clifford Geertz

Clifford Geertz' monograph Negara: The Theatre State in

Nineteenth-Century Bali (1980) takes us further away from India
but not from Hinduism and Buddhism. Geertz doesn't only write
about cults; he has become a cult figure himself. One thinks of
expressions such as " d e e p p l a y " (Geertz 1972) or "thick de-
scription" (Geertz 1973), that have not only become slogans
(from Sanskrit sloka, a verse in a popular meter) but that can
 Anthropology Without Mantras 327

hold their own even with mantras such as om in suggesting

profound and hidden significance.
Early in his career, Geertz put Dutch scholars of Indonesia to
shame by writing a brilliant book entitled The Religion of Java
(1960). Using purely anthropological and synchronie methods of
fieldwork, he showed that Javanese religion on the village or
"small t o w n " level consists of three strata, corresponding to the
three levels that historians had distinguished as having origi-
nated in Islam, (Buddhism-) Hinduism and what had been
vaguely referred to as "animistic beliefs." One of the results of
this work was that it became clear that traditional characteriza-
tions of Hinduism and Islam along the lines of the Christian
concept of religion, implying exclusiveness and incompatibility
with other religions, are inappropriate, at least in Java. Geertz is
aware of some of the conceptual problems inherent in the
Western concept of religion, but being interested less in what
" t h e religion of J a v a " is not than in what it is, he concentrates
on features of a different sort, for example (Geertz 1960:238):

The three major foci of prijaji "religious" life are etiquette, art, and
mystical practice. I admit to using "religion" in à somewhat broader
sense than may be typical, but there is nothing else to do when these
factors are so fused as to make their separate consideration nearly
meaningless.

This usage of the concept of religion is not only broader, but also
narrower than the typical sense; it excludes, for example,
considerations of doctrine or belief—a significant fact to which
we shall return.
If we compare Geertz' work with the monographs on Indone-
sian Islam by earlier Dutch authors (e.g., Drewes 1925, Kraemer
1921, van Nieuwenhuijze 1945, Rinkes 1909, Schrieke 1916—
none of them referred to by Geertz) we find, not surprisingly,
that the picture based upon contemporary anthropological ob-
servation shows a much greater degree of integration of histor-
ically distinct features than the picture based upon the philolog-
ical study of texts. Most of the books mentioned deal with Sufi
texts from Java and Sumatra that closely resemble the Islamic
328 Mantras

mysticism of Gujarat on the west coast of India from which they

derive. Islam took root in Gujarat soon after the Muslim
conquest of Sind, viz., within a century after M u h a m m a d ' s
death (the campaign started in year 92 of the Hijra, i.e., 711
C E . ) ; and it began also, almost immediately, to undergo India-
nization. Thus we find, for example, the followers of Shäh
Madär, a Sufi born in Aleppo in 1051 C . E . who spent much of his
life in India, smoke bhang with Indian ascetics (there are stories
in their writings about Muhammad using it in paradise: Husain
1929:29). Under such circumstances, it is not surprising that
Hindu and Muslim currents merged; according to Husain, the
Madäri cult was " t r è s répandu parmi les paysans sans distinc-
tion de religion." Combining philology, history, and anthropol-
ogy we thus find that the situation in Java as observed by Geertz,
in respect of emerging syncretism, may not have been so very
different from what it had been in India several centuries earlier.
After settling in Java, Indian Sufism, already a mixture, was
further enriched and modified by local elements, some of them
also of Indian provenance, others purely Indonesian.
Another of Geertz' books, Islam Observed (1968), provided
an account which implied that the types of Islam met with in
Indonesia and in Morocco, i.e., at the geographical extremities
of the Muslim world, have been so thoroughly moulded by local
culture and history, and have accordingly so little in common,
that the appropriateness of applying a single religious label such
as " I s l a m " again becomes questionable. This book also illus-
trates that certain features of civilization are stronger and more
permanent than religion. This was already evident from India, as
we have just seen; applies especially to caste (a feature that is
often preserved, for example, when an Indian converts to Islam
or Christianity); and was later confirmed with specific reference
to ritual ( " t h e invariance of ritual under religious transforma-
t i o n , " or " t h e invariance of ritual across religious b o u n d a r i e s : "
see pages 261, 390-394; Schipper and Staal 1988).
The Bali b o o k dates from 1980, and followed a series of more
specialized studies on Bali (1964, 1972, 1973, etc.). The term
negara, which in Bali and Indonesia generally means " s t a t e , " is
 Anthropology Without Mantras 329

none other than the Sanskrit nagara, or " t o w n , " which Bud-
dhist monks added to " v i l l a g e " so as to form a pair in opposition
to the tradition of " f o r e s t . " In Bali (as in India), negara
contrasts with desa as " s t a t e " or " t o w n " does with " c o u n -
tryside."
G e e r t z ' book deals primarily with politics, but since in Bali
political theory is hard to come by, he has sought it in the ritual
performances that are the subject of his final chapter, entitled:
"Political Statement: Spectacle and C e r e m o n y . " In this chap-
ter, ritual ceremonies are interpreted as symbolic expressions of
power. Geertz concentrates on the more spectacular and popu-
lar "outstanding f e a t u r e s " of these ceremonies—e.g., in the
case of cremation, " a social one, the procession; an aesthetic
one, the tower; and a natural one, the fire" (page 118). A
categorization in terms of " s o c i a l , " " a e s t h e t i c , " and " n a t u r a l "
is Western, which does not mean that it must be inapplicable—
unless, of course, it is. The available evidence suggests precisely
that. For much is indeed known about the ritual structure of
these ceremonies from other sources, some of which are quoted
by Geertz himself, especially a series of monographs by C.
Hooykaas on Balinese Hindu and Buddhist rituals which are
almost as detailed and precise as our monographs on Vedic ritual
because they are based not merely upon H o o y k a a s ' observation
of performance but also upon competence, that is, the knowl-
edge and traditions of the officiating priests. (Hooykaas was not
familiar with Vedic ritual until a few years before his death in
1979; he was much impressed by Diehl's 1956 book on South
Indian agamic rituals: see Hooykaas 1964:17).
H o o y k a a s ' oeuvre (which is a culmination of earlier investi-
gations by Dutch scholars such as Goris, Grader, Swellengrebel
and others) has established that Balinese culture is in one respect
very similar to Indian, and especially Vedic, culture: what we
tend to label as "religion" consists there almost entirely of
complexes of mantras and rites. While processions, towers and
fires are among their more spectacular manifestations, these
complexes themselves are elaborate constructions that are in
many specific respects similar to the Vedic ceremonies. They
330 Mantras

may also be of equally ample proportions. For example, an

extended ceremony called Ekädasarudra ("The Eleven
R u d r a s " ) , reputedly celebrated only once in a century, was
performed in April 1963 in Besakih on the slope of Gunung
Agung (cf. Hooykaas 1973:167-249). Allegedly Buddhist, it
refers to the same Rudra who is addressed on the eighth day of
the Agnicayana by the Adhvaryu when the Udgätä sings his
Flow of Milk (above, page 303). (Such cultural diffusion or
Indianization is familiar to Indologists. Another ceremony from
the Agnicayana, " F l o w of W e a l t h " [vasor dhärä: above, page
75], turns up in Nepal, centuries later, as a popular Buddhist
goddess.) These ritual complexes are celebrated on all kinds of
occasions and may be used to express all kinds of things: myths,
history, ideologies, power relationships, status, political and
personal ambitions, etc.; but in no case is there an intrinsic and
necessary relation between the rites or mantras and what they
are called upon to express. That Geertz had an inkling of this
may be inferred from two paragraphs of the Bali monograph that
are of particular methodological interest and deserve to be
quoted in full (Geertz 1980:103).

The Balinese, not only in court rituals but generally, cast their most
comprehensive ideas of the way things ultimately are, and the way that
men therefore should act, into immediately apprehended sensuous
symbols—into a lexicon of carvings, flowers, dances, melodies, ges-
tures, chants, ornaments, temples, postures, and masks—rather than
into a discursively apprehended, ordered set of explicit 'beliefs.' This
means of expression makes any attempt to summarize those ideas a
dubious business. As with poetry, which in the broad, poiesis ('making')
sense is what is involved, the message here is so deeply sunk in the
medium that to transform it into a network of propositions is to risk at
once both of the characteristic crimes of exegesis: seeing more in things
than is really there, and reducing a richness of particular meaning to a
drab parade of generalities.
But whatever the difficulties and dangers, the exegetical task must be
undertaken if one wants to be left with more than the mere fascinated
wonderment—like a cow looking at a gamelan orchestra, as the Balinese,
put it—that Helms, for all his responsiveness and powers of description,
displays. Balinese ritual, and most especially Balinese state ritual, does
embody doctrine in the literal sense of "teachings," however concretely
 Anthropology Without Mantras 331

they are symbolized, however unreflectively they are apprehended.

Digging them out for presentation in explicit form is not a task for which
the Balinese, aside from a few modernists nowadays, have ever had any
interest. Nor would they feel, any more than a translated poet ever feels,
that any such presentation really gets to the heart of the matter, gets it
really right. Glosses on experience, and most especially on other
people's experience, are not replacements for it. At the very best they
are paths, twisted enough, toward understanding it.

But has Geertz proved his point? Are there teachings, is there a
metaphysics ("most comprehensive ideas of the way things
ultimately a r e " ) or a system of ethics ( " t h e way that men should
therefore a c t " ) underlying this " l e x i c o n " of carvings, flowers,
dances, and so on? Do the ritual ceremonies to which Geertz
refers (and rather synoptically characterizes) express teachings
about power? And is to deny this facile assumption equivalent to
being a cow, or being like that nineteenth century British
globe-trotter and sightseer L. V. Helms who authored a travel-
ogue entitled Pioneering in the Far East and Journeys to
California in 1849 and to the White Sea in 18481 One rather
doubts it.
Let us take a closer look at that proverbial cow. What do the
Balinese mean when they refer to a cow looking at a gamelan
orchestra? That she misses its metaphysics, ethics, teachings or
meaning? Well, so do most people, because a gamelan orchestra
possesses none of these. What the cow misses and what the
connoisseurs appreciate is the structure of the music (its
" s y n t a x " ) and the tonal qualities of its melodies (its " p h o -
n e t i c s " ) . That music expresses meaning like language is a
nineteenth century Western belief, inspired by program music,
that certainly does not apply to gamelan—a belief about which
the Harvard Dictionary of Music (1970:696a, quoted above page
176) stated: " A b o u t 1900, many persons, particularly writers on
music, believed that in order to be understandable music must
'express something' or 'tell a story' . . . Today such views are
things of the past . . . " Why, then, should we return to such
superstitions when interpreting gamelan, or ritual, for that
matter?
332 Mantras

I do not deny that what Geertz says about power and politics
may be relevant to Balinese culture; but that it constitutes the
implicit teaching or the implicit meaning of the ceremonies he
refers to is at least far-fetched. In any case, it would need to be
demonstrated. The Balinese are intelligent enough to express
their political notions explicitly and in direct terms. They are not
cows; they have language and can express their ideas without
having to sing or dance them. True, there are all kinds of
connexions and relationships, but these hold between all the
phenomena that Geertz has enumerated—the rites, the flowers,
the chants, the ornaments, the temples, and the political ideas as
well. The relationships between all these entities constitute a
rich and intricate structure that will undoubtedly be difficult to
unravel. That would be a task worthy of an anthropologist, and
Geertz himself hinted at the desirability of such an investigative
analysis when he claimed that the Balinese cockfight is a
1
'cultural d o c u m e n t , " " a t least as important a revelation of what
being a Balinese l is really like' " as Balinese art and social
organization (I quote after Jonathan Lieberson's 1984 review of
Geertz' Local Knowledge to which I shall return). However, in
the Negara book, Geertz does the opposite: he isolates and
concentrates upon one term as the key that must unlock all the
others. Such a move can serve no purpose other than to reveal
o n e ' s own particular interests and prejudices. Any author is free,
of course, to write about whatever he likes. Geertz' book is
about politics; but that does not imply that Balinese ritual is
about politics. Moreover, to collect ideas and meaning on one
side and relegate all concrete entities to the other, and then to
postulate one-to-one-correspondences between these two col-
lections, smacks of the kind of naive dualism that Western
philosophers are trying so hard to overcome. (Note that this is
different from Tambiah, who started with concrete entities
[amulets] and postulated an interpretation; Geertz starts with
[religious or political] meanings and offers them as interpreta-
tions for a variety of concrete entities.)
After having first (in Cockfights) claimed that everything
 Anthropology Without Mantras 333

hangs together, without demonstrating it, Geertz now (in Ne-

gara) claims the opposite, viz., that one thing holds the key to
•everything, or at least much else, again without demonstrating it.
The question is not only whether this theory is true; the question
is, does it even make sense? H o w could people be so obsessed
by politics and yet so collectively inarticulate as to express it
entirely through ceremonies without being able to refer to it in
words? Where is the " h e r m e n e u t i c c i r c l e ' ' ( r e f e r r e d to on the
same page, 103)? Why can we not simply recognize that these
ceremonies are rule-governed activities, which we may begin to
understand only when we know the rules? Geertz himself
suggests this implicity (on the next page) when he compares
them, rather more aptly, to a baseball game. Must a baseball
game necessarily also teach a lesson, have a meaning, a meta-
physics, and an ethic? We all know better: if we wish to know
and understand baseball, we must know what the rules are.
Subsequently w e can discuss its relationships to art, literature,
religion and p o l i t i c s / A n d of course, we may learn about its
psychology or anthropology by including spectators' responses
and much else. Alan Dundes has done it brilliantly for American
football (Dundes 1978).
Long before Negara, Geertz had a much better idea about
" t h e religion of B a l i . " In a 1964 article, reprinted in 1973, he
wrote:

In Java, where the pressure of external influences has been relentless,

and where traditional social structure has lost much of its resilience, not
just one but several relatively well-rationalized systems of belief and
worship have developed, giving a conscious sense of religious diversity,
conflict, and perplexity still quite foreign to Bali. Thus, if one comes, as
I did, to Bali after having worked in Java, it is the near total absence of
either doubt or dogmatism, the metaphysical nonchalance, that almost
immediately strikes one. That, and the astounding proliferation of
ceremonial activity. The Balinese, perpetually weaving intricate palm-
offerings, preparing elaborate ritual meals, decorating all sorts of tem-
ples, marching in massive processions, and falling into sudden trances,
seem much too busy practising their religion to think (or worry) very
much about it (Geertz 1973:175-176).
334 Mantras

Geertz then remarks that " t o say that Balinese religion is not
methodically ordered is not to say that it is not ordered at a l l , "
and goes on briefly to describe three "relatively well-defined
ritual complexes which exhibit, in turn, a definite approach to
properly religious issues no less respectable for being implicit"
(my italics). It is a pity that Geertz has not pursued the study of
these ritual complexes on their own terms, and instead has
constructed from his own imagination (that is, religious back-
ground) the allegedly implicit but "properly religious" issues.
But now we see something surprising happen. In the Negara
book, these issues are no longer religious but turn out to be
political. It is difficult to reconstruct the meanderings of other
minds, but can it be that we witness the result of a merger
between two erroneous assumptions, viz., (1) ritual must ex-
press something and (2) ritual must express religion, in the
manner, say, of a "hermeneutic c i r c l e ? " That would explain the
"hopping back and forth between the whole conceived through
the parts that actualize it and the parts conceived through the
whole that motivates t h e m " (Lieberson 1984:40, apparently
after Geertz). It might also explain the contradiction between
ritual expressing political teachings—the main thesis of the
book—and such expressions as " a royal cremation was not an
echo of a politics" (page 120) or " t o construe the expressions of
the theatre state, to apprehend them as t h e o r y " is " p r e j u d i c e "
(page 136).
So let us return to things "properly religious." Any reader
would expect such things to possess at least a proper religious
affiliation. But in Bali, as in Sri Lanka, Burma, Thailand, the
other countries of Southeast Asia, and indeed in Nepal and on
the Indian subcontinent itself, it is often impossible to make a
sharp distinction between Hindu, Buddhist, and " a n i m i s t . "
Geertz had noted this with respect to Javanese Islam, which is
permeated by Hindu and animist notions, and even though he
sometimes refers to " H i n d u B a l i , " he is careful when mention-
ing Hinduism ("is that what the Balinese great tradition should
be c a l l e d ? " page 46). What we actually find in Bali is a ritual in
which features that can be differently labelled when one looks at
 Anthropology Without Mantras 335

their historical origins have merged—-just as in Sri L a n k a (see

Obeyesekere, passim), and just as in Thailand where we meet
with an amalgam of Buddhism and "-animism" as Tambiah
pointed out in his refutation of Spiro's attempt to keep the two
distinct.
One reason that labels such as Hinduism and Buddhism are
not to be taken too seriously is that doctrines and beliefs are not
as important as in Western religion; what counts instead is ritual
affiliation, lineages, initiation, cults and practice (see A G N I II:
xiv-xv). This is precisely what was missing in Geertz' observa-
tion quoted above page 327, and explains why things that appear
incompatible from a doctrinal point of view are easily combined:
they are not doctrinal, but ritual. If Geertz had pursued his study
of Balinese ritual complexes and not persisted in his search for
religious meaning, he would not have missed a discovery that
would have been a fitting climax to his two earlier findings, viz.,
that the "religion" of Java is not like a religion in the Western
monotheistic sense, and the " r e l i g i o n s " of Indonesia and Mo-
rocco, though both labelled " I s l a m , " are not a single religion in
that sense either. For in Bali, Geertz was on the verge of making
an even more momentous discovery, viz., that there, as in many
of the traditional societies of South, Southeast and East Asia,
rituals are not necessarily associated with doctrinal superstruc-
tures or belief systems; that is, they are rituals without religion.
The use of such words as " p r o p e r " and " p r o p e r l y " is very
revealing about the prejudices of an author. Part of what Geertz
apparently had in mind before he ever set foot on Balinese soil,
although he need at no time to have been fully aware of it, is that
the Balinese must have a proper religion. It is ironic that
H o o y k a a s made a similarly revealing slip in his own domain; and
Geertz did not fail to notice it. Amin Sweeney drew my attention
to the memorable exchange (memorable, that is, if one tends to
remember abusive oratory) that took place between Geertz and
H o o y k a a s in the French journal Archipel (1976) on the occasion
of H o o y k a a s ' review of the Geertzes* book Kinship in Bali
(1975). H o o y k a a s demonstrated in subtle, not so subtle, and
downright awkward ways that the Geertzes had many facts
336 Mantras

wrong and were not very familiar with the scientific literature on
Bali. Anyone who checks the list of alleged mistakes will
concede that Hooykaas was right in almost every case; but
Geertz, 4 declaiming eloquently on anthropology and philology in
his felicitous style, got the better of his adversary and the
uncritical r e a d e r ' s sympathy will rest with him. In two cases
(pages 25 and 134), H o o y k a a s ' criticism is simply lost on Geertz
because when the latter wrote his rebuttal he still did not know
he did not know the sources that H o o y k a a s criticized him for not
knowing.
In the course of his review, Hooykaas quoted the third
sentence of the Geertzes' book in the following context:

Several years ago a team of TV reporters interviewed the Paris public,

trying to find out what it still knew about Stalingrad after a quarter of a
century. The results were devastating. I have only to remind the author
of the third clause of his own book: Two fully cooperative and intelligent
Balinese from the same village may give completely variant accounts on
matters that the ethnologist believes to be crucial for his formulations/
Quite so, but then why not consult proper books? (page 241).

Geertz replied that his sentence " s t a t e s the conditions of an

ethnographer's work, the very shape of the 'fact' he is investi-
gating, and the challenge and possibility his subject offers him,
not something to be got round by consulting 'proper b o o k s ' "
(page 225).
That H o o y k a a s ' bias is " b o o k s " is not surprising; but that
Geertz' bias is "religion" is. For, as an anthropologist, Geertz
should be especially cautious when it comes to projecting
concepts from his own culture. We have seen that Geertz was
cautious in his Java book, despite its easily misleading title; in
the Bali book he is less prudent although the Western concept of
religion is even less applicable there.
At a closer look, the exchange between Geertz and H o o y k a a s
demonstrates that the two fighting cocks agree at least to some
extent on the importance of written sources; and this agreement
reveals a shared bias which might not mar our understanding of,
say, Islam or China, but which is another obstacle to the
 Anthropology Without Mantras 337

understanding of Balinese—and indeed, most South and South-

east Asian—culture. The issue is an interesting one. Hooykaas
quotes Geertz at some length, and I shall do the same especially
because the quote comes from an earlier (1964) paper that in the
meantime has become well-known and has been reprinted in
Geertz 1973 (where the quote now occurs on page 185). Hooy-
kaas writes (page 237) with reference to this 1964 paper:
Here Geertz tells us about post-war Bali, to be exact the year 1957, that
new era (after centuries of manuscript copying) of mimeographed and
sometimes even printed pamphlets. Some of them are done in Balinese
and others in Roman script, some of them in Balinese language and
others in Indonesian. Nearly weekly they were produced by publishers,
in the island-capital Denpasar mainly, but also in the provincial capitals
Gianyar, Klungkung and Tabanan (names not to be forgotten). 'When I
bought some books of this sort,' Geertz writes on p. 297, 'and left them
around our house in the village, our front porch became a literary center
where groups of villagers would come and sit for hours on end and read
them to one another, commenting now and then on their meaning, and
almost invariably remarking that it was only since the Revolution that
they had been permitted to see such writings, that in the colonial period
the upper castes prevented their dissemination altogether. This whole
process represents, thus, a spreading of religious literacy beyond the
traditional priestly castes—for whom the writings were in any case more
magical esoterica than canonical scriptures—to the masses, a vulgariza-
tion, in the root sense, of religious knowledge and theory. For the first
time, at least a few ordinary Balinese are coming to feel that they can get
some understanding of what their religion is all about; and more
important, that they have a need for and a right to such understanding.'

If I had to review this passage I would probably comment on

those righteous, anti-colonial and democratic Americans with
their open houses and canonical scriptures. Perhaps I would add
that it is at least conceivable that the visiting anthropologist's
house was more comfortable and spacious than many others in
the village; or remark on the, at least apparent, inconsistency
inherent in the notion of a priestly class that does not understand
the meaning of the very writings they themselves composed. But
H o o y k a a s , who was more knowledgeable about Bali than I shall
ever be, and also more kindly disposed even under duress,
reacted as follows:
338 Mantras

The painful reality is that the author here straight-away believed a few
young village lads, probably not yet born when in 1928 the Kirtya
Foundation for palmleaf manuscripts was created in the then capital
Singaraja. Soon it had its own modest building: a spacious room for
clerks and European books of reference on Bali and Lombok, a very
sufficient store for the well-preserved manuscripts, a well-visited room
for anybody interested, opened by the then acting Governor-General.

H o o y k a a s continues in this vein, informing us how that Kirtya

library, which now lodges some 3700 manuscripts and which
during the colonial period wrote and financed a dozen publica-
tions, was a standard reference library on Bali and L o m b o k .
About the frequent visits he made there in the late thirties and
early forties, Hooykaas says, " I cannot remember that the
readers' room was ever unoccupied."
Geertz replied that the " y o u n g village l a d s " who spent time
on his porch were in fact " m e n of all ages, including ones born
not only before 1928 but even before H o o y k a a s " . . . and so on
it goes. One hopes that, should his essay be reprinted again,
Geertz will not withhold from his readers this information on the
Kirtya library (about which he wrote in his reply that it " h a s
done invaluable w o r k " ) . It is tempting to say more about this
exchange, especially about colonialism—for example, that the
Dutch, British and French, whatever their atrocities, left a
splendid scholarly record of the cultures, languages, and histo-
ries of the countries where they committed them, whereas the
Americans, who were almost successful in destroying the coun-
try, never even left room in their system of higher education for
studying Vietnam (see Staal 1970b).
In the present context, the important issue is the following,
Geertz' assumption that the Balinese could never get any
understanding of " w h a t their religion is all a b o u t " until they
could read their canonical scriptures clearly shows that he does
not understand at all what Balinese "religion" is. H o o y k a a s
does not commit such a basic mistake, although both he, the
philologist, and Geertz, the anthropologist, share the same
Western prejudice in favor of written books. But H o o y k a a s
objects at least to the expression "holy writings" (page 239)
 Anthropology Without Mantras 339

which Geertz uses freely. In fact, Hooykaas objects on several

counts to Geertz' uses of " h o l y " and " s a c r e d " — e x p r e s s i o n s
that H o o y k a a s quite rightly says are "highly laden technical
terms from the field of divinity/theology and should not be used
easily." All these errors are amusing and entertaining, but
Geertz' error has in addition far-reaching implications. F o r if a
people can get no understanding of what their religion is all about
without reading their canonical scriptures, the Indians must
have been ignorant about their religion for the better part of
2,500 y e a r s *
We have seen that Tambiah was not averse to introducing a bit
of technical jargon even if it is not always clear what good it
does. Geertz goes much further and likes to flirt with contem-
porary ideas whatever their nature or provenance. The " h e r m e -
neutic c i r c l e " is as dear to him as " d e e p s t r u c t u r e , " though
" d e e p p l a y " has nothing to do with the latter; for " d e e p
s t r u c t u r e " is a technical term with a very precise meaning
whereas " d e e p p l a y " is a nice way of expressing the claim
Geertz first made and then abandoned, as we have just seen. In
linguistics, a " d e e p s t r u c t u r e " is an abstract underlying syntac-
tic structure, which has to meet certain conditions of well-
formedness, and which is postulated in order to provide a
starting point for derivations of "surface s t r u c t u r e s " with the
help of rules that also meet specific conditions of well-formed-
ness. In Geertz, " d e e p p l a y " is an activity that seems innuoc-
uous and irrelevant, but that in fact expresses important issues
(such as, in the case of the Balinese cockfight,, honor and status).
H o w e v e r , this terminology has no interesting implications as in
the case of linguistics: for while Chomsky and his followers
studied deep structures in order to find universal structures,
which underlie all human languages, on G e e r t z ' view " t h e r e may
be no interesting general laws of culture, or of anthropology, to
d i s c o v e r " (Lieberson 1984:45; cf. Geertz 1985). " T h i c k descrip-

* Even in Islam, "reading canonical scriptures" is not a prerequisite for

being a good Muslim: Quran means "recitation"and "the main stress is always
on recitation and on the rote learning necessary to it" (observed by Geertz
himself in Morocco: 1983:110-112).
340 Mantras

t i o n , " similarly, is a spin-off from Gilbert Ryle, as Geertz

acknowledges, but without the subtleties that Ryle attached to
what is after all a piece of semantic observation that stands in
need of a theoretical analysis—an analysis that neither Ryle nor
Geertz made any attempt at providing.
The emphasis on religion and accordingly on meaning that we
meet with in Geertz' later work raises a host of philosophical
problems. We have seen in Chapter 3 how logicians and linguists
have tried to come to terms with problems of meaning. This can
be done provided we are working within a theoretical framework
that is well developed—as are logical syntax and semantics. But
ideas or phrases taken out of their (theoretical) context will
never be helpful, as even a minimum of analysis can demon-
strate.
Let us consider the idea from the confines or suburbs of
communication theory that has especially appealed to Geertz, as
we have already seen, of " a message sunk in the m e d i u m . "
N o w , a message is always sent in order to convey meaning; a
meaningless message is no message. A message also needs a
medium, by means of which it is conveyed—e.g., a telephone.
Let us take the simple example of a message left on my
telephone answering machine, saying: " M r . W. will not come
this afternoon." What would it mean, in such a context, to refer
to that message as " s u n k in the m e d i u m ? " It seems to convey
something like: there is no message apart from and beyond the
medium. In other words, the message that Mr. W. will not come
Has disappeared in the medium, viz., the telephone. Ordinarily,
a telephone message is a message sent through a telephone. The
expression " s u n k in the m e d i u m " suggests that the telephone
message is not independent of the telephone; in fact, in the
limiting case, when the message is completely " s u n k , " the
telephone is all of the m e s s a g e / T h e r e are not two entities,
message and medium; there is only one left. But this does not
make any sense, unless it merely means that there is no message,
but that we have a substance that is interpreted as a medium but
that is not a medium through which anything passes: the
telephone interpreted as a message by itself. Like someone
 Anthropology Without Mantras 341

arriving h o m e , seeing his telephone and exclaiming: it must be

that Mr. W. has telephoned and left the message that he will not
come this afternoon! But since there is no message, this excla-
mation is mere fantasy—an expression of disappointment, per-
haps, or a wish—that has nothing to do with either medium or
message.
It is likely that Geertz did not have telephones in mind when
he invoked the idiom; but the communication engineers who
introduced the terminology were very much concerned with
such media. It is accordingly unclear to what extent any part of
their theory applies to anthropology. Why, then, should Geertz
wish to make use of such unhelpful constructions,'that become
obscure as soon as they are detached from their original context?
There are several possible answers, but the best one can do is
study illustrations he provides himself. Geertz refers to poiesis,
" m a k i n g " or " d o i n g , " as such an activity where the message is
" s u n k in the m e d i u m . " What he means, apparently, is that it is
impossible to do justice to a poem by formulating its meaning as
something separate from the particular form in which it has been
expressed in the poem. But why emphasize meaning at all? Why
not accept what poets say themselves, namely, that poems are
made not with ideas (meanings) but with words? If one starts
with the primacy of meaning, and then picks on a particular way
of expressing it, one is finally forced to insist that, in some cases,
the meaning is not to be separated from its particular expression.
It is obviously preferable from a methodological point of view to
proceed differently, and to accept a poem as a datum that may be
interpreted on one or more levels, as the need arises, and
without ignoring the concrete formal structure which is in any
case part of it. That method has been put to good use by
linguistic interpreters of poetry, e.g., Roman Jakobson. Such
interpretations show that the formal structure of a poem makes
a specific contribution to the explanation of its expressive
power. Each piece of analysis that Jakobson has given contains
such a specific discovery.
Geertz believes that art cannot be understood by a formal
structural analysis of " s o u n d s , images, volumes, themes, or
342 Mantras

g e s t u r e s " (1983:96). Such statements, however, are uninterest-

ing as well as unconvincing. They are uninteresting because they
can never be demonstrated, and unconvincing because in spe-
cific cases they have already been refuted. Moreover, such
positive cases of analysis as were given, for example, by
Jakobson, may lead to a general theory, unlike any belief that " i t
cannot be d o n e . "
I have spent some time on this discussion about poetry
because it throws light on the problems in anthropology we have
been discussing. The study of ritual deals with activities, and
poiesis is the general word in Greek for *'activity." Poiesis
corresponds to Sanskrit karman, from the root kr-, " d o , "
" m a k e , " which signifies " a c t i v i t y " and in particular "ritual
activity." The corresponding Latin is actum, " d o n e , " the past
passive participle of agere, " t o d o . " The plural acta, "(things)
d o n e , " corresponds to Greek dramena, " a c t i v i t i e s , " which also
refers to ritual activities.
In ritual we are primarily dealing with sounds and acts, and
these correspond to each other. Our main confusions in this area
are due to the facts that " s o u n d " reminds us of language, and
that " a c t " seems to be something simple. About sound and
language I have said enough, but with regard to acts it should be
emphasized that they are not simple. I can illustrate this no
better than by quoting Austin (1961:126-127): " T h e beginning of
sense, not to say wisdom, is to realize that 'doing an action,' as
used in philosophy, is a highly abstract expression—it is a
stand-in used in the place of any (or almost any?) verb with a
personal subject, in the same sort of way that 'thing' is a stand-in
for any (or when we remember, almost any) noun substantive,
and 'quality' a stand-in for the adjective . . . So we come easily
to think of our behaviour over any time, and of a life as a whole,
as consisting of doing now action A, next action B , then action
C, and so on, just as elsewhere we come to think of the world as
consisting of this, that and the other substance or material thing,
each with its properties. . . . If we are to continue to use this
expression in sober philosophy, we need to ask such questions
as: Is to sneeze to do an action? Or is to breathe, or to see, or to
 Anthropology Without Mantras 343

checkmate, or each one of countless others? . . . Further we

need to realize that even the 'simplest' named actions are not so
simple—certainly are not the mere making of physical move-
ments, and to ask what more then, comes in (intentions?
conventions?) and what does not (motives?), and what is the
detail of the complicated internal machinery we use in
'acting"-—the receipt of intelligence, the appreciation of the
situation, the invocation of principles, the planning, the control
of execution and the r e s t . " And in another essay, published in
the same volume (page 224), Austin writes: "Philosophers at
least are too apt to assume that an action is always in the last
resort the making of a physical movement, whereas it's usually,
at least in part, a matter of convention."
It might seem at first sight that anthropologists would be the
last to make the latter mistake; after all, they are not (or rarely)
inspired by the physical sciences and are professionally inter-
ested in " c o n v e n t i o n s . " However, they do tend to assume,
mistakenly, like many other practitioners of the humanities and
the social sciences, that actions must be done intentionally,
purposefully, and meaningfully, and must therefore possess a
meaning or even incorporate a message that someone wishes to
send us. This is precisely what contradicts the chance-like
association between mantras and acts discussed in Chapter 24,
to which I referred as "inner chance d e v e l o p m e n t . "
The anthropological studies discussed in the present chapter
exhibit another chance-like development which supplements the
former and to which I shall refer as " o u t e r chance develop-
m e n t . " It corresponds to the accretion referred to by Schipper
(see Schipper and Staal, forthcoming), and is found in each of
the three anthropological studies as when Obeyesekere writes
with respect to mythology (above page 324): " A n y mythological
tradition comes from historically diverse s o u r c e s . " As we noted,
this is characteristic of ritual traditions as well. It is obvious that
the result of chance encounters cannot be meaningful in the
manner ritual is generally described and always conceived. To
assume that we are dealing with a meaningful unit would be just
as unreasonable as to suppose that the Christmas tree refers to
344 Mantras

or is the same as the tree of knowledge of good and evil

mentioned in the paradise story of Genesis.
Outer chance developments arise when different cultures or
features of different cultures come together by chance and
combine into new structures. We have seen such elements—
" H i n d u , " " B u d d h i s t , " " M u s l i m , " " a n i m i s t " — c o m e together
and form " m i x t u r e s , " " a m a l g a m s , " " t o t a l i t i e s , " " s y n c r e -
t i s m s " — w h a t e v e r scholars have called them—in Thailand, Sri
Lanka, Java and Bali, and we could adduce similar instances
from other areas of South, Southeast and East Asia. Although
these combinations can sometimes be explained from or at least
against a historical background, they are obviously not develop-
ments that may be described as intentional or meaningful
activities. They are chance encounters. That they can be com-
bined at all is due to the fact that the resulting structures are
ritual. There are hardly any restrictions as to the combination of
ritual acts with each other. We can combine the making of an
oblation with the singing of a song, the killing of an animal, the
lighting of a candle or the crossing of a bridge. Doctrines or
beliefs, on the other hand, do not equally easily combine
because they have to be consistent with each other in order to do
so. We shall study this in greater detail in Chapter 28A, but we
can already conclude that the development of ritual structures
from external elements is often due to chance.
Inner chance developments arise within a ritual tradition as in
the cases discussed in Chapter 24 where the verse of Rigveda
9.1.1 with its various ritual applications yielded different exam-
ples of chance events, which cannot be explained in terms of
intentions, purpose or meaning. In each case the verse was
associated with a rite for no apparent reason—like the wind-
blown seed that settles on a particular spot. One could not go so
far as to claim that, on each of these three occasions, any other
verse would d o ; but there are certainly many verses that could
be used. We conclude that the development of ritual structures
from both external and internal elements is often due to chance.
Discussions on culture and nature are more often inspired by
metaphysical issues than by empirical data. In some cases,
 Anthropology Without Mantras 345

however, definite conclusions have been established, and they

do not always point in the same direction. The recent debates on
animal language have clearly led to the conclusion that animals,
although certainly capable of communication with each other
and thus of conveying messages and information, do not possess
language in the human sense. The human animal, thus charac-
terized by language, is often actually obsessed by it; but that
does not imply that every activity he engages in is language-like,
and accordingly meaningful. Unless we can show that in all
cases they are not, we must therefore allow the possibility that
some human activities are meaningless and governed by chance.
To construct a world of meaning where there is none is mythol-
ogy and not a substitute for finding the truth. Ritual ceremonies
provide the unbiased observer with so many unexplained mys-
teries, that is it likely that we should conclude that some are
simply due to chance. We have found two kinds of evidence that
clearly supports such a conclusion.
Our evidence has been drawn from Asia, but if relevant, it
must also occur elsewhere. Geertz' Bali book provides an
example, together with a final illustration of his own bias. F o r
Geertz does not like chance, and when the evidence clearly
points at it, he looks the other way. In a note in the Bali book
(page 216), he quotes the following passage from Giesey (1960),
a student of royal funerals in fifteenth- and sixteenth-century
France: " T i m e and time again . . . I have emerged with the
conviction that some crucial, innovation in the ceremonial first
occurred quite haphazardly . . . and later generations when
re-enacting it embellished it with clear-cut symbolism. This is to
say, on the level of events themselves, chance frequently
reigned; but symbolic forms affected the thought. . . . " Geertz
agrees with this "picture of irregular evolution . . . regularized
by post hoc interpretation and adjustment," but he a d d s , signif-
icantly: " ' c h a n c e , ' I think, is not quite the right w o r d . "
What is remarkable about Giesey's formulations (which in-
clude " h a p h a z a r d l y ' . ' as well as " c h a n c e " ) is not that they
simply picture "irregular evolution," but that they closely
resemble biological explanations—presumably without any
346 Mantras

knowledge of such types of analysis on the part of Giesey (since

otherwise he would doubtless have mentioned it). For "crucial
innovations in the ceremonial first occurring quite haphazardly
. . . later embellished with clear-cut symbolism" is strikingly
similar to "variations arising by chance . . . later selected in
accordance with the demands of the e n v i r o n m e n t . " T h e latter
expression has been used to characterize those fundamental
tenets of Darwinism that have not been modified or abandoned
in the expanded versions of Darwinism that took shape in the
1930's and 1940's (Stebbins and Ayala 1985:72). When a.biolog-
ical explanation of a ritual fact is so plausible in terms of
available methodologies, it makes good scientific sense to pur-
sue it and not to take refuge in semantic theories that are
unsubstantiated.
Our conclusion that chance plays an important role in ritual is
now well established but it is not altogether new. Jonathan Smith
(1982:53) quoted two stories from Kafka and Plutarch, respec-
tively, each illustrating the same:

Leopards break into the temple and drink the sacrificial chalices dry; this
occurs repeatedly, again and again; finally it can be reckoned on
beforehand and becomes a part of the ceremony.

At Athens, Lysimache, the priestess of Athene Polias, when asked for a

drink by the mule drivers who had transported the sacred vessels,
replied, "No, for 1 fear it will get into the ritual."
 Part IV

The H u m a n Sciences
 26

T h e Science of Ritual

T o most Western scholars and Westerners generally, science

is a characteristically Western achievement. Whether science is
considered a panacea e r a curse, it is the West that receives the
praise or blame. Confirming the myth of " l e miracle g r e c " , the
origins of science have been sought and found in Greece. It has
long been customary to claim that Greek science, especially
geometry, in its abstract and pure forms, presents a basic
contrast to the practical concerns of the ancient Egyptians, who
had to partition lands along the Nile, but never obtained more
general insights or searched knowledge for its own sake. True,
both Egyptians and Greeks knew how to construct a right angle
with the help of a triangle with sides of 3, 4, and 5 units' length;
but while the Greeks derived this as a theorem, to the Egyptians
it allegedly remained a mere carpenter's rule. In this miserable
role, the Egyptians have been appointed as representatives of all
non-Greek civilizations of this world.
Modern research on the history of science has shown that
such a picture is neither adequate nor justified. Babylonian
mathematics and astronomy, for example, have been proved to
be subjects of considerable interest. It is true that m o d e m
science, as it developed in the West since the sixteenth century,
far surpasses almost anything that came before. H o w e v e r , it
neither follows from this that science originated only in the
West, nor that its later achievements took place only there. That
the early developments of science are not confined to Europe
has been established beyond any doubt in numerous publica-
tions. Though N e u g e b a u e r ' s work on the Ancient Near East
may be pronounced a close second, this is nowhere put forward
more impressively than in Joseph N e e d h a m ' s volumes on 'Sei-
350 The Human Sciences

ence and Civilisation of China', a series that is still in progress.

Many volumes of this monumental work deal with the physical
or 'positive' sciences, though a fair amount of space is taken up
by what we would regard as technology. For India, we do not
possess any such standard work. There are numerous mono-
graphs in the field, from early attempts such as Brajendranath
Seal's 'The Positive Science of the Ancient Hindus' of 1915, to
contemporary publications of the Department of Mathematics
and Astronomy of the University of L u c k n o w , comprising a
series of texts edited by Ram Ballabh, Kripa Shankar Shukla and
others, also in progress; an ongoing series of publications by the
National Institute of Science at N e w Delhi, and another ongoing
project of manuscript inventories entitled Census of the Exact
Sciences in Sanskrit by David Pingree. For comparisons be-
tween Greek, Indian and Babylonian mathematics, see Seiden-
berg 1962, 1972, 1978 and 1983.
Most of these publications deal with the so-called " p o s i t i v e "
sciences, but we have seen in Chapter 5 that the ancient Indians
developed also a science of a different kind: the science of
linguistics. The present section deals with another Indian sci-
ence: the science of ritual. I could have called it " T h e Indian
Science of R i t u a l " were it not for the fact that no other culture
or civilization has developed any such science, as far as I know:
the only science of ritual mankind has evolved is the Indian
science of ritual. Both sciences—of language and of ritual—are,
at least apparently, " h u m a n s c i e n c e s . " If we assume that there
is a hierarchy among the human sciences, we should expect that
the science of language is situated at the base of this hierarchy,
for it studies the most basic feature in which humans differ from
the other animals. At the top of the hierarchy one would hope for
sciences to deal with the more lofty features and expressions of
the human spirit such as religion, art, mysticism, and ritual. The
two sciences of ritual and language at the opposite ends of such
an as yet entirely hypothetical hierarchy hold promise for the
development of a full range of intermediate sciences that deal
with the human animal in all its richness and complexity.
It has been claimed that many of the characteristic features of
 The Science of Ritual 351

man cannot be dealt with by scientific methods; an argument that

we shall subject to closer scrutiny in Chapter 29. So far we have
found (in Chapter 2) that science is not as narrow as many
empiricists, positivists, phenomenologists and hermeneuticians
have assumed. It is not confined to behavior, appearances,
quantities, tangibles, or measurable entities. As for "scientific
m e t h o d , " almost anything goes, provided we arrive at hypoth-
eses that account for the data. To support the claim that the
ancient Indians did not only develop a science of language, but
also a science of ritual, it is necessary to say more about the
concept of science that underlies such a claim. The following
sketch is not a philosophy of science, but lists a few rules of
thumb intending to show that, if it is justified to call chemistry,
economics, archeology, or demography sciences, it is justified
also to regard the occupation with ritual that we find in the
Srauta Sütras as such.
I regard the following four features as characteristics of
science, and present, though in varying degrees, in the sciences
I am familiar with:

(1) A science consists in part of a body of statements, rules,

theorems, or theories which aim at the true description
and analysis of some part of the world. There must be a
measure of empirical adequacy to these, which can be
established, not necessarily for all of them, by tests,
verifications, or falsifications, directly, indirectly, or at
least in principle.
(2) Complementing and extending this empirical adequacy,
there are abstract statements that go beyond the data, and
may be hypothetical, postulated, or speculative. Such
statements may include predictions, and be based upon
anything (e.g., a scientist's intuitions or dreams).
(3) This entire edifice of descriptions and generalizations
should be consistent to some extent; in particular, contra-
dictory statements can never be regarded as final. How-
ever, in most cases it is not easy to ascertain whether two
statements, even if fully explicit, are contradictory.
352 The Human Sciences

(4) There is a methodology of argument which addresses itself

to different viewpoints and propositions, and proceeds to
a shifting of options so as to reach a generally accepted
conclusion. The abstract statements mentioned under (2)
are sometimes arrived at through such a process. Logical
derivations within a theory may be regarded as formalized
specimens of such argumentation.

These characteristics are fairly vague, not operational, and

not free from redundancy. There may be sciences that are not in
accordance with some of them (e.g., mathematics). H o w e v e r ,
these characteristics are sufficient for my present purpose. As a
fifth one might add a feature that emphasizes the pure and
theoretical nature of science. This feature was much admired by
the Greek philosophers. Aristotle commented favourably upon
the notion of ûeœpia as 'desinterested contemplation'. Since
some of my colleagues who claiin to be scientists also claim that
they deal with what they deal with because it is ultimately useful
or relevant, I shall not include this feature. When necessary, we
may simply distinguish between ' p u r e ' and 'applied' science.
The Indian disciplines of ritual and grammar, which I have
come to regard as sciences, were in ancient India called sâstra.
This term has been translated as 'traditional discipline' or
'traditional science'. However, I would not regard everything
that used to be called sâstra as 'science'. At an earlier period,
ritual and grammar were considered vedânga, or '(disciplines)
auxiliary to the V e d a s ' . There existed a certain complementarity
between these two. In order to explain this, we need some
conceptual and terminological clarification.
Grammar, Sanskrit: vyäkarana, is a discipline which has
language for its object. In the work of Pânini, the object-
language is to some extent chandas, 'the metrical portions of the
V e d a s ' , viz., mantras; and to some extent bhâsâ, the spoken
language of Pânini's time, which has been shown to be co-
extensive with the domain of Vedic prose (primarily, the
Brährnanas, Äranyakas, early Upanisads, and Srauta Sûtras).
Ritual is an ambiguous term. On the one hand it refers to a
 The Science of Ritual 353

discipline, the 'science of ritual', which is embodied, in the

Vedic realm, in the Grhya and Srauta Sütras; and on the other
hand it refers to the object of this discipline. The two chief types
of ritual text correspond to the two chief types of object-ritual
(page 66): the grhya or domestic ritual which comprises so-called
life-cycle rites or sacraments and the srauta ritual, which is more
abstract and complex. Like in the rest of this book, I am mainly
concerned with the latter, because it exhibits the general fea-
tures of ritual more clearly and explicitly, and has also been
studied in the Srauta Sütras in a more thorough and scientific
manner.
The main technical device that the Indian sciences of language
and ritual introduced and continue to use is the concept of sütra
that corresponds to the contemporary concept of rule developed
in contemporary linguistics especially since Chomsky (Chapter
6). The convergence of these two concepts is remarkable for
although both operate with linguistic evidence, the Western
notion was partly inspired by the mathematical sciences and
logic, whereas the Indian notion originated in the context of
ritual. The related abstract concepts that contributed to the
development of Indian linguistics: categories such as meta-rules,
other meta-linguistic markers, features and elements, ordered
sequences and systems of rules of Chapter 5 also originated in a
ritual context. These devices and mechanisms enabled the
Indian ritualists to describe and analyze the complex structures
discussed in Chapters 10 through 12.
The art of composing in sütra style originated in the Srauta
Sutra of Baudhâyana (not later than the sixth century B . C . E . :
Caland 1903:11) and culminated in Pänini, who went farthest in
the direction of formalization (Renou 1963:180,198). The ^meta-
r u l e s " (paribhäsa) of Indian science also occur for the first time
in Baudhâyana. They were grouped together at the end of his
Srauta Sütra in a special section called karmänta, " e n d of the
karman s e c t i o n " (cf. the Vedänta or " e n d of the V e d a s , " which
came into being much later). Gonda has recently translated and
summarized the beginning of this section (Gonda 1977:509-510).
According to Gonda, this section illustrates the original meaning
354 The Human Sciences

of paribhäsä as a "discourse round t h e t e x t . " The following

excerpts are based upon G o n d a ' s renderings:

One should understand the ritual procedure from a group of five, viz.
from the metrical texts of the Veda (chandas, i.e. the mantra portions of
the Taittirîya Samhitâ), the brähmana, conviction or certainty as a
ground or means of action, by means of the method (nyäya), and by
means of the structure (samsthä). When we say 'From the metrical texts
of the Veda' that means that one should, in accordance with the order of
the mantras observed in the tradition, conclude: 'this act must be
performed first, this later'. Moreover, the very mantra announces the
ritual act, arçd explains the act . . . What one cannot execute by means
of the metrical texts one should try to execute by means of the
brähmana, for the brähmana prescribes with authority the purport of the
undefined mantras, viz. 'he performs this act with this mantra, that with
that'. For example, if the text mentions the mantra Tor refreshment
(food) thee, for strength thee' (TS 1.1. la) he cuts off a certain branch . . .
Moreover, a brähmana prescribes also the purport of acts that are not
accompanied by mantras, when for instance it reads: 'At a distance of
eight steps a brahmin should establish the sacred fire . . . .' ([Taittirîya
Brähmana] 1.1.4.1) . . . As to the expression 'by means of structure',
when the Soma has been stolen one should extract juice from ädära or
phälguna plants.

The last phrase seems to refer to substitutions, which are

always justified provided the structure of the ritual is retained.
Gonda continues with paraphrases of the text:

Section 2 deals with some fundamental concepts relating to mantras,

sacrificial rites etc. which are different in practice and applicability
(adhikarana). . . In Section 3 an explanation is offered of the distinction
between the 'warp' (tantra) and the 'woof (äväpa) of a sacrificial rite,
that is of the framework, standing model, or those components which it
has in common with other rites, and those that vary from ritual to ritual
and are therefore the special characteristic features. In Section 4 the
author deals with certain basic forms . . . defines and explains terminol-
ogy, etc. . . . The seven Soma rituals are enumerated . . . Section 5
answers the question of the difference between the two terms used in the
Baudhäyana corpus pürvä tatih and uttarä tatih 'the antecedent and the
subsequent series of ceremonies'. The standard ritual is pürvä tatih and
what one arranges (modifies) is uttarä tatih; e.g., the establishment of the
ritual fires is pürvä tatih, the re-establishment uttarä tatih', of the
vegetarian sacrifices (isti), the full and new moon sacrifices are the pürvä
 The Science of Ritual 355

ta f//z, all the optional rites uttarä tatih . . . In Section 6, the author turns
to the use of mantras, for instance: "One should not, for the sake of a
ritual act, interrupt a mantra". In case of the immolation of a victim the
mantra that is handed down is short, the act long, in other cases however
the act is short but the mantra long . . . .

Baudhäyana is still relatively verbose, but his meta-rules, like

all other rules, are formulated more succinctly in the later ritual
Sütras. As a simple illustration of the Indian science of ritual, I
provide here a translation of a portion of the twenty-fourth
section of the Äpastamjba Srauta Sütra of the fifth or fourth
century B . C . E . , . first translated into German by Caland (1928),
which primarily lists meta-rules:

Äpastamba Srauta Sütra 24A.I-26

(1) yajnam vyäkhyäsyämah 'We shall explain that ritual',

(2) sa trayänäm varnänäm hrähmanaräjanyayor vaisyasya ca
i t pertains to the three classes, brâhmana, ksatriya, as
well as vaisya',
(3) sa tribhir vedair vidhïy at e 'it is enjoined with the three
Vedas',
(4) rgvedayajurvedasämavedaih 'with the help of Rgveda,
Yajurveda, and S ä m a v e d a ' ,
(5) rgvedayajurvedäbhyäm darsapürnamäsau 'the full and
new moon rituals with Yajurveda and Rgveda',
(6) yajurvedenägnihotram 'the Agnihotra with the Yajur-
veda',
(7) sarvair agnistomah 'the Agnistoma with all (three)',
(8) uccair rgvedasämavedäbhyäm kriyate '(the rites) are per-
formed with Rgveda and Sämaveda (recited and chanted,
respectively) with a loud voice',
(9) upämsu yajurvedena 'with Yajurveda softly',
(10) anyaträsrutapratyäsrutapravarasamvädasampraisais ca
'except with the ritual call (o srävaya), answering call
(astu srausat), pravara (lineage), samvâda (ritual dia-
logues) and c o m m a n d s ' ,
356 The Human Sciences

(11) antarä sämidhenisu anücyam 'in the case of the firewood

verses the utterance is intermediate',
(12) mandrena präg äjyabhägäbhyäm prätahsavane ca 'at a
low pitch before the äjyabhäga oblations (in an isti) and at
the morning pressing',
(13) madhyamena prâk svistakrto madhyandine ca 'at a middle
pitch before the svistakrt oblation and at the midday
(pressing)',
(14) krustena sese irüyasavane ca 'at a high pitch elsewhere
and at the third pressing',
(15) väksamdravas ca tadvat 'and t h e pace of the voice is
similar',
(16) rgvedena hot a karoti 'the Hotä performs with the
Rgveda',
(17) sämavedenodgätä 'the Udgätä with the S ä m a v e d a ' ,
(18) yajurvedenädhvaryuh 'the Adhvaryu with the Yajurveda',
(19) sarvair brahmä 'the Brahman with all (three)',
(20) vacanäd vipratisedhäd vänyah kuryät 'when it is stated
explicitly or in case of contradiction another should per-
form',
(21) brähmanänäm ärtvijyam 'the priestly functions belong to
brahmins',
(22) sarvakratünäm agnayah sakrd ähitäh 'in all rites the fires
are set up o n c e ' ,
(23) juhotïti codyamàne sarpirâjyam pratïyât 'when 'he makes
an oblation {juhotiY is enjoined, 'of ghee' should be
understood',
(24) adhvaryu kartâram 'as agent the Adhvaryu (should be
understood)',
(25) juhüm pätram 'as implement (ladle) the juhü\
(26) vyäprtäyäm sruvena 'when (the juhû is already) used, with
the sruva'.

Most of these rules and metarules are clear but some are of a
more technical nature. The first serves as a heading or caption
for the section that follows, a customary practice in Indian
" t e x t s " because they are not texts but oral compositions.
 The Science of Ritual 357

Metarules 2 through 9 are straightforward, and 10 deals with

specific exceptions, several pertaining to the Isti (Chapter 9A).
12—14 capture a generalization that applies at the same time to
three subdivisions of the Isti and to the three pressings of Soma.
15 in its brevity announces the later sütra-style which is more
formulaic: it states, when expanded (as is done in vrtti, 4 'ex-
p a n s i o n " glosses) that the pace of the voice on the three
occasions in both rites (the Isti and the Soma pressing) is,
respectively, slow, medium and fast.
Metarules 16-19 pertain to the "users":.-they are " p r a g m a t i c "
rules, if we decide to extent the usage of the term in this manner.
With 20 we reach a special case of the principle of contradiction
that recurs in similar form and with numerous applications in
linguistics, logic, philosophy, and other disciplines (see Staal
1988a: 109-128). Metarules 21 and 22 are self-evident. In 23-26,
lastly, we have a striking parallel to the structure of Pânini's
grammar: when nothing else is stated, the entities mentioned
(ghee or clarified butter, the Adhvaryu priest, l\\zjuhü ladle) are
understood. When something else is stated, it prevails. For
example,
4
'he makes the X oblation"

.always means:

"the Adhvaryu makes the X oblation of ghee with thejuhù."

But if the Sütra text states, for example,

"the Pratiprasthâtâ makes the X oblation"

it means:

"the Pratiprasthâtâ makes the X oblation of ghee with the juhü."

This'is parallel to Pänini's metarule 2.3.1 (anabhihite: page 44),

which states that, in the kâraka theory, nothing is expressed
more than once. The ritual parallel is more or less obvious.
358 The Human Sciences

While in grammar, the rules block an ungramrnatical sequence

by preventing its derivation, in ritual it prevents impossibilities
like two different priests or ladles doing the same thing at the
same time and plac€.
Harold Arnold has drawn attention to another metarule which
occurs somewhat later in Äpastamba's paribhäsä section:
24.2.13: antaräni yajnängäni hähyäh kartärah " t h e ritual limbs
should be inside, the officiating (priests) o u t s i d e " . " L i m b s "
refers primarily to the implements, and " i n s i d e " and " o u t s i d e "
mean closer and less close to the fire. The Srauta Sütra of
Satyäsädha uses similar expressions. K ä t y ä y a n a ' s Srauta Sütra
(1.8.31-32) expresses similar restrictions, but in a different way:

(31) havispätrasvämyrtvijäm pürvam pürvam antaram 'with

respect to closeness to the fire, each comes before the
next in the sequence oblation-implement-patron-priest',
(32) rtvijäm ca yathäpürvam 'and the order of the priests is
the order of their election' (viz, in which they were
elected at the beginning of the performance).

These distinctions, introduced by the ritualist Kätyäyana, are

related to an important distinction introduced by the grammarian
Kätyäyana (perhaps the same as the ritualist: Thieme 1937-1938)
between the concepts of antaranga and bahirahga. These apply
in linguistic morphology to internal and external constituents.
For example, a rule applicable within a word is antaranga with
respect to a rule which applies across word boundaries. The
general principle which governs the use of these concepts is
defined in terms of the technical concept of asiddha, itself
defined as follows:

"Rule A is asiddha with respect to rule B" if and only if "Rule A is to

be regarded as not having taken effect when rule B is to take effect."

This concept and the specific order of rules to which it applies

play a crucial role in grammar (see, e.g., Joshi and Kiparsky
1979). Having defined asiddha, we define the use of the pair
antarahgalbahiranga by means of a metarule:
 The Science of Ritual 359

Bahiranga rules are asiddha with respect to antaranga rules.

Renou has studied the terminological and stylistic similarities

between ritual and grammar in an important article (Renou 1941-
1942, republished in Staal 1972:434-469). He deals not only with
rules and metarules, but also with the use of particles expressing
generalizations, options, exceptions or implications, the role of
substitutions, primary and derived elements, prototypes and
extensions or variations, the distinction between essentials and
incidentals, the construction of larger entities from smaller
elements, systems of relations, etc. Anyone who wishes to
obtain an idea of the connexions between the Indian sciences of
ritual and grammar will benefit from this study.
R e n o u ' s conclusions with regard to the terminological and
stylistic similarities between the two sciences can be extended to
the structural, logical and methodological connexions we are
concerned with in the present context:

These stylistic and terminological parallels between ritual and grammat-

ical theory show that we are dealing with disciplines which originated in
the same circles, but which answered complementary needs. Both
pertain to the practice of the sis ta, the specialists . . . When dealing with
a particular term, it is not easy to establish whether it originated with the
grammarians or the ritualists: in the absence of a fixed chronology of
texts, and with the general parallelism of techniques in ancient India,
such a search becomes arbitrary. However, in the majority of cases it is
clear that the point of departure lies in the religious texts. Grammar
appears as a specialized investigation within the larger domain of explicit
ritual science. The extent and importance of the religious literature, the
undeniable priority of the mantras and of the ritual forms which they
presuppose, invite us to look for origins in that domain.

R e n o u ' s views (which make better sense if we substitute

" V e d i c " for the two occurrences of the ail-too Western term
" r e l i g i o u s " : see Chapter 28) suggest that the Indian science of
ritual was earlier than the Indian science of language.
What remains to be shown is that the Indian traditional sästra
disciplines of ritual and grammar are indeed sciences in accor-
dance with the four characteristics of science laid down at the
outset of this section, viz., empirical adequacy, abstraction,
360 The Human Sciences

consistency, and methodology. I shall take up these features one

by one.

(1) Empirical adequacy. In the Introduction to his

Mahabhâsya, Patanjali has formulated the 'potter's principle'
(page 40): 4if you want pots, you go to a potter, but if you want
w o r d s , you don't go to a grammarian'. Where do you go? To
loka, the world, for example, the market place where people
speak the language. Grammar is a curious enterprise, that may
appear circular to those who do not understand it: the grammar-
ians try to derive correct forms, but the correctness of the forms
is already known from those who know the language. In fact, in
case of conflict, it is the speakers of the language who decide. If
grammar provides a derivation of a form that is not used, it is
grammar that is wrong. Thus grammar aims at empirical ade-
quacy, and all its efforts are aimed at reaching this goal. N o t e
that this is not different from the situation in physics, that
paragon of modern science: physics tries to account for facts
with which we are already familiar—e.g., that apples fall.
Ritual science is in precisely the same position. The Srauta
Sütras account for rituals and rites that are performed in
accordance with specific traditions. Accordingly, neither gram-
mars, nor ritual manuals are teaching manuals. They can only be
understood by people who are already familiar with the subject.
They are composed for the sake of the subject only. Other
reasons given are mostly rationalizations.
This characterization implies further that both the Astâdhyâyï
and the Srauta Sütras are primarily descriptive and not prescrip-
tive. But at this point we have to keep in mind, that what was
originally descriptive, has become increasingly prescriptive.
Pânini's grammar originally reflected the usage of the sistas,
those who spoke Sanskrit properly; but to later generations, its
expressions became law, and thus contributed to the fixation of
Sanskrit. The ritual sütras have similarly become prescriptive.
Some contemporary ritual performances have been revivalistic
in precisely this sense: attempts have been made to follow
particular Srauta Sütras. However, there are also living srauta
 The Science of Ritual 361

traditions in India. The Nambudiris, for example, follow their

own tradition. When told that Baudhâyana describes certain
rites differently, they say 'Interesting', but would not for a
moment consider a change in their own proceedings.
The transition from description to prescription indicates in-
creasing dogmatism. We found this already indicated by the
jnäpaka notion, and the resulting jnäpakasiddhaparibhäsäs. The
general reverence of the later grammarians for the trimuni, 'the
three sages', Pänini, Kätyäyana, and Patanjali, illustrates this
further. Even Patanjali himself accepts Pänini as an authority, in
whose grammar no single element may be without meaning or
purport {anarthaka). However, Ojihara (1978:227) has rightly
pointed out that this does not imply that Pänini's rules are
beyond criticism, but that they have to be subjected to a
thorough examination in order to find out what is their ultimate
motivation (prayojana). Wherever precisely we draw the line,
the claim that ritual and grammar were sciences in ancient India
pertains to the earliest works, the Astâdhyâyï and the Srauta
Sütras. It does not hold true of the later developments, which did
become increasingly dogmatic and scholastic.
One feature of description that has contributed to the devel-
opment of ritual and grammar in India is the formal nature of
these disciplines. Emphasis on form is a general characteristic of
Indian civilization. The Vedas refer more to the forms of
language and rites, than to their meanings or function. The
Brâhmanas introduce large scale interpretations of ritual, but
these can often be shown to b.e failures. Language is of course
concerned with meaning, and semantics is basic to Pänini's
grammar; however, once the derivations start, they are fully
explicit and formal. The Srauta Sütras do not provide t h e rites
with any meaning. This inherent formality has contributed
significantly to the scientific character of the study of ritual in
India.

(2) If grammar, physics and ritual accounted only for what we

knew already, the accounting might be interesting but the results
would not fail to disappoint us. It is here that abstraction
362 The Human Sciences

becomes significant. Much of the discussions around the texts

initiated by the paribhäsäs are concerned with this dimension. In
another famous passage in the Introduction to his commentary,
Patanjali has explained that the forms of language are infinite,
and so cannot be enumerated; only general rules and exceptions
can account for them (page 40). He elucidates this with a
reference to ritual, which is equally significant: For the same
holds for the satt ras, or rituals of indefinite duration (page 89).
Hillebrandt had ridiculed these thousand-year performances,
which obviously no human being can engage in. But they give
expression to the recursive procedures which constitute the
essence of the ritual of the Srauta Sütras, and without which
they would constitute not a science, but a mere inventory.
The principle of brevity, expressed by a famous metarule
about the saving of half a mora, may contribute to the expression
of generalizations, but it is at best a mnemonic device or a game,
not a scientific principle. Buiskool, Cardona, and others have
shown that Pânini's grammar is not actually governed by this
principle, but formulates rules through functional generaliza-
tions. The minimization of syllables has been an end in itself only
among minor grammarians.

(3) Consistency is a matter of description or a feature of a

theory; facts are always consistent. In grammar, I have already
referred to the rule vipratisedhe param käryam, which would
have safeguarded the consistency of the grammar if it had been
universally applicable. Whether the Astâdhyâyï is indeed con-
sistent is a matter of continuing discussion. Attempts to save it
can be made from various points of view: recourse may be had
to emendations of the text, to postulating interpolations, to
decomposing and dissecting the text into various layers and
portions, or to the jnäpaka gymnastics of the later commenta-
tors. It is important to note that any of these strategies can only
be fruitful when we are dealing with a text such as the
Astâdhyâyï, which is basically set up as a logical and rational
account of a specific area of experience.
In ritual, the situation is different. There is less scope for
 The Science of Ritual 363

inconsistencies to appear because of the variety of options and

schools. The Srauta Sütras do not recommend or argue for
specific options; they merely describe and analyze the tradition
they have selected and opted to follow. Do they not then accept
the Vedic injunctions {vidhi) enjoining certain rituals for certain
purposes? N o , not explicitly, for the matter has been entirely
ritualized. The Vedic vidhi is described only in as far as it is part
of the formal declaration of intent (samkalpa) which the
yajamäha makes at the outset of a ritual performance. Ritual
options, moreover, are compatible, unlike some articles of faith,
religious convictions, or philosophical doctrines.
This compatibility is consistent with the characteristics of
outer chance developments (page 344) and with the explicit
declarations of Indian philosophers, e.g., Sankara in the Intro-
duction to his Vedänta Sütra Bhäsya where he stated that
opposing injunctions prompting rites are compatible as options,
but opposing truths are incompatible since they contradict each
other:
There is no option as to whether a thing is thus or thus, is or is not.
Option depends on human notions. Knowledge of the nature of a thing
does not depend on human notions. It depends only on the thing itself.
To say with regard to a pillar "it is a pillar or it is a man or it is something
else" does not result from correct knowledge. To say that it is a man or
something else results from false knowledge. To say that it is a pillar
results from correct knowledge, because it depends on the thing itself.
Therefore the means of knowing objects, that are existent things, depend
on the things themselves (quoted in Staal 1988a: 119).

(4) Methodology, lastly, is implicit in Sahkara's observation

just quoted: the principle of non-contradiction which is alluded
to here and explicitly formulated elsewhere in Indian logic and
philosophy was first stated as a metarule in Äpastamba Srauta
Sütra 24.1.20. Other logica! distinctions and principles were
formulated in Patanjali's Mahäbhäsya or " G r e a t C o m m e n t a r y "
on Pânini's grammar. It must be assumed that several of these
technical insights and devices were known to ritualists and
grammarians before Patanjali: for the latter, including Pänini,
confined themselves to formulating rules and did not adopt the
364 The Human Sciences

more discursive commentorial style which Patanjali introduced.

In India, logic developed from ritual and grammar; its history,
though as checkered as the history of Western logic, developed
along different lines (Staal 1988a, pages 37-48). Ritual rules and
metarules were subsequently transmitted, via the Mimämsä
philosophy of ritual, to other sciences and to the legal systems of
the Dharmasästra whence they entered Indian jurisprudence.

It is clear that we are entitled to conclude that the Indian

works and disciplines dealing with ritual and language may be
regarded as specimens of science. The degree of scientifically in
all these works need not be the same, and their efficiency and
successfulness may vary, but this applies to sciences and
scientists everywhere.
There are three final observations to be made. It may come as
a surprise to many that ritual has anything to do with science. Is
ritual not pervaded by magic and superstition? Is the great
achievement of science not that it has emancipated from ritual?
In India itself there is ample historical reason for such feelings of
ambiguity with respect to ritual. The Upatlisads, Buddhism, the
Vedänta, as well as modern secularism—all declare unani-
mously that ritual is useless and in fact unworthy. By contrast,
the authors of the Srauta Sütras obviously believed in the
efficacy of ritual. At this point we have to pay attention,
however, to a remarkable fact: whatever were the beliefs of the
authors of the Srauta Sütras, these beliefs did not interfere with
their science at all. I would go further than this, and claim, that
there is no modern or contemporary scholar of ritual—whether
a student of religion, a social scientist, or a psychologist—who is
as little influenced in his studies by his beliefs as the authors of
the Srauta Sütras were by theirs. However, what is true of the
Srauta Sütras, is neither true of the Brähmanas, nor of the later
Mimämsä. The Brähmanas indulge in unhampered speculation,
and are as arbitrary as contemporary trends in the study of
religion. The Mïmâmsâ, on the other hand, is a system of
philosophy that adheres to the axiom that the Vedic dkarma as
expressed by the vidhi injunctions is conducive to the highest
 The Science of Ritual 365

good. The Mïmâmsa has paid much attention to specific ritual

problems; but the Indian tradition is on the whole right in
classifying it as darsana, 'philosophy'. It became increasingly
philosophical when it took to arguing with the Buddhists. The
Srauta Sütras are different from the Brähmanas and Mïmâmsâ
both, and I regard only their efforts and achievements as
scientific in nature.
The mixture of scientific and un- or less scientific works and
activities is characteristic of almost any period of the history of
science or civilization. It explains Hermann Oldenberg's expres-
sion vorwissenschaftliche Wissenschaft, "pre-scientific sci-
e n c e , " which was intended as a characterization of the
B r â h m a n a literature (Oldenberg 1919). Brian Smith (1986:93,
note 3) has written that my " s c i e n c e of ritual" " w o u l d seem to
be an updating of Oldenberg's t e r m . " In fact, I have tried to
show that the Srauta Sütras are scientific, and the Brähmanas
un-(rather than: pre-) scientific. Of course, the distinction be-
tween the two classes of works is not as hard and fast as the
characterization suggests, but this is not at issue. Whether
vorwissenschaftliche Wissenschaft is a contradictio in adjecto is
another question. In any case, the term " s c i e n c e " should be
confined to what is scientific and not pre-scientific. If Smith were
right, Pänini's grammar would also have to be regarded as
"pre-scientific."
At the outset of this chapter I referred to attempts that were
made to establish that science is exclusively Western. In the
course of the chapter I have not shown that science in the
conventional English sense—that is, " p o s i t i v e " science—did
actually originate in India. The reason is simple: it didn't. India
has produced great mathematicians and astronomers, and has
evolved chemistry, botany, medicine, and other natural sci-
ences. In all these respects, India is on a par with other great
civilizations. However, with respect to the human sciences of
ritual and grammar, the Indian contributions have not only been
outstanding, but unique.
In our contemporary world, the human sciences have been
much neglected. It has become a cliché to state that man
366 The Human Sciences

understands the stars but not himself. The strength of contem-

porary science lies in the mathematical, physical, and biological
sciences. The contributions made by the contemporary human-
ities and social sciences are comparatively dismal. The best that
some of them have evolved is a narrow empiricism. Rational
speculation has generally been absent, and theories have been
rare and haphazard. There is excitement about fashions but no
sense of progress. It has in fact been argued that the sciences of
man cannot be sciences, and that they require unique categories
such as " V e r s t e h e n " that conflict with all rational approaches—
as if man could in principle not be studied objectively, that is,
studied at all (Chapter 29).
Linguistics is the only field of study that has finally escaped
from the traditional humanities and the behavioristic biases of
the social sciences, and has emerged as a scientific discipline. It
is therefore not in the least surprising that in recent years
significant similarities have been discovered between contempo-
rary linguistics and the Indian science of grammar.
In the field of ritual the situation is different. There is nothing
in contemporary research that resembles or even approaches the
scientific achievements of the Srauta Sütras unless it is derived
from these sütras, such as Hubert and Mauss (1897-1898). This
is not due to the fact that humans do not engage any longer in
ritual. We do, and so do our animal relatives, whose rituals have
actually been neglected less. What we lack is not rituals, but
ritualists, and a science of ritual. For these reasons it is not
merely interesting that the ancient Indians had a science of
ritual. Nor is it simply a fact that may help us to better
understand India and Indians. The science of the Srauta Sütras
is the only science of ritual that we are in a position to study and
contemplate. It will inspire us once we realize ourselves that we
need to create and develop such a science.
The third and last observation I wish to make is this. It is clear
that no science of ritual would easily fit into the customary
subdivisions of academic disciplines. Though Sanskritists have
greatly contributed to the study of ritual, its science does not
belong to the humanities as it is independent of language and is
 The Science of Ritual 367

moreover concerned with other animals than only people. Nor is

it a natural or à social science. Social sciences lack the tools to
analyze it (Chapters 14 and 25). Some of these comments apply
to linguistics. In the United States, linguistics has been incorpo-
rated among the social sciences partly because these were at one
time believed to attract financial support more easily than the
humanities. Practical considerations aside, the case of ritual
demonstrates that existing classifications of the sciences—and
especially the alleged distinction between sciences and human-
ities—lack a serious foundation.
 27

Oral Traditions

The human sciences could have originated anywhere *and

under any circumstances, but from the available evidence it
seems that it happened in India some time between 1,000 and 600
B.c.E.. We can now pinpoint some of the reasons. Indians were
careful observers of what was going on inside them: thus they
discovered that different sounds were produced by different
movements of the mouth and the tongue (page 36). They also
paid attention to breathing, an attention that sometimes turned
into an obsession, and that pertains tó the pronunciation of
sounds of language, the recitation and chanting of mantras,
Yoga, and pulmonary development in general. Long breath and
the breathing of mantras such as OM and HIM (page 275) are
therefore not only probable precursors of the sounds of language
but also conducive to health—in any case, not worse than y
jogging.
Other civilizations have also paid attention to breath. In
Sufism, mystics speculated on dhikr although they rarely went
beyond the ascertainment that the term was used in eighteen
senses in the Koran. In Taoism, speculations on respiration
techniques went farther (see, e.g., Maspero 1950:107-114;
Schipper 1982:182,202,206) and empirical observations resulted
in sexual gymnastics and healing techniques connected, for
example, with accupuncture. In India, similar speculations on
präna led to similar techniques, connected with diet, massage,
Yoga, and sexual gymnastics. Historical connections between
China and India, across the (eastern?) Himalayas, have to be
assumed. According to Filliozat 1946, followed by Eliade 1969:
413, the influence went from south to north, or west to east,
370 The Human Sciences

according to Needham 1956:427^428, the other way round. But

such isolated techniques, cures, and speculations do not make a
science as does a system of structured explanations of the kind
characterized in Chapter 26. These observations, therefore, do
not refute the idea that the earliest known examples of human
sciences are the Indian sciences of ritual and language.
H o w and why did these sciences originate? " I n t r o v e r s i o n " is
not enough. To explain their origins in detail would be a large
undertaking (for more details see Staal 1986b), but the historical
development in India shows that their origins are closely related
to the fact that ritual and mantras, like language, were orally
transmitted and that safeguards for the dependability and fidelity
of their transmission with exclusive emphasis on formal correct-
ness were required. Both sciences originated from the twofold
analysis of the continuous recitation of the Vedas into their
4
'word-for-word" recitation called Padapätha (Jha 1973,1975,
1976) and the setting to music of that Padapätha immediately
after its constitution, that is, around 1,000 B . C . E . (Chapter 5).
These investigations led to the analysis of mantras in their ritual
setting and to formulating a system of sandhi and other phono-
logical rules. Both types of analysis made use of rules and led to
general, theoretical discussions on the- notion of rule.
* We can only understand this development and what happened
subsequently when we take note of a remarkable fact: the art of
writing was not known in ancient India. It is possible that a
writing system had existed in India before the Indo-European
semi-nomads arrived and composed the Vedas. For in the course
of excavation of the remnants of the earlier Harappa civilization,
a few hundred seals were found which contain symbols or
symbolic shapes that could be interpreted as a form of script. At
the present state of our knowledge these symbols, which have
not been deciphered, may be regarded as a script, ownership
marks, astrological formulas, or something else. Whatever their
nature, it is clear that there are no links between these i n s c r i p -
tions' and the later Indian scripts that are based upon forms
introduced into India from the Near East, probably not long
before the third century B . C . E It is also clear that there were
 Oral Traditions 371

originally no connections between the semi-nomadic Indo-Eu-

ropean groups that entered India from the northwest and com-
posed the Rigveda, and these large and sedentary indigenous
city civilizations of earlier centuries and millenia of which only
material remnants are left.
The earlier literature contains several references to script, but
we do not know what kind of writing the authors had in mind.
What is clear is that the first Indian uses were confined to royal
edicts and commercial transactions. Writing was alright for
keeping accounts but it continued to be emphatically and metic-
ulously excluded from the ancestral traditions which were
considered too pure to be written down. The low evaluation of
what was regarded as an alien and barbaric invention is illus-
trated in a variety of quotations, some assembled by Ghurye
1950, others by Staal 1961 (1,1). For example, Aitareya
Âranyaka 5.5.3: a pupil should not recite the Veda 'if he has
eaten flesh, or seen blood, or a dead body, or done what is
unlawful . . . or had intercourse, or written . . . \
The most remarkable feature of the Indian scripts is not their
shapes but their scientific arrangement which is basically t h e
same in all the many forms with which we are familiar. Instead
of the haphazard ABC Y of the West, the Indian scripts begin
with the series of vowels—basically # , / , « , e, o, ai, au—
followed by the consistently ordered consonants beginning with
ka, kha, ga, gha, nga, etc. It has been noted long ago that this
arrangement shows that the scientific analysis of the sounds of
language was completed in India before any script was intro-
duced, and was so widely known that the newly introduced
invention was adapted to this analysis as a matter of course.
Renou and Filliozat (1953:668) went one step further: fcOn doit
même remarquer à ce propos q u ' u n e écriture alphabétique du
type sémitique aurait pu entraver les études phonétiques si elle
eût alors existé dans l'Inde, car elle aurait donné le modèle d'une
analyse commode mais non scientifique des sons du langage'
('One is forced to observe in this context that a Semitic type of
writing would have hindered phonetic studies if it had existed at
the time in India, because it would have provided a model of
372 The Human Sciences

analysis of the sounds of language that was practical but not

scientific').
Ritual is not only often transmitted without the help of writing',
it is often transmitted without the help of language. Like many
other features of culture and civilization—cutting, digging, aim-
ing or planting; features of musical scales and melodies, visual
patterns, motifs and shapes, dances, stellar constellations, cook-
ing, the construction of ploughs, weapons and altars, the ele-
ments of arithmetic and geometry—ritual is often transmitted by
demonstration: see Frontispiece and above page xv. Language is
remembered together with meaning but mantras are remem-
bered from their sound and a demonstration or oral exposition of
their ritual setting.
The oral transmission of the Vedas inspired many special
techniques. In addition to the Padapâtha and Samhitäpätha, the
Prätisäkhya literature introduces 'modifications' (vikrti) that are
based upon the Padapâtha, and that 'strengthen' the oral tradi-
tion, that is, minimize the chance of a single word being lost. The
first of these is the Kramapätha, in which each word is repeated
once but in such a way, that it is first linked through sandhi with
the word that preceeds, and then with the word that follows. In
other words, if the separate words of the Padapâtha are referred
to by numerals as follows:

1 111 3 I AI'5 I . . .

the Kramapâtha becomes:

12/23/347 45/ ...

It is clear that sandhi combinations of the Samhitâpâtha are

reintroduced here, for they obtain within each of these pairs, but
not between the pairs. At the same time, it becomes more
difficult to forget a single word: for if a pair were forgotten, the
continuity between the succession of pairs would show a break.
Note that this does not apply to the recitation of the Samhitä or
Padapâtha themselves: if a word is forgotten in their recitation,
it does not leave a trace. T h e study of the Kramapâtha,' says
 Oral Traditions 373

the Prätisäkhya of the Atharveveda (4.108), 4 has for its object

the fixation (dârdhya) of Samhitâ and Pada.'
Subsequent modifications render the oral transmission even
more firm and stable by introducing methods that resemble the
scanning of a tape by a computer, e.g.:
Jatäpätha:
1221 1 2 / 2 3 3 2 2 3 / 3 44 334/. 4 5 5 4 4 5 / " . . .
Ghanapätha:
1221 123321 1 2 3 / 2 3 3 2 2 3 4 4 3 2 2 3 4 /
3 4 4 3 3 45 543 3 4 5 / . . .
Such techniques of oral transmission introduce new sandhi
combinations (e.g., -2-1'., '3-2') that did not occur in the
Samhitäpätha and thus further minimize the probability of a
single word being lost. It is not surprising that as a result of these
widely practised mnemonic techniques of oral transmission,
there is less variation among the oral traditions of the Vedas than
among the manuscripts of much later date that Western scholars
have used and that are themselves based upon these oral
recitations.
The K r a m a p â t h a is fully explained in the Rkprätisäkhya and is
mentioned by Pânini; it must therefore be at least as old as the
6th century B.C.E. The other modifications are of later date. All
survive at the present day and can still be heard in many parts of
India (cf. Staal 1961, Chapters 2, 5 and 6; Levy-Staal 1968).
The fixation of the oral tradition by these mnemonic tech-
niques pertains only to the form of the mantras; there is no
corresponding tradition that fixes and preserves their meaning.
The interpretations of the tradition of ritual and mantras, and the
meanings assigned to them have therefore always changed. This
succession of interpretations is part of the history of the philos-
ophies of India. The exclusive emphasis on form also explains
the doctrine of the ritualist Kautsa, according to whom the
mantras were meaningless (anarthakâ manträh: above page
112). This view should not be interpreted in terms of scepticism,
positivism, or behaviorism. It is a purely ritualistic stance that
374 The Human Sciences

limits the function of mantras to the ritual use which is their

proper home, just as language is the proper home of meaning.
K a u t s a ' s position is easier to understand when we remember
a feature of the Vedic compositions that is closely connected
with the emphasis on their form and the ensuing techniques of
their formal transmission: their obscurity. Unlike epic descrip-
tions, which are explicit and full of attention to detail, the Vedas,
and especially the Rigveda, were allusive even in their original
context. Thieme has shown that in apparently simple songs such
as Rigveda 1.32, which extolls the heroic deeds of Indra, the
poet does not describe events in d e t a i l ' a n d in their natural
succession, but refers and alludes to them in order to show how
heroic they were, how deadly to the enemy, how memorable for
the future, and so forth (Thieme 1957:88-89). Such descriptions,
addressed to insiders and presupposing specific knowledge soon
begin to be misunderstood.
Forms that are not understood tend to be transmitted with
emphasis on their formal properties, a feature that is also
connected with their ritual use. That a formal analysis developed
from this formal transmission is related to another difference
between the transmission of the epics and that of the Vedas: the
former were transmitted by bards, addressing villagers in open
settings; the latter were handed down from father to son or from
teacher to pupil in strict and often secret isolation. Such facts
and others (Staal 1986b) prevent the facile application of con-
cepts that elsewhere have been successfully applied to more
popular oral epics. They also show that the thesis that science
and literacy are necessarily connected (Goody 1968) is not
generally valid, and throw doubt on the distinction between
literate and non-literate forms as a significant distinction be-
tween advanced and less advanced forms of civilization.
To understand Kautsa let us visualize—or rather: imagine the
audition—of an actual ritual performance. When the blindfolded
Grâvastut priest recites his Rigveda verses at the midday press-
ing of the Soma (page 76), he seems in fact—as his name
indicates—to be addressing the pressing stones {grâvastut).
Contemporary scholars of religion are prone to accept strange or
 Oral Traditions 375

exotic facts without question as possibly endowed with spiritual

value—a variant, at first sight, of Quine's "principle of c h a r i t y " :
we should always interpret statements in such a manner that
they make as much sense as possible (Quine 1981:41; above,
page 26). Such " t o l e r a n c e " may signal a kindly disposition but
is not necessarily flattering to the ritualist. It is more reasonable
to assume that ritualists are rational beings who do things that
make sense. That is the real import of Quine's principle, and it
is exactly what Kautsa did.
Kautsa did not question the Grävastut's act, but its customary
interpretation: for how can a person in his right mind address
inanimate objects? He pointed out that mantras do not merely
seem to address stones or herbs; they also seem to refer to things
that do not exist (e.g., a being with four horns, three feet, two
heads and seven hands) and be redundant and self-contradictory
(e.g., one mantra asserts: " T h e r e is only one Rudra, there never
was a s e c o n d , " and another refers to " t h e innumerable thou-
sands of R u d r a s " ) . Finally, he emphasized what we have just
observed: that there is a tradition for mantras to be learned by
heart, but no corresponding tradition to teach their meaning.
All these contradictions are based upon the assumption that
mantras refer or convey meaning like the declarative statements
of ordinary language. Kautsa concluded that the assumption is
wrong and that mantras are meaningless although they have to
be recited on the proper occasion. Though K a u t s a ' s views are in
accordance with the S rauta Sütras, they were criticised in an
early work on etymology, the Nirukta, and also in the later
Mïmâmsâ Sütra. Yet Kautsa's views exhibit the orthopraxy or
insistence on right practice that characterizes most Indian tra-
ditions. The philosopher Sarikara, for example, approvingly
quotes a text which declares: " h e who teaches a mantra or
officiates at a ritual with mantras without knowing their compos-
er-seer, meter, deity and brähmana, will run his head against a
pole or fall into a pit" (Vedänta Sütra Bhâsya 1.3.30). In this list
of requirements, Saiikara does not include knowledge of mean-
ing: all he demands are formal data transmitted together with the
376 The Human Sciences

recitations. The composer-seer, meter and deity are all recited

names. The brähmana is merely another recited text.
An oral transmission that excluded meaning and emphasized
form created ideal conditions for the origination of science. Max
Weber called attention to this "formal rationality" and regarded
it as " t h e essential differentiating factor of Western civilization"
(Goody 1968:65). But India and China refute his contention. In
the Indian science of ritual the formalities are even more
formidable than in grammar. Within the long ritual recitations of
the Soma sequences (Chapter 9B), the order of mantras is
always different from the order in which they originally occurred
in the Veda from which they were taken. This is either due to
chance (Chapter 24) or follows from the application of formal
criteria pertaining to sound, length, meter, etc. We thus find that
the original meaning has disappeared and that the resulting
assemblage conveys no meaning either.
A prerequisite for the traditional study of ritual is that the
student knows his own Veda by heart. He must know it
thoroughly, from beginning to end. When given any couple of
w o r d s , he must be able to continue the recitation from there. If
he is good or takes pleasure in games, he can recite it backward;
recite every other word; do with the words anything that a
computer can be programmed to do; single out or couiit their
occurrences, group them together according to certain criteria;
in brief, perform precisely the kinds of exercise of which the
vikrti 'modifications' are simple examples. On this foundation he
can learn to change the traditional order that he has committed
to memory; and here we witness the beginning of those extraor-
dinary exercises that are the bread and butter—or rice and
ghee—of Vedic ritual. Most of these make no sense in terms of
meaning, and often little sense even in terms of form; because
many of them were, at the outset and at least in part, either due
to intuitions that arc no longer recoverable, or simply due to
chance. Once put together, these exercises can be learnt. There
may be elements that facilitate their study, for example, the
occurrence of certain words; such as the word for dawn—usas—
that the pupil will be familiar with even if he need not know what
 Oral Traditions 377

it means. Or 'Agni,' for that matter; much more c o m m o n and

familiar; yet to the young scholar who is beginning to find his
way in the ritual maze, primarily nothing but a sound. Mantras
recognizable because of a name or term they contain are the
"perfect f o r m " mantras that address (page 196).
To illustrate the formal complexity of ritual recitation, I shall
discuss the "Morning L i t a n y " (prätaranuväka) recited by the
Hotâ priest before dawn on the Soma pressing day. This
recitation lasts about an hour and consists (in the Kausïtaki
recension: AGN11:600-601) of 360 mantras. I shall not list them
all but the following assemblage lists the first 158 mantras (all
references are to the Rigveda by " c i r c l e , " " h y m n , " and verse):

10.30.12 (three times) 7.16.1-12

1.74.1 .-9 3.16.1-6
1.1.1-9 3.10.1-9
6.16.15-27 8.23.1-30
2.5.1-8 1.150.1-3
4.7.2-11 1.140.1-7
4.2.1-20 5.11.1-6
7.12.1-3 5.6.1-10

Committing the 360 mantras of the "Morning L i t a n y " to

memory in the right order is not the end, but the starting point of
what comprises ritual competence. For now the student has to
know the ritual 'application:' that is, he has to know when,
where and by whom the Litany is to be recited. H e has to know
what the other priests are doing at the same time, what other
acts, recitations or chants may have to be engaged in earlier, at
the same time or later, and who is responsible for all of these.
Moreover, all this knowledge often constitutes an element
treated as a complex unit by means of which larger ritual
structures are constructed at another time. The 'Morning Lit-
any' is a case in point. For toward the end of the Agnicayana
ritual, which was performed by Nambudiri brahmins in 1975 and
is the object of A G N I , another recitation takes place which
incorporates a modification of the Morning Litany. This is the
378 The Human Sciences

'Recitation for the Asvins' {âsvinasastra) performed also by the

Hotâ priest and long before dawn, but on the final day of the
ritual. It contains 1,000-verses instead of 360, and there is a close
relationship between the two recitations. This may be described
in approximate terms as follows: a specific number of mantras
are omitted from each group of metres of the Morning Litany,
and others are inserted in order to arrive at the number that is
required for the Asvin Recitation.
The description of this latter recitation in A G N I proceeds in
the customary artificial fashion of modern scholarship; it refers
to the verses by the numbers that Western scholars have
assigned to the various elements. But A G N I also specified the
verses that were omitted from the Morning Litany, because
these alone make the enterprise intelligible. Since the issue that
is relevant in the present context has been treated at some length
in A G N I I, I will quote from that book:

It might be asked why omitted mantras should be part of the descrip-

tion of the 1975 performance . . . Only the omission and insertion of
particular mantras can explain the extraordinary feat of memory that is
here on display. The hotä knows thoroughly the Rgveda Samhitâ, from
beginning to end. Throughout the recitation, he never hesitates when he
is within a hymn, or at the end of a hymn when he is about to recite the
next hymn. But when he is about to recite another hymn, or other verses
than the ones that traditionally follow, he pauses at his last breathing
pause, i.e., in the middle of the last verse. At that time, obviously, he
concentrates on what is to bé done next. Once he remembers it, he
continues with the next part of the verse, and continuous immediately,
without taking breath, with the other hymn or verses that are prescribed.
. . . The Recitation for the Asvins is similar to the Morning Litany. Since
he has learnt how to recite the Morning Litany by deviating from the
order of the Rgveda Samhitä as it is handed down, he has learned to
further deviate from the litany when he recites the Recitation for the
Asvins (AGNI 1:685-686).

The entire procedure that is described here exemplifies the

general structure of Vedic learning that has been outlined at the
outset of the same work. It relates several concepts we have
already met with to each other and places them in their proper
perspective:
 Oral Traditions 379

The oral transmission of the Vedas from father to son or from teacher
to pupil is known as adhyâya, 'learning' or 'recitation.' This is con-
trasted with prayoga, 'ritual application,' which refers to the general usé
of Vedic texts in ritual, or viniyoga, which refers to the recitation of a
particular mantra at a particular point in the ritual. Vedic ritual is
primarily characterized by the recitation, by one or more priests, of
Vedic passages. The structure and organization of this recited material
follows the requirements of the ritual. As a result, sentences and verses
are often taken out of their original context (which is preserved in the
adhyâya only) and adapted to new surroundings. A reciter who is
familiar with the prayoga has learned different arrangements of frag-
ments of the oral tradition that he has already memorized, and knows
where to insert them into ritual structures. This new dimension of
learning may be handed down orally too, and without any connection
with ritual activity. And so we meet with three kinds of knowledge,
handed down orally, each presupposing the former. Most reciters
preserve the Vedic texts in their original, or presumedly original, order.
Some among them have, in addition, learned how recitations have to be
modified and rearranged for use in the ritual. A few have preserved the
ritual practice itself, and know what, how, where, and when to act as
well as to recite (AGNI 1:31-32).

The sciences of ritual and grammar were born in the context of

these oral traditions and it is likely Jhat they were also created
without the help of writing. That the Rigveda itself was orally
composed—no one has ever doubted it. That the science of
ritual, which relies so heavily on memory, might have been
orally composed, can be imagined, though barely—depending
on who does the imagining. But that Pânini's grammar, 'one of
the greatest monuments of human intelligence' (Bloomfield
1933:11), could have been orally composed is an idea that has
never appealed or even made sense to Western scholars. The
only important exception was Max Müller—an exception not
easily brushed aside for he was one of the greatest pioneers of
precisely these studies that are at the heart of the Indian oral
tradition. It is well known that Max Müller was one of the great
nineteenth century polymaths, not only a Sankritist but also the
pioneer of the new 'Science of Religion (page 258).' But he was
also the first editor and translator of the Rigveda, which he
380 The Human Sciences

published in six volumes that appeared between 1849 and 1873,

and in which he included the Rkprätisäkhya.
Boehtlingk discussed the part that writing may have played in
the composition of Pânini's grammar in the Introduction to his
edition and translation. But Boehtlingk did not say: 'may have
played;' he said: 'must have played.' Boehtlingk was also a great
scholar; he did not only know a great deal, but possessed an
original and disciplined mind; moreover, he relied on argument
and not on hearsay or fashion. When such a person feels strongly
that something must be the case, but has no real argument to
support it, he relies on a device that serious scholars rarely use:
the exclamation mark. The exclamation mark occurs once in
Boehtlingk's Introduction; it is used on the one occasion when
no argument is given and when he refers to Max Müller's opinion
that the Brâhmana's and Sütras 'ohne Kenntnis der Schrift
verfasst worden seien!' ('were allegedly made without the
knowledge of writing!')
Sixteen years after Boehtlingk's edition, Caland demonstrated
that the Sütra of Baudhäyana was an oral composition (Caland
1903, reprint 1966:3). This has been universally accepted (e.g.,
Kashikar 1968a:43, Gonda 1977:514). Panini's grammar is also a
Sütra work; in fact, it has been called (by Renou), 'l'apogée du
genre.' Max Müller does not explicitly refer to Pänini when he
refers to 'the Sütras;' but he did, of course, include him. The
difficulty of his implicit assumption that the grammar was an oral
composition is that the sütras are interdependent in an extraor-
dinary complex manner—more so than the ritual sütras. One
change in a sütra in any of its chapters will necessitate numerous
changes in several other sütras in several other chapters. Even a
change in the order of two sütras—a simple inversion, for
example—would have far reaching implications and conse-
quences. A circumstance, incidentally, that exemplifies an im-
portant linguistic fact: the inversion of two rules in the deep
structure of a grammar may have the most dramatic effects on its
surface manifestations—just as an inversion of the amino-acids
within a gene may lead to diametrically opposed characteristics
in the phenotype.
 Oral Traditions 381

Let us consider a simple example of the effects of such an

inversion of rules and describe it in general and abstract terms.
In a system in which the rules are ordered in such a way that
each rule has to be applied before the next, the sequence of
rules:
' a -> b I
(1)
b -> c J
has the same effect as the single rule:
a - » c. (2)
But if the rules are interchanged, as in:

not (2) but something quite different results: there are now two
kinds of ' b ' : one that was already present but is now replaced by
' c ' ; and another that is ' n e w ' for it comes into being whenever
we start with ' a ' . The second 4 b' might as well be given another
name, say, ' d \ so that the sequence of rules (3) becomes:
b-*c
a - » d.
N o w , if we reverse the order again to see what difference it
m a k e s , we obtain something that is quite different from (1), viz.:

b —> c.
I have little doubt that a mind like Pânini's could solve these
kinds of problems that lesser minds would find difficult or
impossible to handle without pencil and paper, and that in our
letter-bound culture cannot even be imagined: he would do it, in
principle, by combining the uncommon analytical gifts he un-
doubtedly possessed with the extraordinary feats of memory
that were part of his culture, having strenghtening this powerful
combination further by exercising it through constant and regu-
382 The Human Sciences

lar practice. After all, Pänini was a trained grammarian who

must from his early days have been steeped in the knowledge of
the grammatical works of his preceptors and predecessors—
works of which at present and thanks to him only the names
survive.
Of course one cannot prove a theory that makes the most of
the elusive notion of genius; I can only try to argue that it is a
reasonable and promising hypothesis, and therefore quite feasi-
ble to pursue it. But if it would ultimately run into serious
problems or not find favor, I can still think of another explana-
tory scenario: Pänini worked in close collaboration with some
colleagues or, more likely, pupils. Let us assume, for example,
that he had more or less completed the rules of vowel sandhi,
and provisionally formulated these in a consistent manner and to
his satisfaction. N o w there appears a problem elsewhere in the
grammar; and the only way in which it can be given a simple
solution is by inverting two of the sandhi rules he had just
formulated. Immediately a host of problems arise, and the rule
system begins to generate ungrammatical forms. How to save it,
safely modify and keep track of it without losing the thread?
The solution is simple: Pänini asked his favorite pupil to
memorize the rules for vowel sandhi he had provisionally
formulated. H e turned his attention elsewhere, and returned to
effect the required inversion. The student who was given the
special assignment heard it, and knew precisely how to react to
it by reformulation. Other pupils who had memorized other
portions of the grammar were eagerly listening in order to find
out how any proposed modification would affect their domain;
and if trouble arose, they immediately took steps to overcome
the problem by changing the rules, their order, their formulation,
or whatever else had to be changed. This led to other revisions
elsewhere in the grammar, supervised and synthesized by Pänini
himself. There are many ad hoc devices for patching up rules
that must have been resorted to on such occasions and that can
in fact explain certain oddities that we meet with in the corners
of Pänini's grammar.
Is the idea of such team-work alien to Indian civilization which
 Oral Traditions 383

is depicted, after all, as a culture of solitary navel-gazers?

Not at all, for that picture is nothing but a caricature. The
Rkprätisäkhya was the first composition to describe the teaching
of Veda recitation to a group of students. A few decades ago,
when the Jaiminïya branch of the Sämaveda was on the verge of
extinction—for it had been transmitted in its full breadth only in
Kerala and only in twenty Nambudiri families (see Staal 1961:
86)—one of the foremost Sämavedins belonging to that school,
Itti Ravi Nambudiri, rose to the occasion and set out to
remember and finally write down the entire Jaiminïya tradition
with the help of his pupils. Docs this phrase indicate that his
pupils helped him with the transcription? N o , not at all; he would
not entrust that to anyone. What he did is chant the songs
together with the others, going over them again and again.
Sometimes one would stop, because he could not remember; at
other times another. Sometimes it was Itti Ravi himself who
seemed to have forgotten an uncommon chant even though it
was he who had originally taught it to the others—but that had,
in some cases, been several decades earlier. It should be noted
that we are dealing here with the ritual domain of the so-called
gäna songs that is much more specialized than the Sämaveda
Samhitâ; there is no such thing as a Padapätha, not to mention
modifications such as Krama or any of the others (cf. A G N I I:
276-278). In these rarified chants, a pupil might remember what
the master himself had forgotten; only if it was recognized by
several others including himself would Itti Ravi accept it as
authentic.
I believe that it would be profitable for Western psychologists
who are studying memory to learn Sanskrit. This would enable
them to go to India and study the mnemonic techniques and
practices of those increasingly rare traditional pandits that are in
popular parlance referred to as 'walking encyclopedias.' It
would be interesting to enquire into a phenomenon that I can
only explain by introducing a notion of 'collective memory 1 : I
am referring to a practice that is common among Vedic reciters
and chanters. Vedic brahmins always prefer to recite in pairs; for
two do not only know more than one; two that recite together
384 The Human Sciences

know more than the same two reciting separately. At first 1 did
not understand or like this practice; it does not make for clear
recordings—especially since the Vedic chanters do not seem to
favor the musical notion of singing in unison. However, I finally
understood that reciting together does not only increase the
confidence of the chanters; it also leads to the recovery of a
larger portion of the oral tradition than could ever be recovered
by single performers.
And so we return from ritual to grammar and attribute to
Pänini the masterminding of an art that is exclusively mental and
oral. It explains, among other things, the extraordinary Indian
insistence on the importance of the guru. (The importance of the
guru is deeply ingrained in the counterculture. However, since
the invention of writing and especially of printing, gurus are
much less relevant and rarely expound a doctrine that is not
already known from texts. Their importance is now confined to
the teaching of practices that cannot be easily learned from
books, such as meditation.)
The Indian art of scientific composition is ultimately explained
by the requirements of the ritual. For ritual requires precision,
accuracy and an extreme degree of formality. The form of the
mantras is all that counts and it is their form that had therefore
to be preserved. The emphasis on formality that characterizes
the science of ritual was equally important to the science of
language. The latter science was also inextricably linked to oral
transmission: for grammar exists raksärtham, 'for the sake of
preservation', as P ä n i n f s commentator Patanjali formulated it
unambiguously in the Introduction to his Mahäbhäsya or 'Great
C o m m e n t a r y / The insistence on formal accuracy, the exclusion
of meaning, and the extraordinary precautions that were taken to
preserve the Vedas; the concomitant sciences of ritual and
grammar—all of these were therefore rooted in ritual. In the final
resort, we have to extend this conclusion beyond the confines of
India: for Western philology and linguistics would not exist
without the Rigveda and Pänini, and these were only preserved
because of the ritual tradition. Thus came into being the two
sciences of ritual and language, one still exclusively Indian;
 Oral Traditions 385

paradigms of what Max Weber called 'formal rationality/ These

sciences, however, were not only Indian; they also were, from
beginning to end and throughout their development, oral.
The strength of oral tradition is extraordinary and in the case
of India the invention or introduction of writing would have
detracted from the substance of Indian culture in its formative
period. If writing had been known, two human sciences would
not have come into being at such an early period. Much of the
strength of oral tradition survives, especially in tribal culture
where it has least been studied and where the historical conti-
nuity often remains hidden. It is unlikely that we shall ever know
how long oral traditions have been maintained in Africa, Aus-
tralia, or the Americas. But in India and China, where the past
is known as well as the present—and sometimes better known—
solid results can be attained provided the data are studied
thoroughly and with an open mind as far as methodology is
concerned. Jyotirindra Jain (1984) has shown that the Rathvas, a
West Indian tribe, maintain in their ritual paintings a cult of
Indra that preserves Vedic features extinct in Hinduism since at
least a thousand years. The Rathvas probably adopted this cult
from the people from whom they learnt farming and agriculture.
The study of such living artistic traditions requires a combina-
tion of the methods of anthropology and of art history, a
synthesis like that between anthropology and the human sci-
ences without which neither ritual and mantras, nor Asia can be
understood (Chapters 14 and 25).
 28

Religions

28A. Seeking Religion in Asia

The study of religion ought to play an important part in the

human sciences, for while language provides the foundation for
most intellectual activity of the human animal, religion hovers
around the loftier realms of human expression and belongs to a
domain that lies beyond language. Religion consists in part of
oral traditions that tend to develop not merely into literary, but
into scriptural traditions. The oral traditions of religion are
generally of the loosely organized epic type that are particularly
suitable for the expression of changing interpretations. When
scriptural, these traditions become fixed, but not in the sense of
the tightly organized Vedic type of tradition that led, at least in
India, to formal analysis and the human sciences. Religious
traditions do not lead to science because interpretations, instead
of being questioned or changed, tend to harden into doctrine.
For the understanding of religion, our point of departure
remains the extended form of Durkheim's account of religion
(above page 126). Durkheim distinguished two categories of
religious phenomena, beliefs and rites, and assumed that the
latter depend on the former. The assumption is unpersuasive
because rites do not depend on beliefs: they lead a life of their
own, which is determined by rules, as we have seen. In many
cases, beliefs depend on rites because they are interpretations of
rites. Durkheim's account is also incomplete because it omits
mystical experience. Nevertheless, his account is a good point of
departure because of its classic succinctness. Durkheim has
logically pursued his preconceptions, and what Quine wrote
about Austin also applies to him:
388 The Human Sciences

Historians of science tell us that science forges ahead not by an

indiscriminate Baconian inductivism but by pursuing preconceptions,
even mistaken ones. I see in Austin's work this kind of progress (Quine
1981:91).

We extend and modify Durkheim's account by distinguishing

three categories: rites, mystical experiences and beliefs. There
are several dependencies between specific members of these
three classes, especially of beliefs on the two others, but
basically the three are independent. Extending Durkheim's
analogy we say accordingly that the "religious" categories of
rite, mystical experience and belief are interrelated like the
" o r d i n a r y " categories of action, perception and thought.
Among these three categories, rites are primary because they
are almost always independent and can be accounted for on their
own terms. They also possess a longer background of evolution-
ary development: ritualization is common among nonhuman
animal species. Both human and non-human animal rituals can
be described in the ethological terms of " F A P " ("Fixed Action
P a t t e r n " : Thorpe 1951) and " M A P " ("Modal Action P a t t e r n " :
Barlow 1977). Rites become "religious" when they are provided
with a religious interpretation. We have accordingly referred to
Asian rituals as rituals "without religion." This expression,
however, stands in need of an explanation because it is contra-
dictory if we follow Durkheim and regard ritual as a necessary
feature of religion.
Mystical experience is another category that goes beyond the
confines of what is generally regarded as religious. I have
discussed this at greater length in another book (Staal 1975) and
shall mention here only one feature of the argument given there.
Mystical experiences are generated not only by so-called reli-
gious practices such as ritual, recitation, and meditation, but
also by practices that are nonreligious, at least in their primary
sense, for example, fasting, breathing or drugs. Indian texts like
the Yogasütra or the Buddhist Abhidharmakosa enumerate such
causes without distinction, and sometimes add others: for ex-
ample, the grace of a deity that is conducive to visions and
incorporates mystical experiences into an existing tradition; or
 Religions 389

4
' b i r t h " to account for those we might call natural mystics,
people who fall into mystical experiences without training or
expectation.
Mystical phenomena possess physiological as well as psycho-
logical features and appear to be biologically determined. H o w
this can be explained will have to be determined by future
research. Some early beginnings were discussed in my 1975
book, but the field is still wide open, and several promising
avenues of research have not been touched or even imagined.
Whether nonhuman animal species experience mystical phe-
nomena, for example, is unknown. This is probable since drugs
affect many species of animals, sometimes in unexpected ways.
Such questions are not beyond the pale of scientific investiga-
tion: much information is available on the perception of color
among animals, to mention one example.
It is safe to conclude that the two categories of ritual and
mysticism are independent from the doctrinal component of
religion. This is in accordance with Geertz' statement that
"religious life" in Java includes "etiquette, art, and mystical
p r a c t i c e , " combined with the observation that it excludes the
dimension of doctrine or belief (page 327). It also fits the well
known fact that Indian religious traditions emphasize correct
practice more than correct doctrine, so that we need the concept
of " o r t h o p r a x y " in many contexts where, in the West, " o r -
t h o d o x y " would be required. A Hindu may be a theist, panthe-
ist, atheist, communist and believe whatever he likes, but what
makes him into a Hindu are the ritual practices he performs and
the rules to which he adheres, in short, what he does.
To understand "religion" in its fullest sense we should now
turn to the third category of religious phenomena. Doctrines and
beliefs are regarded as religious when they involve belief in a god
or gods, in paradise and hell, faith, salvation, and similar
religious concepts that are characteristic of the three monothe-
istic religions of the West. But now we meet with a difficulty. It
is gradually becoming more widely known that most of the other
" r e l i g i o n s " of mankind are deficient in one or another respect
when studied within this perspective: in Buddhism and Jainism
390 The Human Sciences

there is no belief in a god or gods; in Taoism immortality is not

located in a hereafter; in Yoga, Mïmâmsa and several other
Indian traditions, gods are accessories or otherwise subordinate;
in Confucianism none of these concepts or ideas exist in even
remotely similar forms. The main reason, however, that Asian
traditions do not fit the Western pattern of religion is that their
emphasis is not on doctrines or beliefs, but on ritual, mysticism,
or both. In so far as doctrines or beliefs are mentioned at all,
they are not primary but added: they are of the nature of
secondary interpretations, often rationalizations and generally
after-thoughts. Not infrequently the doctrinal dimension, alleg-
edly incorporated into a sacred book, has been created in
response to Western demands and expectations. This is in
accordance with the fact that terms for religion that refer to its
doctrinal content are relatively rare in the languages of Asia and
are invariably of recent date. Hajime Nakamura expanded on
this during a 1984 conference (Staal 1985b:54, note 24): in India,
the term dharma has been used in the sense of "religion" in
expressions like hindü dharma, bauddha dharma, jaina dharma
only during the last few centuries. The same holds for the
Chinese tsang-chiao and the Japanese shükyö ( <£ %K ). The
concept of " H i n d u i s m , " incidentally, came up in the thirties of
the nineteenth century in English literature on India (Heester-
man 1985:207, note 19).
Matters of doctrine or belief range over a wide spectrum.
Mythology, for example, resembles ritual in that it often devel-
ops and grows by incorporting arbitrary elements following an
"outer chance d e v e l o p m e n t " (pages 324, 344). Doctrine, on the
other hand, possesses a philosophic dimension that prevents
unlimited growth: consistency has to be preserved. To the belief
in a single god or in transmigation one cannot add the belief in
another two gods or in a day of judgment. Kautsa excluded
mantras from the category of meaningful statements for pre-
cisely this reason: for if they were, some of them would affirm
the existence of one Rudra, others of many, and this is contra-
dictory (page 375).
Amalgams of religious phenomena are also found in Western
 Religions 391

religion, but play a minor role because of the insistence on

doctrine. Major incorporation of ''foreign" material would lead
to conflict. Obvious examples of ritual elements that are not felt
to be inconsistent with existing tenets of belief are the incorpo-
ration of " C h r i s t m a s " trees or " E a s t e r " eggs in the Christian
calendar and in Christian practice. In Asia, such amalgamation is
found on a much larger scale. Before looking at Asian traditions
from a more systematic point of view, I shall discuss five
examples, some already met with in other contexts.

(1) The rudiments of Vedic ritual were brought to India by the

Indo-European intruders who had already added to their Indo-
European fire cult the Indo-Iranian ritual of Soma. To the
resulting mixture they added the art of building large altars from
kiln-fired bricks which was inherited from the Harappan or Indus
Civilization (Converse 1974; Staal 1978a; 1982:39-53; A G N I I:
125-166).

(2) The Vedic fire cult with many of its mantras spread all over
Asia together with Tantrism although its relationship with other
Tantric rites is tenuous. We find accordingly, even at present,
Vedic rites and mantras, in Tibet, Japan and Bali (Skorupski
1983; Strickmann 1983; Hooykaas 1983 a and b ; Payne 1986). In
Bali, the majority of these are classified as Hindu but some are
regarded as Buddhist; elsewhere they are looked upon as
Buddhist if the need for a label is felt at all.

(3) The Balinese mumukur ceremony takes place after the

cremation of a deceased's body and intends to separate the soul
from the physical remnants of the body. This rite is related to the
Javanese shadow play of wayang kulit, originally a summoning
of ancestors for magical purposes. It also corresponds to similar
ceremonies found in other parts of Indonesia and used to be
enacted after the entombment of the ancient Javanese kings.
Many large monuments on Java were originally built as dwelling
places for the king's soul. Subsequently they were interpreted
392 The Human Sciences

and constructed as " H i n d u - J a v a n e s e " temples although they

are neither Hindu, nor temples (Stutterheirn 1931).

(4) In Japan, the government separated Shinto and Buddhist

divinities during the Meiji Restoration of 1868. Grapard (1982)
has shown that prior to this separation, Shinto and Buddhist
elements had entirely fused. At many cultic centers, what had
begun as a Buddhist cult, became and remained syncretic
throughout the middle ages and became " S h i n t o " only in 1868.
At Tonomine, for example, where the ritual cycle had been
mostly Tendai Buddhist, food offerings began to be made to the
Shinto deity Kamatari starting from an auspicious day in 1465,
chosen by divination and transmitted to the Buddhist authorities
by an imperial messenger. Dances and sumo wrestling were also
offered, and Kamatari was regarded as an " a v a t a r a " of the
" P u r e - N a m e " Buddhist lay master Vimalakïrti. Here as else-
where, ritual, political and cultural features and elements were
inextricably merged.

(5) Taoism has incorporated Buddhist Tantric rites and concepts

that came from India although they may originally have been of
Chinese origin. Whatever the historical relationship (page 369),
"it is possible to find detailed parallels of much precision
between Taoism and T a n t r i s m , " although " t h e field has been so
uncultivated that much research will be r e q u i r e d " (Needham
1956:427-428). That there is not merely parallel growth but a
historical connexion is clear especially in those cases where the
similarities are based not upon factual relationships but upon
fantasy. The most telling example is a practice on both sides of
the Himalayas still advocated by contemporary gurus and swa-
mis. The latter associate this practice with the belief in the
possibility that " t h e entire seminal fluid accumulated in the
testicles starts flowing upwards toward the heart. Then it gets
purified by the gastric fire and moves on into the brain, strength-
ening all the sensory nerves and greatly enhancing a yogi's
powers of memory and intelligence" (Muktananda 1974:98;
Staal 1978b). The Taoist practice consists of exerting pressure
 Religions 393

" o n the urethra between the scrotum and the anus, thus divert-
ing the seminal secretion into the bladder, whence it would later
be voided with the excreted u r i n e " (Needham 1956:149).

The inapplicability of Western notions of religion to the tradi-

tions of Asia has not only led to piecemeal errors of labeling,
identification and classification, to conceptual confusion and to
some name-calling. It is also responsible for something more
extraordinary: the creation of so-called religions. This act was
primarily engaged in by outsiders and foreigners, but is some-
times subsequently accepted by members of a tradition. The
reasons lie in the nature of Western religion, which is pervaded
by the notion of exclusive truth and claims a monopoly on truth.
It is professed by " P e o p l e of the B o o k " in the apt' phrase the
Koran uses to refer to J e w s , Christians and Muslims. In most
parts of Asia, such religions do not exist, but scholars, laymen
and Western converts persist in searching for them. If they
cannot find them, they seize upon labels used for indigenous
categories, rent them from their original context and use them
for subsequent identification of what is now called a " r e l i g i o u s "
tradition. Thus there arises a host of religions: Vedic, Brahma-
nical, Hindu, Buddhist, Bon-po, Tantric, Taoist, Confucian,
Shinto, etc. In Asia, such groupings are not only uninteresting
and uninformative, but tinged with the unreal. What counts
instead are ancestors and teachers—hence lineages, traditions,
affiliations, cults, eligibility, and initiation—concepts with ritual
rather than truth-functional overtones.
Several baffling consequences follow from this stage of affairs.
Conversion, for example, is different in Asia from its Western
counterpart and a person may adhere to two or more "reli-
g i o n s , " Conversions, moreover, are not confined to the religious
domain; they are well-known in the comparative study of ritual.
They resort under that subdivision of rites that has been referred
to as "rites de p a s s a g e " or " l i m i n a l , " because they seem to
constitute, effect or accompany a transition from one state to
another. In the realm of Vedic ritual, such rites are called grhya
or " d o m e s t i c " rites. One of them is upanayanam, the ritual
394 The Human Sciences

through which a member of one of the three highest castes

becomes " t w i c e - b o r n . " Its successor in the history of Indian
ritual is the Buddhist upasampadä, which transfers a layman
{upäsaka) to the monastic life so that he becomes a monk
{bhiksu). These two rites are similar in many respects, and there
are further similarities with the dlksä or " c o n s e c r a t i o n " cere-
monies that exist in Vedic as well as Tantric forms (cf. Staal
1982b: 47, and A G N 1 , I: 317-333), and with initiation rites of the
Indo-European sodalities and secret brotherhoods that left many
traces in the monastic communities of Jainism and Buddhism (cf.
Bollée 1981). The prototype or paradigm of all these ceremonies
is Shamanistic initiation, which is labelled variously but remains
essentially the same in accordance with the maxim: rites are
invariant under religious transformation.
Though mass initiations are occasionally met with (a new
variant in Madras City is group upanayanam for poor brahman
boys), a characteristic feature of most of these rites is that they
pertain to a single individual. In Asia, the ritual character of
these " c o n v e r s i o n s " predominates, and this was often the case
in the West, as is shown, for example, by the similarities
between the Jewish circumcision ceremony and Christian bap-
tism in the early centuries C E . to which attention has been
drawn by J o n a t h a n ^ . Smith. These similarities also illustrate the
invariance of rites under religious transformation. In more
recent centuries, we still hear from time to time that practice is
more important than preaching. Said Dr. Johnson: " S u n d a y
should be different from other days; people may walk but not
throw stones at b i r d s . " However, the individual and subjective
experiences that accompany conversion rites have on the whole
been increasingly emphasized in the West. We find here a wide
range of possibilities. At one end of the spectrum there is Ian
Barbour's solemn declaration which is intelligible only within an
existentialist-Christian perspective: "Religious c o m m i t m e n t . . .
is a self-involving personal response, a serious decision impli-
cating o n e ' s own life, a willingness to act and suffer for what one
believes i n " (Barbour 1980: 239). At the other end of the
spectrum there is the flirtation with Rome that was fashionable in
 Religions 395

Oxford during the days of Oscar Wilde, who wrote about it in a

letter:

If I could hope that the Church would wake in me some earnestness and
purity I would go over as a luxury, if for no better reason. But I can
hardly hope it would, and to go over to Rome would be to sacrifice and
give up my two great gods "Money and Ambition" (quoted by Auden
1974: 305).

In the modern West, conversions such as these are considered

witty when they occur in literature, but Americans raise an
eyebrow when they are real. Not so in Asia, where conversion
has never been merely psychological or " s p i r i t u a l . " A good
example is that of the Thakali, a group of traders living in the
valley of the Kali Gandaki, one of the deepest gorges in the
world which runs roughly north-south between the mountain
massifs of Annapurna and Dhaulagiri in Nepal (cf. Jest, 1964/65).
This community was engaged in the trade between Tibet and
India via Kathmandu. One of their traditional transactions was
to purchase thousands of goats in Tibet, and sell them in the
central valley of Kathmandu where they were then sacrificed
during the festival of Dasain. Originally a variety of Tibetan
Buddhists with a strong admixture of Bon-po and local rites and
deities, the Thakali learned Nepali and adopted Hinduism when
they found that this earned them the respect of their trading
partners in Kathmandu. However, with the increase of the
power and prestige of the Buddhist kingdom of Mustang, which
is situated at the northern end of the gorge and protrudes into
Tibet, they reverted to Tibetan and Buddhism. When the trade
with Tibet stopped as the border was closed, Mustang lost its
prominence', they became Hindus again, and had saptäha,
"weekly r i t e s " , performed for them by brahmans. Presently,
with the ntew popularity of T i b e t a n ' a n d everything Buddhist
especially among tourists, mountaineers and other foreign visi-
tors, the Thakali are once more in the forefront of Tibetan
Buddhist culture.
Such repeated transitions are not easy. Said Churchill, speak-
ing from experience: " A n y o n e can rat once; but it takes a certain
396 The Human Sciences

ingenuity to rat t w i c e . " Nor should these transitions be brushed

aside as mere opportunism. They are no more opportunistic than
is the desire to live close to one's work. Gods, Buddhas, ritual
transitions and political parties are there to assist man in his
endeavors and occupation. Todd Lewis (personal communica-
tion) visited Thakali homes in Pokhara and saw their domestic
shrines which display Tibetan lamas as well as Hindu and other
local deities. This is not a rare form of syncretism; such
phenomena are common in all parts of Asia.
Adherence to more than one "religious" tradition is especially
common in China. I confine myself to a quotation from Michel
Strickman:

What then are China's religions? So far we have taken notice of

Buddhism, an import, and Taoism, a native creation. The oft-mentioned
"Confucianism" is, in my opinion, best seen as something else, not
strictly comparable—more a Paideia, a system of culture and its gradu-
ated inculcation, rather than a faith professed by a guild of ritual
technicians claiming otherworldly authority. The paradigm later styled
"Confucian" was simply a broad base of Sinic letters, a fundamental
view of cosmos and society which we might better term "common
Chinese," for it was als'o fully shared by Taoist priests and Chinese
Buddhist monks. Indeed, in mediaeval times, Taoist abbeys and Bud-
dhist monasteries often sheltered academics providing training in this
common Chinese literary culture, full assimilation of which, whatever
one's regional or even ethnic origins, certified a man as civilised:
respectably Chinese. The embarrassment of modern scholars who retain
the old Chinese rubric of the "Three Teachings" is apparent when they
have to deal with the cult of the dead ("ancestor worship"), and sacred
sciences such as divination and geomancy. It is customary to vacillate,
assigning these wayward phenomena now to one of the great Three, now
to another. But in fact they belong to none of them exclusively, and are
freely used, when needed, by all. This might also serve to describe the
attitude of most Chinese to Buddhist and Taoist institutions, and for the
mediaeval period, as well as the modern, a functional rather than
sectarian approach to the question of religious identity, might best serve
for the great majority of the population. (Strickmann 1982:55)

Though India has always been regarded as a land of religion,

the situation is very similar to what we find in China. Most of the
alleged differences between India and China are due to Western
 Religions 397

projections. India became fashionable during the Romantic

Period, and has therefore always been expected to respond to
religious yearnings. China, on the other hand, was a favorite of
the Enlightenment, and has therefore been expected to be
without religion. In the Indian cultural area, however, Hinduism
and Buddhism often overlap (Staal 1982b, with special reference
to the Himalayas). This was true in Southeast Asia, character-
izes much of Tantrism (where there is often a one-one-corre-
spondence between Hindu and Buddhist notions), and is still
widespread in Nepal or Bali (where we meet, for example, with
"Buddhist Brahmans": Hooykaas 1973). If "Vedism" is re-
garded as a separate "religion," we have to record that in many
parts of India, especially in the South, "Hindus" continue to
perform "Vedic" rites (AGNI, 11:199-251).
In India, the chief conceptual tangle is proffered by Hinduism
itself. For Hinduism does not merely fail to be a religion; it is not
even a meaningful unit of discourse. There is no way to abstract
a meaningful unitary notion of Hinduism from the Indian phe-
nomena, unless it is done by exclusion, following the well-worn
formula: a Hindu is an Indian who is not a Jaina, Buddhist, Parsi,
Christian, Muslim, Sikh, Animist, . . . (the list is indefinite). I
used to know a Hindu who considered himself a Seventh-Day
Adventist. When faced with such data, should we abandon the
concept of religion altogether? Basically, there are two possible
procedures. We can either start with a rather narrow concept of
"religion," based upon the three Western monotheisms, and see
to what extent such a concept can be used in Asia. Or else we
can try to formulate a wider and more flexible concept, and see
just where that leads us. In either attempt the result may be
positive, in which case the term "religion" may be fruitfully
employed; or negative, in which case it is best discarded.
I shall begin with the first procedure. To start with a definition
would be rash: definitions come in the end, when a domain of
experience has been subjected to systematic analysis and con-
ceptual issues have been fruitfully discussed (page 63). Not even
the area of Western religion is such a domain. However, even if
we do not seek to provide a precise definition, it is not entirely
398 The Human Sciences

unclear what would be involved in a concept of religion based

upon Judaism, Christianity, and Islam. It would involve such
notions as a belief in God, a holy book, and (at least in two cases
out of these three) a historic founder. Taking our cue from this
last exception (the fact that Judaism has no historic founder), we
can meaningfully ask whether it is feasible to apply to Asia a
concept of religion that requires the presence of at least two of
these three characteristic's. What we find, even with these
relatively flexible characteristics, is that none of the so-called
religions of Asia is a religion in this sense. Buddhism, for
example, has a founder, but neither a belief in God, nor a holy
book. Even Ruegg, who is inclined to consider Buddhism a
religion (though dharma, according to him, expresses a concept
that applies to more levels than fctreligion"), writes:

It is a well known fact that for Buddhism, both 'early' and Mahayanist,
gods, like all other beings, are subject to Karma and its ripening and they
accordingly also belong to one of the five (or six) forms of transient
existence in Buddhist cosmology. In Buddhism there is no supernatural
and supreme God who creates the world and beings and who presides
over or modifies Karma in favour of beings (Ruegg 1981:424).

As for holy books, the attempt to assign such a role to the

Dhammapada—like the similar attempt with regard to the
Bhagavad Gîta in " H i n d u i s m " — i s nothing but a response to
Western missionaries or an attempt to meet Western expecta-
tions. Taoism does not have a belief in God, but is alleged to
have a founder and a holy book; however, the latter is but a
small fragment of the Taoist Canon. Tantrism does not appear as
an independent movement. It is allied with Buddhism or
Saivism, and shares characteristics with thé Yoga which enters
into similar alliances. Shintoism lacks all three characteristics.
Confucianism possesses only one: it has a founder. And so our
conclusion can only be that any notion of religion that is based
upon characteristics; of the three Western monotheistic religions
is inapplicable in Asia.
Turning now to the second procedure we can try to introduce
a more general and flexible concept: the extended Durkheim
 Religions 399

account with which we started this chapter. Any definition based

upon its three categories should be sufficiently flexible to be
applicable to at least some of the Asian phenomena. To begin
with, almost all Asian phenomena that have been regarded as
religious are replete with ritual. It may be objected that Bud-
dhism does not pay much attention to ritual, for the Buddha
declared that the Vedic ritual of his brahman countemporaries
did not assist in the attainment of nirvana. But whatever the
opinion of the Buddha on this point (a topic to which I shall
revert), it is clear that ritual has always played an important part
in Buddhism. Buddhist monks have always spent a large amount
of their time performing rituals, whether for their own benefit, or
for the benefit of others. This can still be observed throughout
the Buddhist world, e.g., in Sri Lanka, Burma, the Tibetan areas
of India and Nepal, Bhutan and Japan. The category that is least
important in Buddhism is the category of belief. Buddhism is
" a n intensely practical religion, inclined to treat doctrinal defi-
nitions . . . with some degree of u n c o n c e r n " (Conze 1969:11).
Snellgrove (1959,1:40) declares similarly: k T o r it is not the
philosopher who gives life to a religion, but the man who
succeeds in practising it, and in India the practiser par excel-
lence has always been the yogin." And specifically about Tantric
Buddhism: ' T o any who conceive of Buddhism as just philos-
ophy, this preoccupation with ritual and techniques of yoga may
well appear a riot of degeneration, but to a Buddhist who
conceived of his religion primarily as an art of yoga and who had
never neglected the use of ritual, it might very reasonably have
appeared as the most effective teaching ever sponsored under
the name of B u d d h a . "
Though Western scholars and believers have paid much
attention to Buddhist doctrines, (and Buddhist philosophy does
not lag behind any other system of thought in the proliferation of
its concepts, ideas and theories), the Buddha himself either kept
silent or was not forthcoming when he was asked a purely
theoretical or doctrinal question. The parable of the man who
was hit by a poisonous arrow, and then asked who had shot it,
what was his caste and occupation, and what wood the arrow
400 The Human Sciences

was made of and from which plant the poison had been ex-
tracted, is there to illustrate the point: that man would die before
his questions could be answered. Such questions that deal with
irrelevant theoretical problems belong to the category of f c 'unan-
swerable q u e s t i o n s " or " i n e x p r e s s i b l e s " {avyäkrtavastüni).
Buddhist aspirants are dissuaded from taking them too seri-
ously.
Like the other so-called religions of Asia, Buddhism is char-
acterized by the fact that ritual (in which all the monks engage)
is more important than mystical experience (which only a few
attain), which is in turn more important than belief or doctrine (a
matter confined to philosophers, scholarly monks or reserved
for Western converts, anthropologists, and tourists). I shall not
try to determine in the present context whether meditation
constitutes a fourth 4 'fundamental c a t e g o r y , " or whether it is a
variant of the ritual category, or even a category that falls
somewhere in between the ritual and the mystical. Meditation, at
any rate, is not gazing upon nothing (except in the limiting case),
but is closely related to ritual and mantras.
H o w is it possible for ritual to occupy such an important place
in Buddhism when the Buddha himself was sceptical about its
efficacy? We can only arrive at a satisfactory answer to this
question if we abandon the view that ritual is symbolic, and
depends on doctrine or belief, and accept that ritual is some-
thing-to-be-done. We have studied several Asian illustrations of
the fact that rituals are transmitted without interpretation or with
constantly changing interpretations, and are engaged in for their
own sakes. In India, this is particularly clear from the Brähmana
literature (Chapter 13A). But even in the Upanisads, Jainism and
Buddhism, the situation remains similar. In H e e s t e r m a n ' s words
(1964:27 = 1985:42):

The question that occupies religious thought appears not to be: sacrifice
or rejection of sacrifice, but rather what is the true sacrifice. Equally the
question does not turn on brahmin superiority or its rejection, but on the
point who is the true brahmin. On these points both orthodox and
heterodox thinkers seem to agree.
 Religions 401

The contribution of the Upanisads is less often the replacement

of karman, "ritual activity," by jnäna, " i n s i g h t , " than the
insistence that karman should be performed with jnäna. The
implications of these facts confirm what we have observed
before. Rituals are not merely remarkably persistent within
so-called religious traditions, where they are provided with
constantly changing interpretations; rituals remain the same
even across so-called religious boundaries: they are invariant
under religious transformation. This is demonstrated by the fact
that the same rites occur in Vedic, Hindu and Buddhist forms,
not only in India but also in China, Japan, Tibet, and Indonesia.
This explains why Buddhist monks engage in rituals, for many of
their rituals, e.g., rites of oblation into a fire, are independent of
Buddhism. The same holds for the numerous initiatory, seasonal
and exorcistic rites that Buddhist monks engage in and that are
as independent of Buddhism as the Christmas tree is indepen-
dent of Christianity.
We arrive at the following conclusions. If we adopt the
" e x t e n d e d - D u r k h e i m " concept of religion, which incorporates
the categories of doctrine (belief), ritual, mystical experience
and meditation (the latter either as a fourth category or as a
subcategory of one or two of the others), we have a concept on
our hands that has all the characteristics of pathological, if not
monstrous growth, tumorous with category blunders. It is worse
than a spider with a submarine, a burning bush, an expectation,
and a human head. We have found that the trio of ritual,
meditation and mystical experience consists of categories that
are more fundamental than the category of religion itself. Only
doctrine or belief may be in a position to constitute a religious
category per se. We have thus arrived at a point where we
should clarify the situation by making a terminological decision.
The simplest decision would seem to involve a return to our first
procedure, make " d o c t r i n e " a defining characteristic of " r e -
ligion," and confine the term "religion" to Western monothe-
ism. This is what I shall do for the t i m e b e i n g . As we shall see,
this is not an inconsequential decision, and the notion of religion
will still turn out to have applications in Asia.
402 The Human Sciences

When we return to the empirical domain to see whether these

conceptual clarifications may be relevant, we find that many
other considerations have to be taken into account as well.
Buddhism is often regarded as something special even within the
Asian context, and in this respect closer to religion in the
Western sense, comparable—if not in rivalry with—Christianity.
It is clear that Buddhism and Christianity are in some respects
similar. Both have a single historical founder and are spread over
large parts of the world. However, we have to be careful not to
interpret Buddhism as a religion within a Christian perspective.
This has not only been done by Western missionaries, scholars
and converts, but also by Buddhists themselves. The challenges
of the Christian missions, combined with colonialism and the
alleged superiority of the West, have led to representations of
Buddhism by Asian Buddhists that have little to do with Bud-
dhism in any of its genuine manifestations. The only way in
which we can arrive at an adequate understanding of Buddhism
is by approaching it first in the Asian context from which it arose
and in which it obviously still fits. Once we do this we discover
that some of the similarities between Buddhism and Christianity
acquire a different meaning, or vanish.
I have already mentioned that the assignment of a single holy
book to Buddhism is nothing but a reaction to Christian and
Western expectations. The next point of similarity that is often
mentioned is the missionary activity that has led to the world-
wide spread of both traditions. These activities are similar to
some extent, but the differences are equally significant. The
spread of Buddhism from India to Central Asia, China and
beyond, is in many respects different from the Christian conver-
sion of Europe. The diffusion was not just of Buddhism, but
included the exportation of Indian philosophy, logic, science,
medicine, astronomy, grammar, and Sanskrit legends, lore and
literature. These events have been collectively referred to, not
without justification, as Indianization. The immense effort of
translation (from Sanskrit into Chinese, Korean, Japanese,
Tibetan, etc.) that these processes involved is comparable only
to a limited extent to the relatively simple effort of translating the
 Religions 403

Bible into several European languages (an activity that was

undertaken in the wake of Protestantism). These activities are
more on a par with the vast translation efforts from Greek into
Latin and other European languages that took place during and
especially after the Middle Ages. These European translations
were chiefly concerned with the non-religious domain, for reli-
gion had been excised from the Greek h e r i t a g e / T h e arts and
sciences that came to China together with Buddhism had also
nothing to do with religion: they were simply Indian.
We are now in a position to note another significant difference
between not only the Buddhist and the Christian expansions, but
also between the translations from the Greek and from the
Sanskrit: the European and Asian recipients were not equally
backward or barbarian. When Greek science reached northern
and Western Europe through the intermediary of the Arabs, it
filled what was almost a vacuum. The northwest European
barbarians had nothing that even resembled the civilization to
which they were beginning to be exposed. But when Indian
medicine, for example, reached China, it came face to face with
a traditional system of medicine that had been thriving for
centuries. That such meetings did not always lead to violent
clashes is due to partial similarities between the Indian and
Chinese systems that still stand in need of explanation. Summa-
rizing the difference, we should say that, in Europe, a religion
that had no links with civilization swept over an uncivilized area
long before civilization arrived from elsewhere. In Asia, on the
other hand, Buddhism came as part of a package that contained
mostly civilization and that met with a civilization in all respects
its equal. The contrast is between religious m^ss conversion on
the one hand and a meeting of civilizations on the other.
There is yet another significant fact that we cannot fail to note
when we try to interpret the Buddhist expansion in Asia in the
Asian perspective that is its natural home. When Buddhist
practices and ideas went from India to Southeast Asia, they
were not only connected with many non-religious ideas and
techniques, but they were also linked with the expansion of
Hinduism. In most of Southeast Asia, Hindu expansion pre-
404 The Human Sciences

ceded the advent of Mahâyâna Buddhism. The two merged to

some extent, as we have seen, and then, in countries such as
Burma and Thailand, were replaced by Sinhalese Theravâda
Buddhism after the twelfth century. This process can also be
referred to as Indianization, but it took on forms that are to some
extent different from the ones that characterize the Indianization
of Central and East Asia. As Coedès and others have shown, the
Indian idea that was most influential in Southeast Asia was the
concept of divine kinship. In Cambodia and Thailand, the King
came to be regarded as a God on earth—whether it was a Hindu
God or a living Buddha. The concept of Indianization cannot do
justice to the historical processes and transformations that
actually took place. In Cambodia, Burma and Fu-nan (on the
lower Mekong valley) there had long been ceremonies during
which a " L o r d of the Mountain-Top" was set up to represent the
unification of territories conquered by the king. The cult of the
divine king, therefore, was originally animistic (using that term
in deference to Coedès 1966: 143, and for lack of a better one),
then received Hindu and finally Buddhist labels—if one insists
on using labels in spite of the fact that they are less informative
and significant than the entities to which they are attached (cf.
also Stutterheim 1931).
The gradual importation of Buddhism from India into China is
especially instructive for being in many respects totally unlike a
religious conversion. I can do no more than refer to some of the
facts that Strickmann has recently collected and analyzed in his
study " I n d i a in the Chinese Looking-Glass" (Strickmann 1982)
from which I have already quoted. We should first of all take into
account what the Chinese were looking for. They were not
waiting for a "religion." They had long regarded the Western
regions as a paradise where the Mother Queen of the West
dwells " i n a splendid palace amid the peach orchards, ringed
with tigers and other heraldic b e a s t s . " They were always
fascinated by the exotic, the marvellous and the fantastic, and
anything that emerged from the Western regions was first of all
measured against those standards. If it was frightening, like most
of the demonic apparitions that plagued travellers on the silk
 Religions 405

route, it could perhaps be propitiated or avoided; in some cases,

it could simply be eaten. ''Ingestion (rather than expulsion) as a
means of dealing with otherwise troublous phantoms has always
recommended itself to the economical and food-obsessed
Chinese."
Another need was therapeutic ritual. " T h e r e is copious testi-
mony to Buddhist professional activity in the treatment of
d i s e a s e . " The treatments were herbal as well as ritual, and
extended to rites for the salvation of the dead, which developed
as a compromise between the monastic requirement that the
monks should leave their families, and the all-embracing grip of
the Chinese family system: these rites "satisfied not only the
family's past, but also guaranteed its beatific future. . . Chinese
Buddhist monks and Taoist priests quickly perceived and re-
sponded to this d e m a n d . " Another Chinese interest was caves,
and this could easily attach itself to the tradition of cavern-
sanctuary, already well attested in early Buddhist India. The
obsession with mountains-with-caves or pebbles-with-holes is
Indian as well as Chinese, and probably of Central Asian origin
(see A G N I , 1:139-166). The Chinese interest developed Taoist
as well as Buddhist forms: " B o t h Chinese Buddhist and Taoist
sources state that the moutains and their cavern-paradises will
serve as sacrosanct places of refuge during the world's final
apocalyptic convulsions."
Strickmann's essay is rich in information that is relevant in the
present context. I shall conclude with a final quotation that may
serve as a summary: " 4 I n Chinese, religious conversion is styled
" t r a n s f o r m a t i o n " (hua). What was the role of religion in this
great Babel of peoples? If the overlooming monoliths of race,
language, and state were heaved aside, should we in fact find
that religion offers a master key to understanding, a straightaway
through the maze of ethnic profusion? Such has not been the
view of historians of China writing in English. It is regularly
implied that religion, for the Chinese, was at best a distinctly
marginal phenomenon, at worst a form of low-class self-indul-
gence."
All these considerations on the non-religious nature of most
406 The Human Sciences

Asian traditions that are generally called religions, explain some

baffling facts of Asian history, for example, why Buddhism is
often undistinguishable from Hinduism, why Buddhism disap-
peared from India and was exported all over Asia where it fused
with different civilizations, and in general, why we find so many
outer chance " s y n c r e t i s m s " in Asia. Even Buddhism, which is
in some respects more similar to Christianity than any other
Asian tradition, is basically a ritual-mystical cult that pays little
attention to doctrine or belief. That it has made such a different
and misleading impression in the West is largely due to the fact
that Western seekers and students have confined themselves
mainly to the study of Buddhist philosophy, and have not heeded
the Buddha's advice that philosophic speculation does not lead
anywhere.

28B. Seeking Out the Buddha

If it is true that Buddhism pays little attention to doctrine, this

raises an important question: did Buddhism introduce a separate
doctrine? The answer is generally assumed to be in the affir-
mative as a matter of course; but the facts are far from obvious.
Unlike the early Christian missionaries in northern and western
Europe, who addressed peoples that can only be described as
savages, the Buddha did not preach in a vacuum. His teachings
were not only preceded by those of Jainism, but they formed
part of the general intellectual ferment that characterizes the
seventh and sixth century B . C . E . context in India. In that context
it should be simple to determine what doctrinal innovations the
Buddha offered. In fact, they are surprisingly difficult to find and
formulate.
If the Buddha preached that there is an endless cycle of
transmigration from which it is possible to escape—it had
already been done in the Upanisads and in Jainism. If he
preached that the true brahman is not he who is born in the
highest caste, but he who is fearless, controlled, free from sins,
etc.—the Upanisads had already stated that a brahman is only he
 Religions 407

who speaks the truth, or knows brahman (see Bhattacharya

1973; 88-89), and the Jaina Uttarädhyayana-sütra had declared:
He who is exempt from love, hatred and fear, and who shines forth like
burnished gold, purified in fire, him we call a brahman . . . (Jaini 1979:
75).

If the Buddha preached that the true sacrifice is not that in which
oblations are made into the fire—the Upanisads had already
declared that only fools engage in such sacrifice, while a much
earlier Vedic Brähmana regarded as the highest sacrifice, the
sacrifice to brahman which consists in study (Bhattacharya
1973:115-116). According to the Jaina Uttarädhyayana-sütra,
true sacrifice is "guarding one's purity by means of the restraints
of a Jaina m o n k " (Jaini 1979:76).
If the Buddha said anything against the Veda—the Upanisads
themselves had not merely done it, but it is nowhere formulated
in more radical terms than in the Brhad Äranyaka Upanisad
(4.3.21-22), which is one of the two earliest and most venerable
Upanisads: " H i s is a condition beyond all desire, free from evil,
without any fear. Like a man in the embrace of his beloved
knows of nothing outside, nothing inside, that person in the
embrace of the self which is wisdom {präjnena ätmanä) knows
of nothing outside, nothing inside. His is the condition of bliss
when all desires are fulfilled, where there is only the desire for
âtman, or no desire at all. There, the father is no father, the
mother is no mother, the worlds are no worlds, the gods are no
gods, the Vedas are no Vedas.' 1
If the Buddha denounced the sacrifice of animals—it had
already been done in the Upanisads, and, more emphatically, in
Jainism. The killing of sacrificial victims by ritualists was
ridiculed in the earliest Vedic texts. It occurs in a verse of the
Rigveda (10.82.7), which returns in the Taittirlya Samhitä of the
Black Yajurveda (4.6.2.2.e):
You will not find him who has created
some obstacle is in your way.
Enveloped in mists and stammering,
taking'life, the reciters wander (cf. AGNI, 1:547).
408 The Human Sciences

Heesterman has shown, in a series of articles (1962, 1964, 1983),

that the trend away from animal sacrifice, which culminated in
the Upanisads, Jainism and Buddhism, is found throughout
Vedic ritual itself: it is not a radical movement, but an unmista-
keable, gradual tendency to move away from bloodshed and
violence. The "interiorization" of ritual is part of this general
development.
It might be said that all these points are not doctrinal but
ideological: that is, replete with social, political and religious
implications. So let us revert to pure metaphysics. Did not the
Buddha propose a major new theory at least in this rarified
domain of speculation: the doctrine of non-self {anätmavädä)!
And did not this doctrine stand in absolute contrast to the
Upanisadic doctrine of self (ätmaväda) which was further de-
veloped in the Vedänta? Here our authorities seem to agree
unanimously. The picture has been drawn, with broad but
magisterial strokes, by T.R.V. Murti in The Central Philosophy
of Buddhism o f 1 9 5 5 .
Yet even here caution is necessary, and the situation is more
complex than it appears at first sight. It is not open to doubt that
many of the later Indian Buddhist philosophers advocated the
anätmaväda; they are quite explicit on this count. The
Nägärjuna who was the author of the Mahäprajnäpära-
mitäsästra, made accessible through its monumental translation
from the Chinese version by Etienne Lamotte, wrote for exam-
ple:

Look for the âtman on heaven or earth, inside or outside, in threefold

time or in the ten regions, but you will not find it anywhere. Only the
meetings of the twelve supports (viz., the six sense organs and their
objects) produces the six cognitions. The meeting of the three (organ,
pbject and cognition) is called contact. Contact produces sensation, idea,
attention, and the other mental dharmas. In our system, ignorance leads
to the belief in the personality. Because ofthat belief, people believe that
the âtman exists. But that belief is destroyed by the vision of the truth
regarding suffering and the relative knowledge with respect to suffering.
When belief in the personality is destroyed, the ätman is no longer seen.
(Lamotte 1949:747; 1976:2005).
 Religions 409

What did the Buddha himself have to say on this topic? He

indicated that the five skandhas, and more generally, all
dharmas, are anätmä (see Lamotte IV, 1976:2005; Bhattacharya
1973, 12, passim). But in the Upanisads, the corresponding five
kosas are equally ephemeral: they do not survive after death and
they do not transmigrate. In the Upanisads, the âtman is
different from the kosas, and is real. In Buddhism, the skandhas
are anätmä, which implies, in Sanskrit, that they are different
from the âtman. But that ätman itself, is it real or unreal?
The few early passages in which the Buddha is quoted as
expressing himself on these issues have been discussed by
scholars for over a century, and both interpretations have been
defended. Those who have adduced reasons for a specific
interpretation comprise some of the greatest Orientalists of the
past, the very same scholars to whom we owe in fact almost all
the basic information we now possess on the Upanisads and on
early Buddhism. H o w then are we to decide?
The obvious answer to such a question is that we should go
back to the texts, and try to read and understand them in the
light of all the information that is presently available. However,
we should critically assess the sources of our information and
the channels through which it has been obtained. These require-
ments are obvious, but taking them into account in the area we
are discussing leads to several preliminary observations. First of
all, we will discover an astonishing schism for which there exists
no intrinsic or scientific justification: the schism between " I n -
dologists" and " B u d d h o l o g i s t s " . It might seem that this division
corresponds to the schism between Hindus and Buddhists, but
this begs the question we are asking, and " I n d o l o g y " would
need to encompass a great deal more than " H i n d u i s m " in any
case. It might be argued that the schism between Indologists and
Buddhoiogists exists for good reason because of necessary
specialization, and ensuing differences in background and edu-
cation between different specialists: a scholar who deals with
Vedic and Hindu texts needs primarily Sanskrit (though a
smattering of other Indo-Iranian or Indo-Aryan languages is
helpful), whereas a scholar who deals with Buddhist texts needs
410 The Human Sciences

in addition to Sanskrit not only Pali, but Chinese (for primary

sources, primarily), and Japanese (for primary as well as sec-
ondary sources), and if possible Korean, Tibetan, Mongolian
and a handful of other languages. Scholars of the latter type are
extremely rare, obviously. Even if we include Japanese scholars
who possess advantages of language and culture, they can still
be counted on the fingers of one hand.
The schism between Indologists and Buddhologists, however,
is not only based upon the presumed need for specialization; it is
also based upon at least one unproven assumption. That assump-
tion is that Buddhism is a religion, and that there is therefore a
certain unity to the subject. We have seen in the preceding
section that the concept of religion is not easily applicable to
Buddhism or to any other Asian tradition. That unity is therefore
imposed from the outside and a priori. The unities presumed to
cover early and late Buddhism, or Indian, Chinese, and other
forms of Buddhism, are functions of the same unproven assump-
tion. Of course, one is bound to see unity when one is convinced
that it must exist. But such an approach will not lead to adequate
evaluations. It is not open to doubt that Chinese Buddhism has
been influenced by Indian Buddhism. But in particular cases,
that influence might be small and insignificant in comparison to
other influences. For scholarship to be adequate, it should not be
based upon such assumptions. Only if we abandon them are we
in a position to discover whether, and to what extent, such a
unity may in fact exist.
Another circumstance complicates the issues we are dealing
with. When Buddhism became first known in the Western world,
it was the Pali sources that were almost exclusively relied upon.
The picture that emerged was that of a simple religion, replete
with ethical notions and similar to Protestantism. The Mahäyäna
insofar as it was referred to at all was regarded as a deteriora-
tion: at best a fresh development in the Far East, at worst a
decadent manifestation influenced by Tantrism and other de-
based Hindu notions of later date. As it happens, the same
specialization that I mentioned has led to a reversal of this
former picture, which has been beneficial because it represents
 Religions 411

a step forward in our knowledge and understanding: it has been

found that the Mahäyäna is much older and that the Chinese
translations from the Sanskrit provide reliable information on
much earlier stages of Indian Buddhism than had been assumed
at first, not only in the domain of the Mahäyäna, but also in that
of the Theraväda. Because of these sources we now have a more
reliable picture of the history of Indian Buddhism than could
ever have been reached if we had confined ourselves to Indian
sources.
Another qualification has yet to be taken into account when
we evaluate these sources and their significance: they do not
take us back to the time of the Buddha. During the early period
of what has been called "original B u d d h i s m , " the order was
hardly organized and there were few monasteries. ' T h e life of
Buddhist ascetics in its incipient stage was fairly different from
the monastic life of monks in later days. It was quite close to the
life of hermits as is mentioned in great epics, such as the
Mahäbhärata and the Rämäyana" (Nakamura 1980:58). The
schism that gave rise to the Mahäsamghika sect took place
during the period of Asoka in the third century B . C . E . The points
of difference that were discussed at that time were largely minor
matters of monastic discipline. In addition, five theses defended
by these first " h e r e t i c s " were directed against the traditional
Arhats: they were declared to be subject to certain impurities,
mental as well as physical (e.g., nocturnal emissions). These
heretical views were supported by laymen (upâsaka) who,
according to the traditional doctrine had no access to nirvana
(Lamotte 1958:312-319). More important doctrinal differences
that characterize some of the Mahäyäna theories, such as
emptiness (sünyatä) are of much later date, and were referred to
as pratirüpakaäharma ("imitation-doctrine": Lamotte 1973:39-
40; cf. N a k a m u r a 1964b).
Even with these new developments on the doctrinal level,
there was no split in the monastic community: monks with
different views continued to live together in the same monasta-
ries and were subject to the same discipline—if we except
minutiae. When Hiuen-tsiang visited India around 630 O . E . , . he
412 The Human Sciences

found several monasteries, especially in north-west India, where

the monks devoted themselves to different schools at the same
time. Lamotte comments memorably on this and on the later
splits within the Mahäyäna: " I n India, tolerance and accommo-
dation are more important than differences of doctrine; for belief
has only relative value. In the West, this would be inconceiv-
able; among us, each difference of opinion leads to a serious
dissension. For a Buddhist, salvation does not come from faith,
but from renouncing the world and bringing peace to the spirit"
(Lamotte 1973:12-13). Lamotte speaks here with double author-
ity: until his recent death, he was the most distinguished member
of that tiny band of scholars who can be counted on the fingers
of one hand; and he was at the same time a Catholic priest.
With all these considerations and qualifications we are now
sufficiently armed to discuss a simple example. At the second
sermon in Banaras, as referred to in the Anattalakkhana-Sutta
and other sources (see Bhattacharya 1978:12 sq.), the Buddha is
reported to have declared with reference to the aggregate of the
five skandhas'. ' T h i s is not mine, I am not this, this is not
m y s e l f (netam mama, ne so 'ham asmi, na m'eso atta). Even
a person without Pali can see how the thought in the final clause
is expressed: this aggregate (eso) is not (na) my (m') self (atta).
The first thing that strikes us is that the term anattâ is not used.
The second fact we should bear in mind is that attälätman is
used in Sanskrit and Pali as the reflexive pronoun which func-
tions in the same manner it does in other languages such as
English. "Self," in other words, is nothing but a reification of
"self." A person who knows nothing about these doctrinal
controversies would therefore translate: "this is not myself." A
person less innocent is tempted to translate: "this is n o t . m y
self," and a translator who wishes to press a point: " t h i s is not
my Self."
There are several similar passages in the earliest sources. A
scholar familiar with later Buddhism will not see any problem:
he will take such statements to mean that the skandhas are not
real, with the implicit understanding that the ätman is also not
real. A scholar familiar with the Upanisads will not see any
 Religions 413

problem either: he will take such statements to mean that the

skandhas are not real, and naturally different from the ätman
which is real. I do not pretend to offer a solution to this difficulty
in the context of the present chapter. My purpose is different: 1
want to show that there is a difficulty. Many contemporary
scholars have simply disregarded it: " B u d d h o l o g i s t s " assume
that the anätma doctrine is indissolubly connected with the
Buddhist religion; " I n d o l o g i s t s " are afraid to touch Buddhism,
and so have said nothing.
In the last generation, two great Orientalists conducted a
controversy on this issue: Hermann Oldenberg defended the
view that the Buddha accepted the ätman, and Louis de la
Vallée-Poussin defended the opposite view. Oldenberg was one
of the last Indologists who wrote authoritatively on Sanskrit,
Vedic, Hindu, as well as on Buddhist matters, whereas de la
Vallée Poussin was one of the first Buddhologists, albeit avant la
lettre, who eschewed non-Buddhist topics.
In the present generation there are at least five scholars who
are familiar with early Buddhist as well as Upanisadic literature,
and have expressed themselves on the issue: Bhattacharya,
Frauwallner, L a m o t t e , Murti and N a k a m u r a (in alphabetical
order). I have already referred to Murti: his book dealt chiefly
with the Mädhyamaka, but he also argued that the original
Buddhist doctrine was anätmaväda. N a k a m u r a defends the
opposite view, though with qualification: " I n early Buddhism,
they taught atvoidance of a wrong comprehension of non-ätman
as a step to the real ätman . . . Therefore early Buddhists never
maintained the non-existence of ätman. They merely opposed
the substantial permanence of a n y o n e ' s ätman. As for the
metaphysical question whether an absolute ätman exists or not,
early Buddhists kept silence" (Nakamura 1964a:90-91). Naka-
mura devoted a separate study to the problem of self in Buddhist
philosophy (Nakamura 1977), which he summarized as follows:
" B u d d h i s m did not deny the self as such, contrary to the general
assumption by many scholars who tend to regard the theory of
Non-Self as a sort of nihilism' v (Nakamura 1980:63-64). Frau-
wallner agrees with Nakamura, but is more explicit: " T h e
414 The Human Sciences

Buddha never ceases to emphasize that none of the five heaps

(skandha), which make up the mundane personality, should be
taken for the I. It was far from him, however, to deny the
existence of the s o u l " (Frauwallner 1956:63).
Bhattacharya is the most explicit among contemporary schol-
ars: his book h1 Ät man-brahman dans le bouddhisme ancien of
1973 is largely devoted to the issues we are discussing, and I
have taken many quotations from this work. Bhattacharya
defends the view that the Buddha accepted the ätman.
What, finally, did Lamotte have to say on the issue? He is
careful, as one would expect, if not over-cautious. After briefly
referring to Frauwallner, Oldenberg, de la Vallée Poussin and
Bhattacharya, he emphasizes that we have no explicit statement
attributed to the Buddha in which he affirms the existence of the
ätman. We should therefore, says Lamotte, confine ourselves to
the golden rule laid down by the Buddha himself: " W h a t I have
not declared, take that to be non-declared, and what I have
declared, take that to be d e c l a r e d . " (Lamotte 1976, IV:2004-
2005).
Earlier (Staal 1985b:44), I concluded from this passage that
Lamotte did not explicitly affirm that the Buddha taught the
non-self. In a footnote, I criticized de Jong's review of Lamotte,
1976, for rejoicing that " M . Lamotte categorically rejects all
attempts at discovering in Buddhism the belief in a Vedântic
ä t m a n " (de Jong 1978:168). In his reply, de Jong (1987:151) drew
attention to other statements by Lamotte showing that according
to the Buddha there is no ätman. Lamotte must accordingly'be
classified with Murti. But the main question remains unresolved:
why was the Buddha so ambiguous and unclear with respect to
what we are inclined to regard as a central issue that the leading
specialists arrive at opposite interpretations of his statements? A
substantial part of the answer must lie in the Buddha's lack of
interest in doctrinal matters, a lack of interest he explicitly
affirmed in the parable of the arrow and elsewhere. This is
important in the present context because it shows that even in
Buddhism, the Asian tradition that is in many respects most
 Religions 415

religion-like, doctrine plays a subordinate role and mystical

experience and rites are basic.
We must conclude that the concept of religion is not a
coherent concept and therefore misleading. It does not hang
together like a concept should and should either be abandoned or
confined to Western traditions. This conclusion is to be drawn
with caution and is not without qualifications. For example,
ritual is more important in Judaism than in Christianity, and in
Catholicism than in Protestantism; and the reverse holds, ac-
cordingly, for doctrine. Much of the emphasis on doctrine in the
study of non-Western traditions is, in fact, connected with the
rise of Protestantism (cf. Nyberg 1953).
A phenomenon more like religion in the Western sense
appears in the later phases of development of several Asian
traditions. Saivism and Vaisnavism, for example, though often
referred to as " H i n d u s e c t s , " may therefore be called "reli-
g i o n s " (as is done in Gonda 1954 and elsewhere). They should
perhaps be regarded as the first two indigenous religions of
India. They are closely related and not entirely distinct, but that
holds for the three Western monotheistic religions, too.
Whereas Saivism remained orthodox in its insistence on jnäna
and orthoprax like all preceding Indian traditions, including
Buddhism, Vaisnavism introduced a new concept epitomized by
the term bhakti or " d e v o t i o n . " The sources can be traced to a
more distant past: the late Svetäsvatara Upanisad (Oidenberg
1915:238; Lauenstein 1943:86-89). With bhakti, a more purely
" r e l i g i o u s " element in the Western sense is introduced for the
first time, and dogmas and religious squabbles come to the fore.
A Hindu is no longer an Indian concerned about what he must do
while thinking anything he likes. A Hindu becomes a believer in
God equipped with faith and a holy book. The development of
the Bodhisattva ideal in Buddhism is connected with this notion
oï bhakti. Thus Mahäyäna also came to incorporate religion—or
perhaps we should say two religions, one in India and one in the
Far East. That Shinran's " P u r e Land B u d d h i s m " was influ-
enced by the Indian bhakti cults and not, for example, by
Christianity, has long been known (Otto 1930:7-9). The tradi-
416 The Human Sciences

tions of " B u d d h i s t " temples in Japan also became indistinguish-

able from the cults of various divinities in the so-called " S h i n t o "
shrines (page 392). Buddhists have finally turned into believers,
not in God but in the Buddha. But such developments are no
longer inspired by the original outlook of the founder, which was
the hallmark of classical Buddhism:

Adhesion to the Buddhist faith does not require that the adept rejects his
ancestral beliefs or denounces the religious practices that are current in
his milieu. Through one of those compromises of which India has given
so many examples, everyone is free to worship apart from the 'Three
Treasures," the deities of his region, caste or choice and perform the
appropriate rites. Throughout history one meets with famous Buddhists
who continue their worship of spirits. Nâga and Suparna, Yaksa,
Vajrapäni, Women and Fairies. Householders, benefactors of the
Sangha, remain loyal to the gods of their class: Kuvera, spirit of riches;
Hâritî, the fairy with the children; the tutelary Couple, etc. The higher
castes continue to demand for the great Vedic and Brahmanic deities:
Indra, Brahma, Mära, etc. The advent of Buddhism has never produced
a "twilight of the gods." Säkyamuni does not oppose the pagan gods of
Hinduism. He recognizes that "the deities, honored and worshipped by
man, will in turn honor and worship him." He does not condemn pagan
rites: he disapproves of sacrifice in which living beings are killed; he
recommends peaceful offerings devoid of cruelty; certain practices based
upon pure superstition, ritual baths, etc. are practically without value.
What is important is to put everything in its place: gifts to pious monks
are better than the cults of the gods; taking refuge in the Three Treasures
is better than giving gifts; the highest achievement of sacrifice is to enter
the order (Lamotte 1958:74-75).

28C. Mythology

Students of religion have increasingly had an inkling of the fact

that "religion" in the Western sense, with its emphasis on
doctrine, is not a single, coherent concept. This insight has led to
a gradually increasing attention to non-doctrinal phenomena
such as mysticism and ritual, to " p o p u l a r " religion and "little
traditions." But anthropologists have failed to render the much
needed assistance in unravelling the resulting conceptual tangles
and confusions. Instead of being critical of the Western concept
of religion with its concomitant emphasis on doctrine or belief,
 Religions 417

they have continued to search for features that fit into Durk-
heim's category of " t h o u g h t . " This has led to the preoccupation
with myths and symbols that still characterizes cultural anthro-
pology and the study of religion. The process has been facilitated
by the incorporation of ideas from (various schools of) psycho-
analysis (cf. page 153).
One reason for the interest in mythology is that " o n e can
prove almost anything of the nature of myths if one selects only
a part of the data and refuses to look at the c o n t e x t " (Ingalls
1960:196). Scientific method is then replaced by a panacea: take
myths, psychology, add a little of anything you like, and you can
get whatever you want. I have paid little attention to mythology
partly because of this elasticity, and partly because myths are
already overdone.
In India, mythology developed from the Brähmanas just as
science developed from the Srauta Sütras. The relation between
Brâhmanas and Srauta Sütras—long disputed—was correctly
estimated by Caland, tested by Tsuji (1952) and has been
expressed by Gonda ( 1977:497) in terms that are no longer
controversial:

The authors of the Brâhmanas endeavoured to explain the origin,

meaning and raison d'être of the ritual acts etc., and to prove their
validity; the compliers of the Sütras, on the contrary, aimed at a
systematic description of every ritual in its natural sequence.

I have tried to show, in addition, that the Srauta Sütras embody

a science, and that the Brâhmanas offer interpretations that are
generally arbitrary and ad hoc. Their authors undoutedly knew
the ritual; but they did not know " w h a t it m e a n t . "
From some scholarly writings ( L é v i - S t r a u s s \ for example) it
would seem that the concept of religion has already been quietly
buried. Its critical assessment in the light of all available data,
and possible excision from scientific usage remains a task for the
future. Departments of religion, programmes of study involving
religion in a historical, comparative or pheriomenological per-
spective, and many related institutions and efforts remain rele-
vant to the study and understanding of Western traditions, i t is
418 The Human Sciences

unlikely that they are very helpful in the study of non-Western

peoples and civilizations. The work that has been done under
such headings has to be re-evaluated or redone after the con-
cepts have been scrutinized and purified, renewed or replaced.
Mythology within or without religion is still a promising subject,
but it goes beyond our present investigation and is treated
elsewhere (Staal 1988b).

28D. Philosophy of Religion

It would stand to reason to expect that conceptual analysis is

the primary task of a discipline traditionally called "philosophy
of religion," taught in most English speaking countries in
departments of religion or philosophy. In this field of studies,
however, the current approach is generally outmoded. The
philosophy of religion is not only exclusively taught in terms of
doctrines and beliefs, but almost entirely confined to what
Western philosophers and theologians have written about Chris-
tian doctrines. Thus attention is paid to the existence of god and
alleged proofs of that existence, to good and evil, reason and
faith or belief, and so forth. Anselm's ontological proof and its
discussion by Western philosophers, culminating in K a n t ' s
discovery that existence is not a predicate, predominate in these
courses. That K a n t ' s discovery does not even make sense in
Chinese was established by A. C. Graham (1959, 1965, 1967).
The real objection to the manner this discipline is taught is that
it has little or nothing to do with the so-called religious traditions
of mankind. A philosophy of religion worthy of its name should
begin with a discussion of the concept of religion and an
investigation into the status of a possible science of religion
based upon what is presently known about the religions or
so-called religions of mankind. Ritual, mantras and mysticism
should occupy at least as important a place in such an investi-
gation as matters of doctrine or belief. Should it turn out that
concepts of Western philosophy are not appropriate for fruitfully
dealing with such topics, the subject is all the more interesting
and worthy of discussion. The philosophy of religion is impor-
 Religions 419

tant for that reason and also because the imposition of the
Western concept of religion on the rest of the world illustrates
how Western imperialism continues to thrive in the realms of
thought.
A philosophy of religion based upon Indian philosophy would
be equally parochial but more adequate because Indian philos-
ophers were familiar with the science of ritual. This holds
especially for the Mïmâmsâ which is at the same time the most
Indian and least studied of the six classical systems. Out of 159
metarules of the twenty-fourth section of Äpastamba Srauta
Sütra quoted in Chapter 26, nearly a hundred correspond closely
to sütras of the Mïmâmsâ Sütra (Garge 1952:54). The värttikas of
grammar, which are often metarules too, are also closely related
to these sütras (Paranjpe 1922:72), which is not surprising in
view of the fact that the Mïmâmsâ is often regarded as an
analysis of sentences (väkya). Since they are either based upon
it or react against it, many Indian philosophies, including Bud-
dhist systems, the philosophies of language and especially the
Advaita Vedänta of Sankara cannot be adequately understood or
evaluated without knowledge of the Mïmâmsâ. Within such a
philosophic perspective, a philosophy of religion could attain
substantial results.
 29

T h e M y t h of the T w o Cultures

In 1867, William James wrote from Germany to his sister

about a person he had met at a dinner party at the house of
H e r m a n n Grimm:

A soft, fat man with black hair (somewhat like the Renan of the
photograph) of uncertain age between twenty-five and forty, with very
small green eyes, he wore the obligatory frockcoat with an exceedingly
grimy shirt and collar and a rusty old rag of a cravat. The professor
overflowed with information about everything knowable and unknow-
able. He is the first man I ever met of a class of men to whom learning
has become as natural as breathing. He talked and laughed incessantly
at the table and gave Mrs. Grimm the whole story of Buddhism, and I
don't know what other bits of the history of religion. After dinner Grimm
and the professor got involved in a heated controversy about the
primitive form of natural religion. I noticed that the professor's answers
became somewhat tired and then his massive head suddenly fell forward.
Grimm called out that he'd better have a proper sleep in his chair. He
eagerly consented. Grimm gave him a clean handkerchief which he threw
over his face and appeared to go to sleep instantly. After ten minutes
Grimm woke him with a cup of coffee. He rose, like a giant refreshed,
and continued to argue with Grimm about the identity of Homer (quoted
from Rickman 1979:26-27 after W. Nohl; the italicised part of the letter
was written in English, the rest in German).

The person James described was Wilhelm Dilthey. If a single

person had to be held accountable for historicism, the re-
invention of hermeneutics, and the introduction of the distinc-
tion between sciences and humanities into contemporary West-
ern culture, it should be him. The terminology of the " t w o
c u l t u r e s " was later introduced by the British author and physi-
cist C. P. Snow. Dilthey's distinction between Naturwissen-
schaften ("sciences of n a t u r e " or "natural sciences") and
422 The Human Sciences

Geisteswissenschaften (literally, "sciences of the spirit") was

accepted without question by most continental philosophers,
from Heidegger to Foucault, including all existentialists and
almost all phenomenologists with the exception of their founder,
Edmund Husserl. Though Anglo-American philosophers have
been less sanguine about embracing it explicitly, the distinction
is now a fact of academic life on both sides not only of the
Channel but of the Atlantic, and is generally felt to be not merely
due to an unavoidable division of labor (as in the "National
Science F o u n d a t i o n " and the "National Endowment for the
H u m a n i t i e s " ) , but to deep and intrinsic differences of organiza-
tion, method and nature. The myth of the two cultures is now so
deeply ingrained that it might be called the Prejudice of Twen-
tieth Century Research.
That it is a prejudice and not based upon facts or a serious
demonstration becomes clear as soon as we look at the matter
more closely. Dilthey was unfamiliar with the sciences and
attributed to them the caricature of "scientific m e t h o d " I have
discussed in Chapter 2. He introduced a special faculty, Verste-
hen, " u n d e r s t a n d i n g " or " e m p a t h y , " characterized vaguely but
abundantly illustrated and exemplified, always in such a manner
as if it were exclusively appropriate to the study of man.
I have already touched upon this subject in Chapter 3 and
cannot, in the present context, give a full account of this notion
which has been very influential not only in the humanities but
also in the social sciences (it was adopted, for example, by Max
Weber). I shall quote a few phrases from the sympathetic
account given by H. P. Rickman in The Encyclopedia of
Philosophy (1967,11:405-406):

Das Verstehen is a technical term with a definite meaning that must be

clearly distinguished from its general use as a synonym for any kind of
comprehension. It is the comprehension of some mental content—an
idea, an intention, or a feeling—manifested in empirically given expres-
sions such as words or gestures.
In his later writings (when kkHow is meaningful experience possible?"
became his central question: page 404), Dilthey placed increasing em-
phasis on two additional conditions.
 The Myth of the Two Cultures 423

The first of these two conditions for understanding expressions is

knowledge of the particular concrete context in which they occur. A
word is better understood, sometimes only understood, in its verbal
setting; an action, in the situation that gave rise to it. From this condition
Dilthey derived the methodological principle that to understand an
expression we must systematically explore the context in which it
stands. For example, to understand a religious movement or philosophic
doctrine better, we must relate it to the climate of opinion and the social
conditions of the time. For example, the philosophy of Spinoza can be
better understood against the background of the rise of science and the
conflict between different religious sects in the sixteenth and seventeenth
centuries.
The second of these conditions is knowledge of the social and cultural
systems that determine the nature of most expressions. To understand a
sentence we must know the language; to understand a chess move, the
rules of the game.

I shall not attempt to subject these expressions to a detailed

analysis; they are second hand descriptions from a general work
of reference. I have used them because Dilthey is never clear on
central issues ("Dilthey himself never gave a systematic treat-
ment of these ' i d e a s / but only what may be termed preludes for
such a treatment, with the topics of life, life context, expression,
interpretation, and understanding as the leading t h e m e s " : Spie-
gelberg 1965,1:122). It is -not clear, for example, what the
difference is between the two "additional conditions," and
many of the explanations are so obvious as to be trivial. The
" m e t h o d " of Verstehen, moreover, is not different from the
point of departure of any scientific account, whether in the
" n a t u r a l " sciences or in the humanities. Out of context, facts,
data and phenomena cannot be understood (Vedic ritual, for
example, was placed in such a context in Chapter 7). But this is
only a point of departure, a first step, which corresponds to the
" p h e n o m e n o l o g y " of physics (pages 12-13): science becomes
interesting when phenomenological description is followed by
abstraction from the concrete context and when a general
hypothesis is formulated. That this is rarely done in the human-
ities is a matter of fact; but that it cannot be done has never been
demonstrated.
424 The Human Sciences

In the social sciences, more attention has been paid to

4
'theory''• than in the humanities. It is common, however, for
social scientists to insist that ritual is " e m b e d d e d " in society. It
is, but interesting results are obtained when we abstract from
this context and observe that the same ritual forms and struc-
tures are associated with different social events and are embed-
ded in different societies. The same holds for the embedding of
ritual in religion. By insisting upon context and prohibiting
abstraction, the nature of ritual and mantras has remained
inaccessible.
Dilthey's distinction between the sciences and the humanities,
therefore, reduces to a chimera based upon the caricature of
scientific method discussed in Chapter 2: what he describes as
characteristic of the sciences of man is also found in the sciences
of nature; only, the latter go beyond it in all interesting cases.
The same holds for most discoveries in the human sciences: this
applies to Jakobson, Chomsky or Lévi-Strauss—to mention a
few names of authors whose work we have discussed.
In some of Dilthey's work, and in other publications (specif-
ically in Heidegger who repeatedly refers to him in the context of
hermeneutics and historicity), the uniqueness of the humanities
is related to the unicity of man, which is, in turn, related to a
tradition of discussions on "subjectivity" in modern philosophy.
The argument runs, briefly, as follows. " M a n " is so different
from stars, rocks, molecules and even other animals that he
cannot be studied by the same methods by which these other
things are studied. He is, after all, not a thing: he is the unique
studying subject himself. This conviction explains, among other
things, Heidegger's effort to introduce " e x i s t e n t i a l s " into the
study of " D a s e i n " (page 26).
The argument is not simple and I shall not attempt fully to
unravel it in the present context. But we must recognize, to
begin with, that every star is also unique. The sun, " o u r " star,
is particularly important to us, and so close, relatively speaking,
that we can learn more from it than from any other star. But we
learn more from it not only about it; we learn more from it about
other stars, too. This leads to hypotheses that are general or
 The Myth of the Two Cultures 425

universal and that can be related to other data. The unicity of

man, or any other entity, is similar. The only apparent examples
of entities that do not have a unique individuality are certain
subatomic particles which seem to be interchangeable and
undistinguishable in principle.
It is true that we ourselves are not only objects, animals, and
beings, but also subjects. Being subjects means that we are in a
position to evolve a general theory or hypothesis about subjec-
tivity—a theory that aims at "objectivity" like anything else that
aims at being true. Dilthey did not accept this, and was bound to
accept relativism. Heidegger, following in his wake, rejected it
explicitly when he asserted that every truth depends on Dasein
(page 28). But whatever they say, most statements made by
philosophers, scholars and scientists aim at truth. Dilthey or
Heidegger do not wish to be interpreted as claiming that what
they say is false. They may deal with unique entities but they try
to say something about them that is not unique but general,
universal and true.
Dilthey is noted for his work in history. Historians seem
especially concerned with what is unique. There was only one
Hitler. What historians provide, in fact, are illustrations of the
problems to which Dilthey drew attention. I shall briefly discuss
two illustrations from Huizinga's Herftsttij der Middeleeuwen
("The Autumn Tide of the Middle A g e s " ) of 1919. The first
example concerns context. As his title indicates, Huizinga
adopted a particular perspective by placing the period of his
investigation (the fourteenth and fifteenth century in France and
the Netherlands) in the context of an age of completion rather
than a new beginning. But this is not different from the astron-
omer who chooses to describe a star as having evolved from a
gaseous cloud rather than as developing into a red giant, or a
biologist who describes the development of an animal species as
the end of an evolutionary movement rather than the beginning
of a new species.
Norbert Elias has criticized the limited time perspective of
sociology by comparing sociology and cosmology in a similar
spirit:
426 The Human Sciences

For sociologists, processes and structures of the nineteenth or eighteenth

century often appear as events of little significance for what happens
today. What happened one thousand or two thousand years ago may
appear to many contemporary sociologists not worth speaking of,
something of no interest whatsoever for sociologists. Cosmologists, on
the other hand, may well be equally interested in what happens today
and what happened millions of years ago. They see clearly the connec-
tion between past and present. They are able to perceive such events 0
together as aspects of a unitary process. Their time perspective, in other
words, is dictated not by their personal involvement but by the facts
themselves, whose connection with each other they try to unravel and to
represent by means of testable theoretical models, mostly of process
models. If the time perspective of sociologists too were fact oriented,
they too would have the task of bringing to light the processes connecting
past events with each other and with present times (Elias 1987:xvi-xvii).

That history cannot be objective is just as unpersuasive a

proposition as that it deals with unique things. Chapter XI of
Huizinga's book describes the churchyard of the Innocents at
Paris, with its skulls and bones in open charnel-houses along the
cloisters that surround the burial ground on three sides. Hui-
zinga uses this and other descriptions to evoke the medieval
image of death. The description is as objectively true as any
other scientific hypothesis: it represents the facts as best the
author knew them, and may have to be reviewed, modified or
rejected in the light of later findings. The same holds for the
hypothesis about the image of death that the author presents.
In the present context it is relevant to recall that Huizinga was
also the author of Homo ludens (1938), a theoretical and
speculative work that attempts to determine the "play element
of c u l t u r e " {het spelelement der cultuur): In American univer-
sities, this is prescribed as a textbook not in philosophy or
history but in departments of physical education because it deals
with games and sports and emphasizes the body, an entity often
neglected by theoreticians (Sartre included it in Being and
Nothing, an ingenious Cartesian adaptation of Heidegger's Be-
ing and Time, which had omitted it). The body is as important in
ritual as it is in most of the games that Huizinga deals with and
that illustrate a greater variety of 4 'language g a m e s " than were
 The Myth of the Two Cultures All

discussed by Wittgenstein, who introduced this notion. In Witt-

genstein, "language g a m e s " remain so vague that they can be
identified with the most diverse entities. In ritual, the mantras of
Kashmir Saivism were interpreted by Harvey Alper as " S i v a ' s
Language-game" (Alper 1988:249-294). The importance of Hui-
zinga's Homo ludens is that he characterized play precisely and
enumerated a number of its features, including the circumstance
that it is generally engaged in for its own sake. He included ritual
in this characterization.
If even a cursory inspection of a historical text suggests that
Dilthey's approach is not applicable to his own discipline, it
should not surprise us that the allegedly fundamental distinction
between the sciences and humanities is not supported by any of
the illustrations or theoretical considerations offered in its sup-
port. Dilthey was a fascinating scholar, but even Husserl, the
founder of phenomenology, "considered him chiefly as a man of
genius for intuition, but not of rigorous science and t h e o r y "
(Spiegelberg 1965,1:123 referring to Husserl 1952:173). Dilthey
was the son of a clergyman and his biography includes a case of
puritanical idealism that his sympathetic exponent Rickman
(1979:28-30) regards not only as " a little repugnant t o d a y " but,
even for a contemporary of Queen Victoria, as " a m y s t e r y . "
Dilthey's ideas are religious in the sense of the Western notion
(Chapter 28) as was established in 1929 by Fritz Heinemann.
Heinemann (1929:178-208) describes the "religious coloring"
of the approach to history found throughout the Romantic
period, in Schleiermacher, Hegel, Carlyle and Dilthey. He
shows how the freedom which Dilthey attributes to Verstehen,
by which one raises above the limitations of place and time
through a creative identification with other eras and personali-
ties, is described in terms of a religious experience. Dilthey
commented on a variety of religious values:

"Germany is a nation which, since the end of the middle ages, has
responded to each strong intellectual impulse with a change in its
religious outlook" {Deutschland ist eine Nation, die seit Ausgang des
Mittelalters auf jeden starken intellektuellen Impuls mit einer Verän-
derung des religiösen Weltansicht antwortete: in Heinemann, page 187).
428 The Human Sciences
kW
My vocation is to grasp the innermost religious life in history and to
express it in dynamic terms at the present time when we are moved
exclusively by the state and by science'1 {Mein Beruf ist, das Innersie
des religiösen Lebens in der Historie zu erfassen und zur bewegenden
Darstellung zu bringen in unseren von Staat und Wissenschaft aus-
schliesslich bewegten Zeiten: in Heinemann, page 191).

Heinemann, who praises Dilthey's contribution to philosophy as

the first understanding of life and history that is free from
"abstract thinking," characterizes it at the same time as " h i s -
torical p a n t h e i s m . " Such " p a n t h e i s m " is different from Chris-
tian theism but conforms to the Western notion of religion which
stresses doctrinal content. Whatever the value of this vision, it
does not provide a solid foundation for a distinction upon which
not only much contemporary research is based, but also the
modern understanding of man and his place in the universe—at
least in the Western world.
The study of ritual and mantras undertaken in the present
book and the conclusions and corollaries derived from this study
do not support nor are they supported by the idea that the
humanities and the sciences are fundamentally different from
each other. We have seen that other areas of the humanities have
also been studied without depending on the assumption of the
two cultures, notably in the science of language: this applies to
the work of Jakobson—the scholar who has " d o n e most to
revitalize the human s c i e n c e s " (Frank 1984:29)—of Chomsky
and his followers, and includes much of Lévi-Strauss' anthro-
pology. The distinction between sciences and humanities is not
clearly appropriate to the study of history, and clearly inappro-
priate in areas of natural science such as biology. This is obvious
from the work of Konrad Lorenz and the development of
ethology which ensued, along with attempts at "socio-biology"
and the work of scholars like Jacques Ruffié in France or
Stephen Jay Gould in the United States. It would be surprising
if methods of study that work in science and in the areas of
language, ritual, philosophy, music, literature, anthropology and
history could not be profitably used in all areas of the humani-
 The Myth of the Two Cultures 429

ties. Our conclusion must be that the human sciences are simply
a group of sciences.
Dilthey was right that history is a necessary component of the
study of man. The importance of history explains, but does not
justify the enduring popularity of historicism or more recent
relativisms such as 4 'post-structuralism" or " p o s t - m o d e r n i s m . "
It is clear that theories that emphasize an exclusively synchronie
approach are incomplete unless they are supplemented by
diachronic theories which include hypotheses on origins. In
linguistics, this has been accepted again after long hesitation: the
strictures of the Linguistic Society of Paris, which prohibited at
the beginning of the century the study of the origin of language,
are no longer in force (see, for example. Parts II and III of the
Proceedings of the conference on "Origins and Evolution of
Language and S p e e c h " : Steklis, ed., 1976).
Although neglected by anthropologists, history, which in-
cludes the study of origins, is relevant to the study of ritual
(Chapter 14). Burkert, originally a classicist, is perspicuous in
this respect. Despite the fact that he is hesitant when stating
what he does:

I once wrote that one should forbid the use of the term "origin," and I
try, sometimes without success, to avoid the word in my own usage.
What Î am trying to find is what I would call formative antecedents,
which of course implies the notion that the earlier is a formative power
for what is latter . . . (Burkert 1987:212).

he is very interesting when speculating about origins, provided

we look at what he actually does (see, e.g., Burkert 1987:152).
T o adopt, like Burkert, a biological approach, is to adopt an
evolutionary, that is, Darwinian perspective, whether developed
along orthodox lines or in a more revolutionary vein. For a
theory about the origin of language, for example, Eldredge and
G o u l d ' s 1972 proposals for " p u n c t u a t e d equilibria" are more
promising than the traditional "phyletic gradualism." For Don-
ald Davidson, speaking at the conference just referred to
(1976:18-19), is right in saying that a language is only a language
430 The Human Sciences

when it is similar to the most highly developed languages (the

only ones we happen to know).
Jonathan Smith's distaste for origins is much more marked
than Burkert's although he, too, has advanced speculations
about origins, adding that they have to be regarded as ajé-w
d'esprit (Smith 1987:206). His stated, theoretical position is
uncompromising:

These considerations underscore the position that a theory of sacrifice

cannot be found in a quest for origins, but can only be found through the
detailed examination of elaborations (1987:195).

Later, after discussing interesting hypotheses about things that

might have happened "before w h i c h , " he writes:

but there comes a point where the nature of the sources simply will not
let you go any farther. In other words, my nervousness on the subject of
"originating," "origins," "original" has to do in part with a stance. In -
another sense, it has very much to do with the nature of the reports that
I, at least, am dependent on—reports designed precisely not to ask the
question that's been asked, namely, the question about history, about
origin (1987:208).

In 1978 I published an article in which I speculated about

features of the Agnicayana that must be regarded as pre-Vedic
(Staal 1978). Kashikar argued that it is impossible and unneces-
sary to engage in such speculations (Kashikar 1979). He objected
to hypotheses that go beyond the data ancl that are not absolutely
certain. My reply (Staal 1982a:44-47) referred to scientific
speculation about the origins of life, the earth, the planetary
system and the-universe. Such speculations are among the most
interesting contributions of contemporary science and are, of
course, not "absolutely c e r t a i n . " They go beyond the observed
data and resort under the domain of the unobservable. In
paleontology, a class of .microscopic beings, called archibacte-
riae, have been postulated to have existed for the last three and
a half billion years, although no one has seen them. Astrophys-
icists theorize about what happened during the first few seconds
 The Myth of the Two Cultures 431

of the " h i s t o r y " of the universe. Why should humanists be

prohibited from doing the same within their own disciplines?
Some speculations are, probably, no more than wild dreams.
They have value only if they are treated as hypotheses amenable
to rational discussion. Since they are liable to being improved,
they are not final. Final truth is sought in religion but does not
exist in science. This does not imply that one should avoid
unobservables. Kashikar or Jonathan Smith simply deprive
themselves and their readers of hypotheses that are of interest
because they contain possible explanations. We shall not make
progress unless we supplement synchronie studies with diachro-
nic studies that do not eschew origins. In all these respects the
human sciences are like the exact sciences and there is no
foundation for the myth of the two cultures,
Dilthey's historicism is a forerunner of K u h n ' s Structure of
Scientific Revolutions (1970) and both lead easily to relativism
although Kuhn himself has not accepted relativist interpreta-
tions of his work. To followers of anti-scientific ideologies,
science is as arbitrary as faith and a scientific hypothesis is no
better than a myth. The distinction is, that hypotheses, unlike
myths, are refuted or rejected by considerations of logic or fact.
The difference is illustrated by Stephen Toulmin's The Return to
Cosmology (1982), which argues, following Kant, that hypothe-
ses concerning " t h e universe as a w h o l e " cannot be scientific
and must therefore be myths. Toulmin supports this argument
with a critique of two French biologists, Jacques Monod and
François Jacob, who had used their discoveries in molecular
biochemistry to bolster the already well entrenched central tenet
of Darwinism, that variations arise by chance and selection is
based upon the demands of the environment.
Toulmin diminishes the importance of chance with which,
says he (page 145), Monod has made "excessive p l a y . " He
criticizes M o n o d ' s and J a c o b ' s "Biological Structuralism" for
depending on "mechanistic Cartesianism." Rhetorical force is
added by attributing a "peculiar F r e n c h n e s s " ("not meant
frivolously": page 15) not only to Monod and Jacob, but also to
the mystic dreams of Teilhard de Chardin. Not being a biologist,
432 The Human Sciences

I may be wrong in failing to detect any difference between this

"Biological Structuralism" and Darwinism apart from the fact
that the former is further developed. But one need not be a
biologist to be able to perceive that hypotheses about anything
"in its e n t i r e t y " (Kant's phrase) are scientific provided they
depend on data and argument, are liable to rational discussion,
and can be refuted by considerations of fact and logic. Toulmin
has not shown that there is a point where the theories of science
must end and make room for mythology.
We have seen that the myth of the two cultures crossed the
Channel and the Atlantic. Let us pray that it does not settle on
the Pacific Rim.
 30

Conclusions

30A. The Origin of Language

The previous chapter has thrown doubt on the doctrine that

there is a fundamental difference between science and the
humanities. Developments in linguistics, biology and other sci-
ences, and the investigations on ritual and mantras in Parts II
and III combine to show that the general scientific method of
gathering data and formulating hypotheses to account for these
data can be fruitfully adopted in all these kinds of research. The
final blow to the myth of the two cultures is delivered by the
existence of Indian sciences of language and ritual (Chapter 26),
which originated from an oral tradition (Chapter 27) without
depending on any form of natural science. These Indian sciences
show that the distinction of the two cultures is not only un-
founded but lacks universality: for science is not confined to the
natural sciences as they developed in the West. This conclusion
is supported negatively by the religious nature of Verstehen and
of the privileged status claimed for it: for the underlying notion
of religion is not universal but Western (Chapter 28). All these
conclusions support the idea that the human sciences are scien-
tific in nature and universal in scope.
We are now in a position to provide the provisional definition
of ritual that could not be given at the outset (page 61). Ritual
may be defined, in approximate terms, as a system of acts and
sounds, related to each other in accordance with rules without
reference to meaning. The qualification implies that neither
ritual nor mantras constitute a " l a n g u a g e . " Ritual rules are
sometimes the same, and sometimes different from the rules of
language discovered in linguistics. Two features characterize
434 The Human Sciences

them: (1) they are recursive and can generate infinitely many
structures; and (2) they correlate sounds and acts in such a
manner as to approximate a one-one-correspondence, even
though we find in actual fact numerous deviations from this
principle. This second feature implies that Parts II and III have
dealt with different sides of the same coin: both " r i t u a l " and
" m a n t r a s " fall under " r i t u a l " in a wider sense. Since ritual is
not a system of symbolic representations that refer t o something
else, it cannot be explained in terms of religion or society.
Though embedded in various contexts and applied to a variety of
events, it adheres to its own rules, follows its own course and
leads its own life.
The independence of ritual from religion and society has been
perceived before. The late Naoshiro Tsuji, for example, who
was not only J a p a n ' s leading Sanskrit scholar but also a special-
ist of Vedic ritual in the tradition of Hillebrandt and Caland,
regarded the traditional intepretations of ritual as irrelevent to its
understanding. Familiar with the full breadth of Indology and
Sanskrit literature, he once remarked that nothing reminded him
more of Vedic ritual than the Japanese tea ceremony (Minoru
H a r a , personal communication).
To what extent is our working definition of ritual applicable to
other data? To answer this question would require a large under-
taking with ramifications leading us in different directions. In
some contexts, rites and mantras must be kept distinct; in others,
the concept of ritual will include both. The definition needs to be
tested in non-Asian civilizations and among non-human animals.
I have generally assumed that animal ritualization is akin to the
human, following the results of ethology (Lorenz 1963 = 1966,
1965, 1967, Eibl-Eibesfeldt 1971, 1975, Hinde 1982 with the
literature cited; cf. also Chapter 23) and the research of such
scholars as Burkert (1979, especially Chapter II), Fehling (1974),
and so forth. The notions of " F A P " ("Fixed Action P a t t e r n " :
Thorpe 1951) and ("Modal Action P a t t e r n " : Barlow 1977) are
helpful in understanding ritual structure among many animal
species. I shall illustrate the relevance of ethology for the study
of human ritual with one example from Vedic ritual.
 Conclusions 435

According to Heesterman (1962, 1983), Vedic ritual must be

explained against a background of battle, the agonistic elements
of which are "identified a w a y " by abstract liturgical elements.
Other explanations of ritual from violence have been given
elsewhere, specifically by Burkert (1972) and Girard (1972).
They are the subject of a recent publication entitled Violent
Origins: Ritual Killing and Cultural Formation (Hamerton-
Kelly, ed., 1987).
Ritual as a replacement for aggression is not uncommon
among nonhuman animals. It is found in tournaments, a term
used by Konrad Lorenz and other ethologists to refer to ritual-
ized battles. Here is an example:

During the breeding season the male marine iguanas become territorial.
They defend a small area of lava rock against other males, whilst females
are allowed to stay. If a male rival approaches the territory, its owner
displays. He opens his mouth, nods with the head and walks stiffleggedly
up and down in front of the rival, showing his lateral aspect. The dorsal
crest is erected and the gular regions extended. If the rival answers by
the same display, fighting is initiated. The opponents rush at each other.
But in spite of the biting intentions shown during the display they never
bite each other, but instead lower their head and butt. The hornlike
scales on the roof of the head interlock and the animals try to push the
other away. This can continue for a while, with pauses in between,
during which the opponents display frontally. The struggle ends when
one of the rivals is pushed from the rock or when he gives up by
assuming a submissive posture (lying flat on his belly). The winner then
stops fighting and waits in threat display for the rival to leave. The fight
is a highly ritualized tournament in the course of which the stronger
wins, without hurting the loser.

Ritualized fighting is fairly widespread in animals that are capable of

inflicting serious damage to the conspecific. Poisonous snakes never bite
each other, but the rivals wrestle according to fixed rules. Cichlid fishes
have developed various forms of mouthfighting and thus avoid mutual
damage. In a number of species, e. g. in wolves, fighting starts as a
damaging fight, but ends by a submissive posture of the loser, which
inhibits further aggression (Eibl-Eibesfeldt 1966:475-476; referring to
Lorenz 1963).
436 The Human Sciences

At this point a problem arises: if ritual as defined, or defined

along similar lines, occurs among nonhuman animals, the sci-
ence of ritual is not a component of the human sciences, but a
link between those sciences and other biological sciences. The
solution to this problem depends on the extent to which human
and other animal rites are different. My present guess is that
human ritual differs from animal ritual in the greater degree of its
involvement with mantras. For this there is empirical support
and it would explain that humans evolved language and other
animals did not. It is too early to propose a solution to this
problem which would depend on many details about which little
is known; but it is worth pointing out that any solution would
have implications for the concept of culture. For if culture is a
system of meanings and symbols (the ''Chicago d o c t r i n e " in the
words of Edmund Leach 1985:156), ritual is not part of culture
(as it is not part of religion), and a wide gap subsists between
humans and other animals (this is in accordance with Western
religious speculation following Genesis: see Staal 1979c). The
alternative is to incorporate ritual in culture along with language,
but regard only language as exclusively human. Available data
support this latter option. The definitions of ritual and culture
affect the problem of the origin of language, whether it coincides
with the origin of the human animal or whether it constitutes
merely an important event in its evolution, as seems more likely.
Ritual data from human and other animal species suggest that
both prehuman animals and humans who had not yet evolved
language, performed rituals basically of the nature we have
studied. I assume that complex ritual systems evolved in which
there were several types of sound and act and numerous
deviations from the principle of one-one-correspondence. This
labyrinth of correspondences became more prominent when
sounds that correspond to acts were uttered even at times when
the acts themselves were not performed. We need assume no
more than that our ancestors sang, hummed and voiced sounds
while imagining or thinking of rites. These latter faculties are
available to many animals: they imagine (danger, for example) or
think (of food or sex). This stage survives or is re-enacted in the
 Conclusions 437

otherwise unexplicable phenomenon of mantra recitation and

meditation outside ritual context. Language originated when
sound-meaning correlations were selected out of the abundance
of sound-act correspondences that exist in animal ritual.
The phonological pattern extracted from Jakobson (page 275):
om, mama, papa illustrates this development from mantras to
language. Sounds did more than accompany rites; they devel-
oped into a language that refers to rites. This survives in the idea
of one-one-correspondence between mantras and rites that is
otherwise difficult to explain. Acts, already distinct, were now
classified by sounds. This system of reference generated mean-
ings that were stand-ins for acts and other entities. Since the
correlation between rites and mantras constituted a complex
system, meanings developed by further extension, accretion and
systematization until language was born with the full complexity
of its features—a complexity that cannot be explained by
assuming that it gradually developed as a rational system for the
sake of reference, expression, or communication. Born by
chance, language was selected and retained, despite its cumber-
some complexity, because of its uncanny power to relate a great
variety of sounds to a great variety of meanings. Like elsewhere
in biology, traces of its origin remain. For although language
correlates sounds and meanings, large chunks of arbitrary ritual
structures survive in it which play no part in this correlation.
They explain the syntax of human languages and other features
of language that are similarly intricate, nontrivial, nonfunctional
and determined by highly restrictive principles.
If this sketch correctly characterizes the development from
ritual via mantras to language, it provides an immediate expla-
nation for the hypothesis that ritual and mantras are not lan-
guages: they are not languages because they originated prior to
language in the scheme of evolution. This hypothesis has other
implications of which I shall mention three.

(1) It is incorrect to assume that mantras are meaningless

because their meaning was forgotten. It would be extraordinary
for any intelligent creature to continue to engage in meaningless
438 The Human Sciences

activities and remember the extremely complex structures con-

structed out of mantras, whilst forgetting the simple meanings
presumably assigned to them in their original context. The
untenability of this assumption is supported by an analysis of
one of the more sophisticated hypotheses about such forgetting.
Paul Thieme (1957) distinguished between the explicit descrip-
tions of the epic and the allusive nature of the Rigveda. Accord-
ing to him, meaning was forgotten because the verses of the
Rigveda were, by their very nature, allusive and obscure (page
374). Thieme also proposed that Vedic ritual originated from a
feast: it is a "stylized feast" (stiliertes Gastmahl). It is not clear
whether there is any connection between these two theories.
Like other reductionist theories (Smith 1987:196 mentions
three types: sacrifice, New Year and initiation), the " f e a s t "
hypothesis does not explain the data. Thieme attributes the
belief in the magical power of mantras to the nature of brahman,
a problem he had dealt with elsewhere. He explains the survival
of mantras and acts without their original meaning, which has
been " f o r g o t t e n , " by the fact that " t h e element of the irrational,
unintelligible, even contradictory plays, after all, in magic a
necessary r o l e " (Das Element des Irrationalen, Unverständ-
lichen, ja Widersinnigen spielt ja in der Magie eine notwendige
Rolle: Thieme 1957: 91). But this " n e c e s s a r y " role cannot serve
as an explanatory hypothesis. It merely states in different terms
( " m a g i c " ) what we are trying to explain.
Puzzling about ritual is not the belief in "magical p o w e r " held
only by a minority of those who continue to engage in its
performance. Its mystery lies buried in the apparently innocent
notion of " s t y l i z a t i o n . " This is not different from the "rituali-
z a t i o n " of activity that is the object of the science of ritual.
Burying problems in an apparently simple notion is reminiscent
of the exchange in linguistics to which I referred on page 54:
Charles Hockett (1968) had argued that C h o m s k y ' s "transfor-
m a t i o n s " were unnecessary because the "subtle facts about
various languages" (Hockett 1968:3) they had brought to light
could be explained by analogy. LakofFs review article of 1969
showed that the notion of " a n a l o g y " could account for the data
 Conclusions 439

Chomsky had dealt with, provided the " t h e o r y of analogy"

would incorporate all the devices Chomsky had used to define
transformations, including rule order, recursiveness, variables,
exceptions, cyclical application, and so forth. A theory of
" s t y l i z a t i o n " that explains all the rules of ritual so far discussed
would for the same reason be an adequate theory of ritual.

(2) A second implication of the hypothesis of the ritual origin

of language is that it explains the emergence of logic. The
redundancy of language was gradually discovered—a process
that still continues. This led to attempts to expunge these ritual
survivals and construct more effective and rational systems of
expression and communication. This explains the construction
of the artificial languages of logic and mathematics. If language
reflected meaning in a straightforward manner, it would be
logical to begin with and there would be no need for logic.

(3) The assignments of meanings to ritual sounds had been an

arbitrary but successful process. Not surprisingly, it picked up
momentum and could not be stopped. Meanings were now
assigned to all kinds of entities. Not only were sounds, arbitrar-
ily but systematically, interpreted as meanings, thus producing
language; but trees, mountains, rivers, the universe, animals,
man himself were also endowed with meanings that were be-
lieved to be intrinsic. Everything became meaningful. These
extrapolations from language explain the origin of many features
of magic, mythology, religion, and philosophy itself. They place
Max Müller's and Wittgenstein's views on these features as
"diseases of language" in their evolutionary context.

We are now in a position to discuss the origins of hermeneu-

tics—a topic that is more interesting than hermeneutics itself.
The development from ritual via mantras to language demon-
strates the insufficiency of any approach that treats ritual and
mantras as languages. But if they are not languages, they
certainly are not texts. We have seen that the Modistae and
other medieval grammarians prevailed over the " a u t h o r s " who
440 The Human Sciences

emphasized texts (page 35), and that contemporary linguistics

similarly prevailed over the emphasis on texts and finite corpora
in Whitney, Bloomfield and others (page 52). A more ancient and
venerable European speculation holds that the entire universe is
a book that needs to be deciphered. In his famous work on
European literature and the Latin middle ages, Ernst Robert
Curtius devoted a chapter to this topos of " t h e book as s y m b o l . "
We recognize here the impact of canonical books in the Western
monotheistic religions which the Koran aptly characterized as
"religions of the b o o k . " Today such speculation is no longer
found in scientific theory though it survives in metaphorical
expressions such as " t h e book of n a t u r e . "
In the human sciences the metaphor has survived much longer
and is more than a metaphor whenever we really deal with
texts—as in philology. To look upon other manifestations of
culture as texts, however, reflects the particular nature of
Western religion. To this belief hermeneutics adheres, and it
explains all its foibles. It is found elsewhere in an attenuated
form, especially among scholars concerned with textual inter-
pretation. Jonathan Smith, for example, although untainted by
the irrationalities of hermeneutics, looks for a text whenever he
seeks support for an argument. This self-imposed restriction
prevents him from looking for a global theory of universals. It is
less fatal, yet misleading when it occurs in a metaphor, for
example: "ritual is primarily a matter not of nouns and verbs,
but of qualifiers—of adjectives and a d v e r b s " (Smith 1987:194).
To sum up—ritual originated when the compound activity of
doing things and making sounds grew into a system. This
happened early, after the emergence of the phylum Chordata.
Language was born when that system was used to express
meaning, probably late in the evolution of homo sapiens. Texts
originated as expressions of language when writing was invented
during the historical period. Like most other features of nature,
ritual, therefore, is not a language; a fortiori, it is not a text.
Hermeneutics is confined to texts and text-like objects. As a
corollary I conclude that attempts to bridge the gap between
" A n g l o - S a x o n " and " C o n t i n e n t a l " philosophies by attempting
 Conclusions 441

a synthesis between analytical philosophy and hermeneutics are

fruitless and misguided. This could be shown with reference to
Richard Rorty or Charles Taylor, but I shall illustrate it with
reference to the philosopher who has done most to interpret the
history of the human sciences within a philosophic perspective.

30B. Michel Foucault

Foucault's Les mots et les choses (1966), an "archeology of

the human s c i e n c e s , " shows, like Quine's Word and Object
(I960)—not to mention Words and Things (Gellner 1959), Words
and Objections (Davidson and Hintikka 1969), or Words {Les
mots: Sartre 1963)—that the chief concern of contemporary
philosophy is with the relationships between language and
reality. One way to study these relationships is by concentrating
on how existing languages refer. Since it is unlikely that language
came into being to enable human beings to philosophize about
reality, it is more promising to devise new languages that can do
better than the existing ones. The artificial languages of logic and
mathematics are such languages. Their development is due to
the insight that the objects of reference need not possess the
structure of ordinary language—the mistake exemplified by the
simple illusion that, whenever there is a word, there must be a
thing to which it refers. Another example is the belief that reality
must consist of substances that possess attributes because
language possesses nouns qualified by adjectives. The entities to
which language refers need not be languages or even language-
like even if human beings are predisposed to look upon them as
such.
Even at the relatively early point of his short career when he
wrote Les mots et les choses, Foucault would not say that
philosophy deals primarily with the relationships between lan-
guage and reality. But he says something similar, in inimitable
Foucaultian, and after referring to Nietzsche and Mallarmé:
" t h e whole curiosity in our thought now resides in the question:
What is language, how can we find a way round it in order to
make it appear in itself, in all its plenitude?" (I quote from the
442 The Human Sciences

English translation: Foucault 1970:306). Foiicault's primary aim

was not to overcome the opposition between analytical and
continental philosophies. Rather, he attempted to go, as the title
of a monograph about him (Dreyfus and Rabinow 1982) puts it:
" b e y o n d structuralism and h e r m e n e u t i c s . " Concern with this
problem is similar but more firmly rooted in the French intellec-
tual climate; it was the chief topic of the dispute between
Lévi-Strauss and Sartre (page 259) which focused on structural-
ism and phenomenology. Foucault also formulated it first with
reference to phenomenology, long before hermeneutics had
become fashionable in the English speaking countries.
There are important differences between phenomenology and
hermeneutics but they are insignificant in the present context.
That Foucault was concerned with developments on both sides
of the Channel, affecting the main concerns of modern philoso-
phy, is clear from his attempt to show that both structuralism
and phenomenology have a common ground in a nineteenth
century development in the course of which "language became
an o b j e c t . " This view is largely based upon a discussion devoted
to Franz Bopp (pages 280-294 of the English translation). The
preceding sections of the same chapter (Chapter 8 of Part II)
dealt with the British economist David Ricardo and the French
anatomist Georges Cuvier. In the next chapter, Foucault dis-
cusses " t h e vast play of language contained within a single
s p a c e " as " a leap toward a wholly new form of t h o u g h t " and
notes: " O n l y those who cannot read will be surprised that I have
learned such a thing more clearly from Cuvier, Bopp, and
Ricardo than from Kant or H e g e l " (page 307).
According to Foucault, the objectification of language led to
three related developments: the study of the Indo-European
languages by Bopp; logical algebra or the formalism of thought
by Boole and Russell; and the discovery of the unconscious by
Freud. Many phrases and passages in which this is formulated in
Les mots et les choses are, at least at a first reading, obscure. I
shall therefore quote the French originals along with their
(anonymous and occasionally free) translation:
 Conclusions 443

At the archeological level, the conditions of possibility of a non-verbal

logic and of a historical grammar are the same. The ground of their
positivity is identical (page 297 of the English translation: au niveau
archéologique y les conditions de possibilité d'une logique non verbale et
celle d'une grammaire historique sont less mêmes. Leur sol de positivité
est identique: 1966:310).
The critical elevation of language . . . implied that it had been brought
nearer both to an act of knowing, pure of all words, and to the
unconscious element in our discourses. It had to be either made
transparent to the forms of knowledge, or thrust down into the contents
of the unconscious. This certainly explains the nineteenth century
double advance, on the one hand towards formalism in thought and on
the other towards the discovery of the unconscious—towards Russell
and Freud. It also explains the tendency of one move towards the other,
and of these two directions to cross: the attempt, for example, to
discover the pure forms that are imposed upon our unconscious before
all content; or again, the endeavour to raise the ground of experience, the
sense of being, the lived horizon of all our knowledge to the level of our
discourse. It is here that structuralism and phenomenology find, together
with the arrangements proper to them, the general space that defines
their common ground (page 299; La surélévation critique du langage . . .
impliquait qu'il soit rapproché à la fois d'un acte de connaître pur de
toute parole, et de cela qui ne se connaît pas en chacun de nos discours.
Il fallait ou le rendre transparant aux formes de la connaissance, ou
Venfoncer dans les contenus de l'inconscient. Ce qui explique bien la
double marche du XIXe siècle vers le formalisme de la pensée et vers la
découverte de l'inconscient—vers Russell et Freud. Et ce qui explique
aussi les tentations pour infléchir l'une vers l'autre et entrecroiser ces
deux directions: tentative pour mettre au jour par exemple les formes
pures, qui avant tout contenu s'imposent à notre inconscient; ou encore
effort pour faire venir jusqu'à notre discours le sol d'expérience, le sens
d'être, l'horizon vécu de toutes nos connaissances. Le structuralisme et
la phénoménologie trouvent ici, avec leur disposition propre, l'espace
général qui définit leur lieu commun: 1966:312).

These phrases are not merely occasionally puzzling; they are,

when clear, so demonstrably mistaken that the argument col-
lapses. One problem is that the formalization of logic in Boole
and Russell was not only foreshadowed by Leibniz, but is
connected with the medieval logicians and goes clearly back to
Aristotle. Its advance in the nineteenth century is due to
mathematics, the science of formalization par excellence-and the
444 The Human Sciences
•V
'
first science that Foucault intentionally omitted from his account
(since it is already given "pride of p l a c e " ' i n the history of
science and thought: page ix of the " F o r e w o r d to the English
edition").
The weakest link in F o u c a u l f s account is the inclusion of
historical grammar or comparative grammar of Indo-European
in this trinity of "positive unconscious knowledge" which marks
the beginning of the modern age and constitutes the chief subject
of Foiicault's work (pages xi-xii). Bopp could only develop these
disciplines because he had been influenced and inspired by the
Indian grammarians (Chapters 5 and 6). The entire convergence
of disciplines took place in the nineteenth century as a matter of
fact, but is due to chance: its roots are diverse and have nothing
in common in terms of their "historicity." If it were true that
these diverse disciplines express a unified "positive unconscious
k n o w l e d g e , " we must assume that the sea-faring nations of
Western Europe that initiated the colonization of Asia were
driven by an unconscious desire to discover a theory of lan-
guage.
The "objectification" of language that Foucault deduces from
these considerations as the common ground of structuralism and
phenomenology was the foundation upon which the Indian
grammatical tradition rested. It was explicitly formulated by the
Sanskrit grammarians in terms that happen to exemplify the
structure of the Sanskrit language. We have seen that the most
commonly used Sanskrit term for Päninfs notion of rule, which
goes back t o . t h e science of ritual, is sütra. But there is a
synonym that is widely used, for example by Patanjali in his
account of the infinity of language (page 40): laksana. This
formation consist of a stem "laks" and a suffix u(a)na" which
together denote an activity directed at an object. The object is
formed, in accordance with the general structure of Sanskrit, by
the suffix " y a " being attached to the stem. This results in laksya
which means: " t h e object of laks" The Sanskrit grammarians
accordingly refer to the rules of grammar as laksana. The
objectification of language, therefore, is not a discovery of early
nineteenth century Europe, but of ancient India. It was the
 Conclusions 445

ground upon which the edifice of the Sanskrit grammarians

rested, and was explicitly formulated as such. Because of
Charles Wilkins, Franz Bopp and others it was incorporated into
traditional philology and led to the origination of linguistics in
the West.
Foucault is to be applauded for including in his philosophic
investigations studies of the history of science, especially of the
often neglected human sciences. This inclusion distinguishes
him from many other philosophers who confine themselves to
philosophy. It might seem unfair to blame Foucault for not
having included the study of Indian science in a perspective that
is already so vast. I am not blaming Foucault but merely pointing
out that the history of science cannot be confined to the West.
The separation between East and West, which turned into an
almost unbridgeable gap after the European Renaissance, is
artificial and counterproductive especially in the domain of
science. The prejudice that science and rational thought are only
Western is not only pernicious but an obstacle to the understand-
ing of any scientific development. Science is not the product of
a peculiar European development; it is universal. For a historian
of science or of ideas, trying to account for Bopp without
referring to Pänini is like a historian of logic studying Tarski
without mentioning Aristotle, or an archeologist digging at
Istanbul without knowing that it is built on Byzantium.
That Foucault accepts the prejudice that rational thought is
only Western is clear from the Preface to his book which quotes
from Borges' imaginary Chinese encyclopedia a taxonomy of
categories " t h a t lack all life and place, but are rooted in a
ceremonial space, overburdened with complex figures, with
tangled paths, strange places, secret passages, and unexpected
c o m m u n i c a t i o n s " (page xix). That such a confused assemblage
amuses Foucault is only to be expected—but why would he
assume that an Asian would not be equally amused? I have
shown elsewhere that such taxonomies constitute a literary
genre that has nothing to do with Chinese or Indian logic or
scientific thought, which have both evolved rational taxonomies
like the Western. The common prejudice that Oriental thought is
446 The Human Sciences

irrational, exotic and weird lacks a serious foundation (Staal

1975: Chapters 1-3; 1988a:20-22, passim).
It might still be claimed that this criticism is unfair because
Foucault's search for historicity is, after all, a relativist under-
taking. Elsewhere Foucault has written that: "effective history
differs from traditional history in being without constants . . .
history becomes effective to the degree that it introduces dis-
continuity into our very b e i n g " (Foucault 1977:153-154).
Philp (1985:78), who quotes this passage, adds: " A t the heart
of Foucault's work lies the conviction that there is no constant
human subject to history—that there is no valid philosophical
anthropology—and thus no basis for claiming that we can
identify a coherent and constant human 'condition' or
' n a t u r e . ' " This passage comes from a later work, and Foucault
has, as he has repeatedly noted, often changed his mind. This is
a mark of sincerity, but the latter view is unfortunately false. It
is obvious that humans have changed throughout history and are
different in different parts of the world; but whatever the changes
and differences, it is equally obvious that humans have features
in common which they do not share with monkeys, fish, trees or
stars.
Foucault's speculation about the common ground of phenom-
enology and structuralism shows no trace of relativism or of this
later theory of " a b s e n c e of c o n s t a n t s . " It is a hypothesis about
events that took place in the nineteenth century and is argued for
with detailed information and sustained (though not always
perspicious) argument. Since he obviously wishes to be taken
seriously, Foucault is open to serious criticism.
À final question arises in this context. Is the " s t r u c t u r a l i s m "
Foucault refers to in fact related to what I have generally called
" s c i e n c e " and tried roughly to characterize in Chapters 2 and
26? Dreyfus and Rabinow have tried to pinpoint the kind of
structuralism Foucault has in mind and how he himself differed
from it. They distinguish two kinds of structuralism
1
(1982:53-55): the atomistic in which the elements are completely
specified apart from their role in a system, and the holistic, "in
which what counts as a possible element is defined apart from
 Conclusions AAI

the system, but what counts as an actual element is a function of

the whole system of differences in which the given element is
i n v o l v e d . " Dreyfus and Rabinow state that Foucault addresses
the second, but rejects it in favor of yet another ism that is more
radical and referred to as pragmatic holism. H e r e , even the
identity of a statement depends on the use that is made of it:
4
'Not only can the identity of the statement not be located once
and for all in relation to that of the sentence, but it is itself
relative and oscillates according to the use that is made of the
statement and the way in which it is h a n d l e d " (quoted from the
1972 English translation of Foucault 1966, page 104). Though
this passage is imprecise (for example, to what can the expres-
sions I have italicized refer if not to an identical " s t a t e m e n t " ? ) ,
the state of affairs that Foucault attempts to express is not
uncommon in science. In some sciences the basic concepts are
defined beforehand; in others, they are not. C h o m s k y ' s symbol
44
S , " for example, which stands for " S e n t e n c e , " is not defined
in advance but is contextually defined by the entire system of
linguistics in which it occurs. The notion of ritual provisionally
used in the present study and not defined, approximately, until
now, has a similar status as do, indeed, Sanskrit terms for ritual
such as kriyä and karman. Foucault's pragmatic holism, which
sounds like something wonderfully new, refers in fact to a
common feature of science.

30C. Concluding Generalities

What remains to be carried out is a systematic, formal analysis

of the rules of ritual. This will be an extensive project of research
for which I have only tried to pave the way. in the absence of
solid results to which such an analysis might lead, no general
conclusion can be drawn and there is only room for scattered
remarks.
The regard for unicity on the part of many phenomenologists
and hermeneuticians has a religious origin. Oscar Cullmann has
shown that the word and the concept for " o n c e " occur with
characteristic frequency in the Bible. His book starts with the
448 The Human Sciences

statement: fc"Original Christian faith and thought are not based

upon the spatial distinction between this world and a world
beyond, but upon the temporal distinction O n c e — N o w — T h e n "
(Cullmann 1946:1). In Christianity there is not only one God,
who has one son; God has created the world once, sent his son
once, and there is one day of judgement. The notion of progress,
that hallmark of the West, partly derives from these ideas. Many
Oriental systems, on the other hand, stress repetition, transmi-
gration or reincarnation, numerous worlds and world periods.
This distinction between East and West is related to that
between rectilinear and circular time. The present relevance of
Asian ideas is connected with the need for a future based on
balance and stability rather than unlimited growth.
Uniqueness is often related to chance, for the products of
chance are generally unique. If a native of Beijing marries a
Sioux from Minnewaukan, North Dakota, their child is likely to
be unique. Its characteristics, on the other hand, need not be
unique at all. Everything is unique to some extent, but nothing is
unique in all respects. Many things are similar in one respect,
fewer in t w o , still fewer in more, and no two things are similar in
all respects. Things that are relatively unique may be of special
interest only on account of features that are themselves non-
unique.
A possible objection to the theories sketched in this book is
that the structural complexities of Vedic ritual should perhaps be
dismissed for being unique, anomalous and hence of limited use
to the general theory of ritual. I briefly addressed this question in
the General Introduction to A G N I (1:16):
Perhaps Vedic ritual is too sophisticated, highly developed, and intellec-
tual. This may be so—I could not tell without a major comparative
survey of rituals. However, I suspect that such criticism is on a par with
saying that it does not matter that a certain theory of language does not
apply to Sanskrit or English, because Sanskrit or English are too
sophisticated, highly developed, and intellectual. If anyone were to make
such a claim, the conclusion would be simple and immediate: his theory
of language is itself insufficiently sophisticated, developed, and intellec-
tual. The same verdict must apply to the theories of ritual we have
reviewed. If they cannot account for Vedic ritual, they must go.
 Conclusions 449

The theories I had reviewed in the General Introduction to

A G N I were rather general; they could probably not account for
any ritual. Moreover, I had not, at the time of writing, under-
taken any comparative study of ritual, although the second
volume of A G N I itself contained data that were intriguing in this
respect (especially in the articles by H o o y k a a s , Skorupski and
Strickmann).
These questions about uniqueness are empirical and can only
be definitively answered when we make detailed comparisons of
Vedic ritual with other rituals that appear similarly complex and
sophisticated. Students of comparative ritual face certain prob-
lems, however, for they cannot do fieldwork all over the world
and must accordingly consult the scientific literature, which is
almost entirely preoccupied with problems of semantics. The
only way to overcome this obstacle is by reading between the
lines. This can be done, provided the fieldwork reports of an
author are very explicit and rich, and replete with what is
nowadays called "thick description." In that ease we may try to
achieve what Malinowski required that an anthropological book
should enable its readers to do: construct a theory that accounts
for the author's data but is different from his own theory.
Attempts of this sort have been made in Chapter 25. Subse-
quently, similarities were found between the historically unre-
lated Vedic and Taoist ritual traditions. These will be studied in
a forthcoming publication by Kristofer Schipper and myself.
Several anthropologists have remarked that the Agnicayana
ritual is unique and that what counts in anthropology is the
" c o m m o n " rites of " o r d i n a r y " people. Burkert has voiced a
similar objection to Jonathan Smith and myself:

Your paradigms seem to be of a different character. I think for Jonathan

it is mainly rabbinic; for Frits it is the brahmanic tradition. But both are
special cases, because you have there groups of people who really spend
their lives meditating on ritual or doing ritual, and they keep their own
identity just by this, in my view, nearly obsessive concentration on
ritual. Î am looking at more humane kinds of ritual that are communica-
tive and are understandable to a certain degree, though perhaps not
totally understandable (Burkert 1987:233).
450 The Human Sciences

The question is: how different is a special case? The obses-

siveness with ritual that Burkert pinpoints sheds as much light
on ritual as Mozart's or Euler's obsessions shed on music or
mathematics. It is clear that musicologists should study sympho-
nies as well as lullabies, and mathematicians cylindrical algebra
as well as fractions. It is unlikely that it would be fruitful to
confine our science to what is ordinary, human, and more or less
intelligible. The most curious and challenging things are neither
average, nor immediately intelligible. It is unlikely that â theory
of ritual, if adequate, will be simple. There will be complaints
about its myriad rules as there have been about Chomsky and
Halle's Sound Pattern of English, Euclid's Elements, or the
Srauta Sütras.
Science includes analysis of the apparently unique, but ac-
counts for it in terms of its opposite: the general and universal.
To make something intelligible is to relate it to general and
universal features. Universality characterizes the best work also
in the human sciences. Jakobson discovered the simplest and at
the same time most well-established universals—distinctive fea-
tures. For Chomsky, the search for universals of language is the
essence of linguistics. For Lévi-Strauss, anthropology seeks
universals of the human mind.
Universals need not be universal in the sense that they must
be found everywhere. The issue, at first sight puzzling, is
discussed in 4 t U n i v e r s a l s " (Staal 1988a) with special reference
to logic and linguistics, but is straightforward within phonology
itself. The click sounds of Xhosa and Hottentot are described in
terms of universal distinctive features but these sounds occur
only in a few language families. Ritual and mantras are particu-
larly well developed in Asia. Mantras seem to be common in
Africa, too, but one has to read between the lines (Dammann
1963:116-131, 228-233). The modern West preserves little more
than abracadabra, amen, aleluja and hosanna, although Pente-
costals add uncounted numbers of meaningless sounds almost
every day. That these ritual sounds involve musical patterns that
we now recognize as recursive was pointed out by the founder of
the phenomenology of religion, G. van der Leeuw. In a chapter
 Conclusions 451

on " M u s i c and Religion" he drew attention to the schema ABA

in: " K y r i e eleison—Christe eleison—Kyrie eleison" (van der
Leeuw 1948:378).
Although the term mantra is of Sanskrit origin, I have tried to
show that the study of mantras deals with features that are
universal. We can now speculate and even predict—other fea-
tures of science not absent from the human sciences. When I
first visited Bali, I predicted that I would hear Vedic mantras.
Within a few days after my arrival, Ï heard a priest at a temple
ritual recite a fragment of the Gäyatri (page 306). In subsequent
conversation, it turned out that he knew the entire mantra. I thus
obtained a verification of my prediction. Later I learned that the
Gâyatrî in its entirety is not used in traditional Balinese ritual,
but had been recently introduced from India by a Hindu mis-
sionary. My hypothesis was falsified—and the episode illustrates
that we should not rush to conclusions.
Hypotheses may be suggested by negative facts, for example,
the fact that the Old Testament does not contain ritual texts.
This does not imply that there was no ritual: the biblical texts are
" w h a t whoever wrote them wanted you to know, as opposed to
what they might k n o w " (Smith 1987:210). Although there is no
information on biblical rites. Smith adds (ibid.): " T h e rabbis
gleefully will supply that lacuna for you, but they don't know
any more than I do on the subject." It is true that an argumen-
tiim e silentio does not prove much, but a hypothesis may be
established with a good measure of probability if it is supported
by a general theory of ritual and a sound knowledge of a
particular context—in this case, the Bible and whatever data
have been made accessible by the biblical sciences.
In anthropology, " l a n g u a g e " occupies the pride of place. This
is as it should be for language is the most characteristic feature
of the human animal, more important, for example, than a
prohibition on incest, variable as to content. The importance of
language, however, does not justify the emphasis on meanings
and symbols adopted in anthropology and religious studies.
Mantras are not sounds of an unknown language that has to be
deciphered, and neither mantras nor ritual acts are symbols that
452 The Human Sciences

express meanings. Both are governed by rules, but they are rules
without meaning. Our confusions in interpreting ritual systems
are largely caused by the disconcerting concatenation of two
independent and unrelated facts: ritual systems are like language
in that they are governed by rules, but unlike language in that
they do not express meanings. For their study, understanding
and analysis, such systems accordingly require syntactic theo-
ries that deal with rules, not semantic theories that deal with
meanings. The Indian Srauta Sütras have paved the way.
Indian and Western civilization have been characterized by
their emphasis on linguistics and mathematics, respectively
(e.g., Ingalls 1954, Ruegg 1978, Staal 1963a, 1965b). It would be
more accurate to say that the originality of Indian science lies in
the discovery of rule-governed activity. The importance of rules
is that they express regularities, similar to the regularities of
nature and the non-human universe in general—whether the
latter are regarded as " l a w s of n a t u r e " or as mathematical
formulas. An important distinction between rules that govern
activities and rules that describe other regularities is that agents
who perform activities may be aware of the rules they follow or
observe. This is not confined to humans; dogs know where they
may not go. A distinction must be made between following a rule
and knowing that (or knowing what) rule one follows. The
planets go around the sun in accordance with regularities de-
scribed by rules, but do not know or consciously follow any of
these rules.
When they are aware of the rules they follow, men and other
animals add a feature to patterns of regularity that already exist
in the universe. If we adopt the distinction of Indian philosophy
between pradhäna or "principal e n t i t y " and guna or "subor-
dinate e n t i t y " (page 230), following a rule is pradhäna, being
aware of it, guna. This particular guna exhibits the nature of
consciousness, that much overrated philosophers' pet (E. W.
Beth called it " h e t troetelkind der traditionele wijsbegeerte"), a
kind of tracing in the mind or brain as yet unexplained. For us
individuals such awareness does not only make a difference; it
may be a matter of life and death. In the development of the
 Conclusions 453

species it plays a less important role or no role at all. The limited

relevance of Wittgenstein's remarks about following rules
(Chapter 6B) is connected with this difference between individ-
uals and the species. Innate knowledge of rules is the result of a
process that may have gone on for millions of years. It may be
unconscious because it may reflect knowledge acquired by the
species that never became conscious in any individual.
Awareness of rules is not only a subordinate feature; it also
varies with individuals and circumstances. In the use of lan-
guage, some speakers or hearers are more clearly aware of the
grammatical rules they follow than others. N o speaker who is
not a linguist is aware of complex features such as rule order or
the rules that account for the difference between " J o h n ' s
eagerness to p l e a s e " and * " J o h n ' s easiness to p l e a s e " (page 25).
In ritual, Heesterman( 1962:18;above, page 154) has discussed a
case that exhibits, similar features. When the victim is tied to the
sacrificial pole, " t h e cord is fastened to the right foot, goes
round the left side of the neck and is then wound round the right
horn and finally fastened to the s t a k e / ' This explains the origin
of the rite: originally, the victim's head was cut off while it was
tied to the stake. In the present context, we make a different
observation: most ritualists who conform to the rule are unlikely
to be aware of the historical origins of what they are doing. Some
might recognize their plausibility if they were pointed out to
them; in some individuals, an awareness or semi-awareness of
their original motivation may be lingering. Whatever the degree
of awareness, what counts in ritual is what the ritualist does.
The meaninglessness of ritual and the myth of the two cultures
are two tips of the iceberg of human nature. What remains under
water is not only the roots of religion, but an even more ancient
and archaic obsession with language and its symbol-generating
and mythopoeic functions. This reflects the nature of man as the
unique creature that is endowed with language and loves to talk.
The origination or discovery of language was the most momen-
tous step the human animal took in the course of its evolution. It
also threw him into immediate and profound confusion. To
attach meanings to sounds in order to express oneself or
454 The Human Sciences

establish communication opens up a world of new possibilities.

To attach meaning indiscriminately is by the same token pro-
foundly disorienting. This explains Wittgenstein's remark that
philosophical problems have an aura of profundity because they
are as deep as our language {Philosophical Investigations §111).
Although every zoologist knows that an elephant does not
have meaning, students of the humanities and social sciences
attach meaning to many expressions and manifestations of
humanity as if they were linguistic expressions. An artificial gap
between us and the rest of the universe is created by myths that
have solidified into doctrines, e.g., that man is unique, created in
the image of a deity, or beyond rational understanding. Under-
lying such ideas is the assumption that man must make sense. To
an unbiased observer, much of that sense seems arbitrarily
assigned, evanescent, or due to chance. The available evidence
suggests that man can make as much sense as he likes, but does
not make intrinsic sense. Like our language, and all other
creatures, we humans originated by chance. The thought may
have occurred to King Solomon when he was contemplating ants
(Proverbs 6:6). It must have been in the mind of the poet of
Rigveda 10.129.7:

Whence this creation has arisen—perhaps it formed itself, or perhaps it

did not—the one who iooks down on it, in the highest heaven, only he
knows—or perhaps he does not know (translation OTlaherty 1981:25-
26).
 Excursus

"Writing in the S k y "

Hermeneutics and Astrology:

A Neoplatonic Parallel
The Enne ads of Plotinus are called after the number {enne a,
" n i n e " ) of chapters in each of its books just as the Astcidhyâyîoï
Pânini is called after the number (asta, " e i g h t " ) of its chapters.
In a passage of] the fourth Ennead, the great Neoplatonic
philosopher, who was born in 205 A . D . in Upper Egypt and
studied in Alexandria, stresses the ritualistic nature of prayer
and argues that its effect is neither due to a deity or heavenly
body nor based on the intention of the person who prays but
derives from the correctness with which it is performed:
In the stars (that is, the gods) there is no will to answer our prayers . . .
their powers are used without will, whether they are provoked (by us) or
not, through a scientifically effective procedure {rkxvy)) • • • The per-
former may be a bad person (KÖKÓS)—it need not surprise us: bad people
fetch water from the rivers. The being that gives does not know that it
gives; it simply gives (IV 4.42.3-4, 6-8, 14-17 Bréhier).

The explanation of the efficacy of prayers lies in the fact that the
entire universe is like a single organism:
Prayer is effective because one part of the universe sympathizes with
another, as in the stretched string of a lyre when the vibration moves
from low to high. Often, when one string vibrates, another perceives it,
as it were, because they are tuned in concord and harmony. Vibrations
move even from one lyre to another, showing how far the "sympathy"
goes. In the universe, too, there is one universal harmony even if it
consists of contraries (IV 4.42.11-14).

This picture is more attractive than Cartesian dualism, but it

does not explain how particular heavenly bodies influence our
456 Excursus

destiny. This is the topic of the third chapter of the second

Ennead:

If heavenly bodies signify (o-rjiAaipoixnv) future events, like we say of

many other things that they are signs icr-q^avriKà) of the future, whence
comes this quality? How does order come about? For nothing could
announce events that are not arranged in order. The stars must be like
letters that are constantly being written in the sky, or that have been
written once and for all, and that move; while performing another
activity, they also, through that same activity, convey meaning
l 3.7.1-7).

The fifth century Roman Neoplatonist Macrobius referred to this

view and explicated it further in his Commentary on Cicero's
Scipio's Dream:

In his treatise Do the stars act?, Plotinus declares that nothing happens
to man because of the force or power of heavenly bodies; but the events
that destiny has in store for us are announced by the seven wandering
stars by means of their locations and rétrogradations. Similarly, birds
signify future events of which they are unaware by the way they fly or
alight, and by their wings or songs. Hence we are right in saying that
Jupiter is auspicious or Mars inauspicious, since happy events are
signified {significantur) by the former and unhappy events by the latter
(translated from the French translation in Duhem 1914, 11:310).

Duhem places this discussion in its philosophic context by

quoting an observation due to Bréhier (1907:1). The Greek
philosophers discussed the nature of causation by specifying
what kind of entity can be a cause. The Pre-Socratics recognized
only causes that are material and visible; Plato and Aristotle
sought the principles of things in elements that are perspicuous
to clear thought, tha* is, intelligible things; the Stoics and
Epicurians returned to the Pre-Socratic view that causes must be
material; the Neoplatonists reacted against this and returned to
Plato by again accepting only intelligible causes. Pre-Socratics,
Stoics and Epicurians were all in a position to accept the claims
of astrology according to which stars are causes of terrestrial
events. Their disbelief in unseen and immaterial causes resem-
bles that of the modern empiricists and positiyists discussed in
 Writing in the Sky 457

Chapter 2, but is, in the case of astrology, not based upon

empirical facts. Plotinus did not accept material causation, but
interpreted the stars as endowed with meaning or significance:
they inform us about events that are caused by other agencies.
The naive astrological view, that the stars cause events, runs
parallel to the naive view that rites cause events. Some
Brähmanas declare that the rising and setting of the sun is caused
by the performances of the morning and evening Agnihotra. But
if one does not believe that stars or rites cause or influence
events,—and neither Plotinus, nor contemporary hermeneuti-
cians do—then what is there left to believe? Skirting the obvious
answer ( ' ' N o t h i n g " ) what we need is an explanation for the fact
that people entertain such beliefs. Both Neoplatonists and
hermeneuticians felt that there must be something left to believe
in. They accordingly postulated " m e a n i n g s " or "significa-
t i o n s , " a scientific procedure provided it is supported by empir-
ical data. Neoplatonists and hermeneuticians justify their belief
by treating a variety of entities as if they were texts. Hermeneu-
ticians stop short of stars, which Plotinus included without
compunction.
That Plotinus did not believe in rituals is related by his pupil
Porphyry in his Life of Plotinus (10.33-38):

Amelius was a lover of rituals (</>IÀO#T)TOS). He would never miss the

New Moon Ceremonies or any other recurring festivals. Once he wanted
to take Plotinus with him, but the latter said: "the gods should come to
me, not I go to them." We did not understand the mentality underlying
these proud words, and we did not dare to ask.

The Neoplatonic parallel indicates that the urge to attach mean-

ings to what is essentially meaningless is a common feature of
the human animal. It occured in the Brähmanas and re-appared
in hermeneutics and related philosophies like " symbolic anthro-
p o l o g y . " This extrapolation from the systematic reference to
meanings that characterizes language must have come into
being—probably not very long—after the origin of language.
 Bibliography

A. Texts

The texts are listed without particulars regarding editions, translations,

commentaries, etc., which are known to specialists. Non-specialists may
derive comfort from the thought that mantras should be heard, not read.

Aitareya Äranyaka
Aitareya Brähmana
Anattalakkhana Sutta
Äpastamba Srauta Sutra
Astädhyäyf
Äsvaläyana Grhya Sütra
Atharvaveda Prätisäkhya
Atharvaveda Samhitä
Baudhäyana Srauta Sûtra
Brhad Äranyaka Upanisad
Chândogya Upanisad
Jaiminfya Aranyageyagäna
Jaiminfya Ärcika
Jaiminfya Grämageyagäna
Kausïtaki Brâhmana
Kätyäyana Srauta Sütra
Kauthüma-Ränäyaniya Grämageyagäna
Mahäbhäsya
Maiträyani Samhitä
MFmämsä Paribhäsä
MFmâmsâ Sütra
MFmamsä Sütra Bhäsya
Närada Siksä
Nirukta
Pancavimsa Brähmana
Prapancasära
Puspasütra
Rgveda Prätisäkhya
Rgveda Samhitä
Sämaveda Samhitä
Satapatha Brähmapa
Stobhänusamhära
460 Bibliography

Taittirfya Àranyaka
Taittinya Brâhmana
Taittinya Prätisäkhya
Taittinya Samhitâ
Ühagäna
Väjasaneyi Prätisäkhya
Vedänta Sütra Bhäsya

B. Other Publications

Aarsletï. H. (1970). "The History of Linguistics and Professor Chomsky,"

Language 46:570-585
AGNI: see Staal 1983a
Alper. H. P. (1988). "Siva's Language-game," in: Alper, ed.: 249-294
Alper. H. P. (ed.) (1988). Understanding Mantras, Albany. N.Y.
Anscombe, G. E. M. (1959). An Introduction to Wittgenstein's Tractatus,
London
Apel. W. (1970). Harvard Dictionary of'Music, Cambridge, Mass.
Aristotle. Categoriae et Liber de Interpretations ed. L. Minio-Paluello.
Oxford 1949
Aristotle. Categories and De Interpretationen transi. J. L. Ackrill. Oxford 1963
Aristotle. Metaphysics, ed. W. D. Ross. I-II. Oxford 1924
Arnold. Harold (forthcoming). "The Hinduization of Vedic Ritual," in:
Strickmann (ed.) 1988
Auden. W. H. (1974). Forewords and Afterwords, New York
Austin. J. L. (1961), Philosophical Papers, Oxford
Austin. J. L. (1962). How to do Things with Words, Oxford
Ayer. A. J. (1936) Language, Truth and Logic, New York
Balasooriya, S. et al. (1980). Buddhist Studies in Honour of Walpola Rahula,
London and Sri Lanka
Baptista. Luis (unpublished), "On Parallels between Bird Song and Speech
Development."
Barbour, Ian (1980). "Paradigms in Science and Religion," in: Gutting, Gary
(ed.) 223-245
Barlow, George W. (1977). "Modal Action Patterns," in: Sebeok (ed.) 98-134
Beck, Leslie (ed.) (1962). La philosophie analytique, Paris
Benfey. Th (1874). Einführung in die Grammatik der vedischen Sprache,
Göttingen
Beth. Evert W. (1959). The Foundations of Mathematics: A Study in the
Philosophy of Science, Amsterdam
Bethel, T. (1976). "Darwin's Mistake," Harpers (February)
Bharati. Agehananda (1965), The Tantric Tradition, Garden City
Bhattacharya. K. (1973), L'ätman-brahman dans le bouddhisme ancien, Paris
 Bibliography 461

Bieeker, C. J.. G. W. J. Drewes and K. A. H. Hidding (eds.) (1951).

Proceedings of the 7th Congress for the History of Religions. Amsterdam
Bloomfîeld, L. (1933). Language, New York
Bloomfield, M. (1906). A Vedic Concordance. Cambridge. Mass. Reprint 1964.
Boehtlingk. Otto (1887). Pänini's Grammatik. Leipzig. Reprint 1964.
Bokenkamp, Stephen R. (1983), "Sources of the Ling-pao Scriptures," in:
Strickmann 1981-1983. 11:434-486
Bollée, W. B. (1981), 'The Indo-European Sodalities in Ancient India/'
Zeitschrift der deutschen morgenländischen Gesellschaft 131:172-191
Boulez. Pierre (1966). Relèves d'apprenti, textes réunis et présentés par Paul
Thevenin, Paris
Bowra, Sir Maurice (1966). "Dance. Drama, and the Spoken Word/* in:
Huxley, ed. 387-392
Bradley, F. H. (1893), Appearance and Reality, London
Bréhier, Emile (1907). La théorie des incorporels dans l'ancien Stoïcisme.
Paris
Brough, J. (1950), "The meaning of niVhnu in the Brähmanas." Siddheshwar
Vanna Commemoration Volume, 1:126-130 '
Brough. J. (1953), The Early Brahmanical System of Gotra and Pravara.
Cambridge
Brough, J. (1971). "Soma and Amanita muscaria."* Bulletin of the School of
Oriental and African Studies 34:331—362
Brown, W. N. (1965). "Theories of Creation in the Rgveda/* Journal of the
American Oriental Society 85:23-34
Brunner, Hélène (1986a), "Les membres de Siva/' Asiatische StudienlEtudes
Asiatiques 40:89-132
Brunner, Hélène (1986b), ""Mandaia et Yantra dans le Sivaisme Agamique:
Definition, description, usage ritual/' in Padoux (ed.) 11-35
Brunner-Lachaux, Hélène (1963-1977), Somasamhhupaddhati. I-IIL Pondi-
chéry
Burghardt, Gordon M. and Bekoff, Marc (1978), The Development of Behav-
ior: Comparative and Evolutionary Aspects, New York
Burkert, Walter (1972), Homo Necans: Interpretationen altgriechischer Op-
ferriten und Mythen, Berlin
Burkert, Walter ( 1979). Structure and History in Greek Mythology and Ritual.
Berkeley
Burkert, Walter (1983), Anthropologie des religiösen Opfers, München
Burkert, Walter (1987), see: Hamerton-Kelly
Bursill-Hall, G. L. (1974), "Toward a History of Linguistics in the Middle
Ages, 1100-1450/' in: Hymes (ed.) 77-92
Caland, W. (1903), "Über das rituelle Sütra des Baudhäyana/' Abhandlungen
für die Kunde des Morgenlandes 12:1-65. Reprint 1966.
Caland, W. (1904-1913). Baudhävana-srauta-sütra, Calcutta. I-III
462 Bibliography

Caland, W. and V. Henry (1906), L'Agnistoma: description complète de la

forme normale du sacrifice de Soma dans le culte védique, I—II, Paris
Caland, W. (éd.) (1926), Ziegenbalg s Malabarisches Heidenthum, Amster-
dam
Carnap, Rudolf (1936-1937), "Testability and Meaning," Philosophy of Sci-
ence 3:419-471 and 4:1-40
Carnap, Rudolf (1947), Meaning and Necessity, Chicago
Carnap, Rudolf, Charles Morris and Otto Neurath (1936—1962), International
Encyclopedia of Unified Science, Chicago
Cassirer, Ernst (1925), Philosophie der symbolischen Formen, Berlin, I-II
Cherry, C. (1974), Pragmatic Aspects of Human Communication, Dordrecht-
Boston
Chomsky, Noam (1957), Syntactic Structures, The Hague
Chomsky, Noam (1961), "On the notion 'rule of grammar'," Structure of
Language and its Mathematical Aspects, Proceedings of symposia in
applied mathematics 12:6-24
Chomsky, Noam (1964), Current Issues in Linguistic Theory, The Hague
Chomsky, Noam (1965), Aspects of the Theory of Syntax, Cambridge, Mass
Chomsky, Noam (1966), Cartesian Linguistics. A Chapter in the History of
Rationalist Thought, New York and London
Chomsky, Noam (1968), Language and Mind, New York
Chomsky, Noam (1975), Reflections on Language, New York
Chomsky, Noam (1980), Rules and Representations\ New York
Chomsky, Noam (1981), Lectures on Government and Binding, Dordrecht-
Cinnaminson
Chomsky, Noam (1986), Knowledge of Language: Its Nature, Origin and Use,
New York
Chomsky, Noam and Morris Halle (\96S), The Sound Pattern of English, New
York
Coedès, G. (1967), The Making of South East Asia, Berkeley-Los Angeles
Converse, H. S. (1974), "The Agnicayana Rite: Indigenous Origin?" History
of Religion 14:81-95
Conze, Edward (1969), Buddhist Meditation, New York
Coward, H. and Sivaraman, K., (eds.) (1977), Revelation in Indian Thought. A
Festschrift in Honour of Professor T. R. V. Murti, Emeryville
Cullmann, Oscar (1948), Christus und die Zeit. Die urchristliche Zeit- und
G e schicht s-Auffassung, Zurich
Curtius, Ernst Robert (1948), Europäische Literatur und lateinisches Mit-
telalter, Bern
Dammann, Ernst (1963), Die Religionen Afrikas. Stuttgart
Dandekar, R. N. (1946-1973)., Wedle Bibliography, Poona, I—III
Dandekar, R. N. (1982), "Some Aspects of Vedic Exegesis," Indologica
Taurinensia\0:7\-&\
 Bibliography 463

Dandekar, C. G., Kashikar, C. G. and others (1958-1973), Srautakosa:

Encyclopaedia of Vedic Sacrificial Ritual, Poona
Darwin, Charles (1872), The Expression of the Emotions in Man and Animals,
London
Davidson, Donald (1976), "Introduction," to Part II in: Steklis, etc. (ed.) 18-19
Davidson, Donald and Hintikka, Jaakko (1969), Words and Objections: Essays
on the Work of W. V. Quine, Dordrecht
Davison, A. T. and Apel, W. (1974), Historical Anthology of Music, Cam-
bridge, Mass, I—II
Demiéville, P. (1930), "Bombai," Hôbôgirin. Dictionnaire encyclopédique du
bouddhisme d'après les sources chinoises et japonaises 1:93-113
Demiéville, P. (1980), "Notes on Buddhist Hymnology in the Far East," in:
Balasooriya, ed.
Diehl, C. G. (1956), Instrument and Purpose: Studies on Rites and Rituals in
South India, Lund
Dilthey, Wilhelm (1949), Grundriss der allgemeinen Geschichte der Philoso-
phie, ed. by H.-G. Gadamer, Frankfurt am Main
Drewes, G. W. J. (1925), Drie Javaansche Goeroe's, Leiden
Dreyfus, Hubert L. and Rabinow, Paul (1983), Michel Foucault. Beyond
Structuralism and Hermeneutics, Second Edition, with an Afterword by and
an Interview with Michel Foucault, Chicago
Duhem, Pierre (1914), Le système du monde: histoire des doctrines cosmolo-
giques de Platon à Copernic, Paris
Dumont, Louis (1959), "Le renoncement dans les religions de Finde,"
Archives de sociologie des religions 7:45-69
Dumont, Louis (1966), Homo hierarchicus: Essai sur le système des castes,
Paris
Dumont, P.-E. (1939), L'Agnihotra, Baltimore
Dumont, P.-E. (1957), "The Full-Moon and New-Moon Sacrifices in the
Taittirïya-Bràhmana," Part I, Proceedings of the American Philosophical
Society 101:216-243
Dundes, Allan (1978), "Into the Endzone for a Touchdown," Western Folklore
37:75-88
Durkheim, Emile (1915), Les formes élémentaires de la vie religieuse. Le
système totemique en Australie, Paris. English translation 1915, London and
New York
Edwards, Paul (ed.) (1967), The Encyclopedia of Philosophy, New York -
London, I-VI1I
Eggeling, Julius (1882-1900 = 1966), The Satapatha Brâhmana, I-V, New
Delhi
Eibl-Eibesfeldt, I. (1966), "The Fighting of Marine Iguanas," in: Huxley, ed.
475-476
Eibl-Eibesfeldt, I. (1971), Love and Hate, New York
464 Bibliography

Eibl-Eibesfeldt, I. (1975). Ethology. New York

Eldredge, Niles (1982), "An Ode to Adaptive Transformation," Nature
296:508-509
Eldredge, Niles and Gould, Stephen Jay (1972), "Punctuated Equilibria: An
Alternative to Phyletic Gradualism," in: Schopf, ed.
Eliade, Mircea (1958), Yoga: Immortality and Freedom, Princeton
Eliade, Mircea (1964), Shamanism: Archaic Techniques of Ecstasy, London
Elias, Norbert (1987), Involvement and Detachment, Oxford
Faddegon, B. (1927), "Ritualistic Dadaism," Acta O rient alia 5:177-195
Fehling, Detlev (1974), Ethologische Überlegungen auf dem Gebiet der
Altertumskunde: Phallische Demonstration—Fernsicht—Steinigung, Mün-
chen
Filliozat, Jean (1946), "Les origines d'une technique mystique indienne,"
Revue philosophique 136:208-220
Filliozat, Jean (1965), "Kailâsaparamparâ," in: Felicitation Volume of South-
east Asian Studies presented to His Highness Prince Dhaninivat, Bangkok
Findly, E. B. (1988), "Mäntra Kavisastä: Speech as Performative in the
Rgveda," in: Alper, ed. 15-47
Firth, Raymond (1967), Tikopia Ritual and Belief, Boston
Foucault, Michel (1966), Les mots et les choses, Paris. Translated as:
Foucault, Michel ( 1970), The Order of Things. An Archaeology of the Human
Sciences. New York
Foucault, Michel (1977), Language, Counter-Memory, Practice, Oxford
Frank. Joseph ( 1984). "The Master Linguist," The New York Review of Books,
April 12
Frauwallner. Erich (1956). Die Philosophic des Buddhismus. Berlin
Freud. Sigmund (1907), "Obsessive Acts and Religious Practices," Collected
Papers 2:25-35
Fung Yu-Lan (1952), .4 History of Chinese Philosophy, transi. Derk Bodde,
Princeton, 1
Gadamer, H.-G. ( 1965), Wahrheit und Methode, Tubingen
Gardner, Martin (1987). "Giving God a Hand," The New York Review of
Books, August 13, 17-23
Garge, D. V. (1952). Citations in Sabara-Bhâsya. Poona
Geertz, Clifford (1^60). The Religion of Java. Glencoe. 111.
Geertz. Clifford (1964a). "'Internal Conversion' in Contemporary Bali,"
Malayan and Indonesian Studies Presented to Sir Richard Winstedt. Ox-
ford, 282-302, reprinted in: Geertz 1973:170-189
Geertz, Clifford (1964b), "Tihingan: A Balinese Village" Bijdragen tot de
Taal-, Land- en Volkenkunde 120.1-33
Geertz, Clifford (1968), Islam Observed, New Haven
Geertz, Clifford (1972), "Deep Play: Notes on the Balinese Cockfight."
Daedalus 101:1-37, reprinted in: Geertz 1973:412-453
Geertz. Clifford (1973). The Interpretation of Cultures, New York
 Bibliography 465

Geertz, Clifford (1976), "Hooykaas on (the) Geertz(es): A Reply/* Archipel

12:219-225
Geertz, Clifford ( 1980), Negara: The Theatre-State in Nineteenth Century Bali,
Princeton
Geertz, Clifford (1983), Local Knowledge: Further Essays in Interpretive
Anthropology, New York
Geertz, Clifford ( 1985), "Waddling In," The Times Literary Supplement, 4,288
(June 5) 623
Geertz, H. and C. Geertz (1975), Kinship in Bali, Chicago
Gellner, Ernest (1959), Words and Things. A Critical Account of Linguistic
Philosophy and a Study in Ideology, London
Gennep, A. van (1909), Les rites de passage, Paris
Gennep, A. van (1911), "De la méthode à suivre dans l'étude des rites et des
mythes," Revue de l'Université de Bruxelles, 502-523. Dutch translation:
Wetenschappelijke Bladen 3:97-110. English translation in: Waardenburg
1973 1:287-300
Ghurye, K. G. (1950), Preservation of Learned Tradition in India, Bombay
Giesey, R. E. (1960), The Royal Funeral Ceremony in Renaissance France.
Geneva
Girard, R. (1972), La violence et le sacré, Paris
Gombrich, Richard ( 1971), Prec ept and Practice, Oxford
Gombrich, Richard (1974), "Eliade on Buddhism." Religious Studies
10:225-231
Gombrich, Richard (1985), "Knowledge of the Unknowable." The 'limes
Literary Supplement, 4,278 (March 29) 359-360
Gonda, J. (1954). Aspects of Early Visnuism, Utrecht
Gonda, J. (1963a), The Vision Of the Vedic Poets, The Hague
Gonda, J. (1963b), "The Indian Mantra," Oriens 16:244-297
Gonda, J. (1965), Change and Continuity in Indian Religion. The Hague
Gonda, J. (1977), The Ritual Sutras, Wiesbaden
Gonda, J. (1979), "Satarudriya," Sanskrit and Indian Studies: Essays in
Honor of Daniel H. H. Ingalls, Dordrecht 75-91
Gonda, J. (1980a), The Mantras of the Agnyupasthäna and the Sauträmani,
Amsterdam
Gonda, J. (1980b), Vedic Ritual: The Non-Solemn Rites. Amsterdam
Gonda, J. (1981), The praügasastra, Amsterdam
Gonda, J. (1982), The Haviryajnâh Somâh: The interrelations of the Vedic
solemn sacrifices. Sänkhäyana Srautasùtra I4J-I3, Translation and Notes,
Amsterdam
Gonda, J. (1985), The Ritual Functions and Significance of Grasses in the
Religion of the Veda, Amsterdam
Goodman, Nelson (1955), Fact, Fiction and Forecast, Cambridge Mass.
Goody, Jack (ed.) (1968), Literacy in Traditional Societies. Cambridge
466 Bibliography

Goody, Jack (1977), "Against 'Ritual': Loosely Structured Thought on a

Loosely Defined Topic," in: Moore and Myerhoff, eds. 25-35
Goris, Roelof (1926), Bijdrage tot de Kennis der Oud-Javaansche en Balinee-
sche Theologie, Leiden
Goudriaan, T. (1986), "Kubjikä's samayamantra and its manipulations in the
Kubjikâmata," in: Padoux, ed., 141-167
Goudriaan, T. and Gupta, S. (1981), Hindu Tantric and Säkta Literature,
Wiesbaden
Gould, Stephen Jay (1977), Ever Since Darwin: Reflections in Natural History,
New York
Graham, A. C. (1959), Being in Western Philosophy compared with ShihIFei
and Yu/Wu in Chinese Philosophy," Asia Major (new series) 7:79-112
Graham, A. C. (1965), " 'Being1 in Linguistics and Philosophy," Foundations
of Language 1:223-231
Graham, A. C. (1967), " 'Being' in Classical Chinese," in: The Verb "Be" and
its Synonyms. Philosophical and Grammatical Studies, ed. J. W. M.
Verhaar, Dordrecht, 1:1-39
Grapard, Allan G. (1982), "Japan's Ignored Cultural Revolution: The separa-
tion of Shinto and Buddhist divinities in Meiji (shimbutsu bunri) and a
case-study: TÖNOMINE" (mimeographed)
Gray, J. E. B. (1959a), "An Analysis of Rgvedic Recitation," Bulletin of the
School of Oriental and African Studies 22:86-94
Gray, J. E. B. (1959b), "An Analysis of Nambudiri Rgvedic Recitations and
the Nature of the Vedic Accent," Bulletin of the School of Oriental and
African Studies 22:499-530
Green, Douglas, M. (1965), Form in Tonal Music, New York
Gren-Eklund, G. (1978-1979), "An upanisad of sâman," Orientalia Suecana
27/28: 148-158
Gulik, R. H. van (1980), Siddham: An Essay on the History of Sanskrit Studies
in China and Japan, Nagpur
Güntz, Th. (1861), Der Geisteskranke in seinen Schriften, Thonberg bei
Leipzig
Gutting, Gary (1980), Paradigms and Revolutions. Appraisals and Applica-
tions of Thomas Kuhn's Philosophy of Science, Notre Dame
Güttinger, Hans Rudolf (1970), "Zur Evolution von Verhaltnisweisen und
Lautäusserungen bei Prachtfinken (Estrilididae)," Zeitschrift für Tierpsy-
chologie 27/8:1011-1075
Halle, Morris (1957), "In Defense of the Number Two," Studies Presented to
Joshua Whatmough on his 60th Birthday, The Hague, 65-72
Halle, Morris (1975), "Confessio Grammatici," Language 51:525-535
Halle, Morris (1983), "On the Origins of the Distinctive Features," Interna-
tional Journal of Slavic Linguistics and Poetics, 27:Supplement, 77-86.
Hamerton-Kelly, Robert G. (ed.) (1987), Violent Origins: Walter Burkert, René
 Bibliography 467

Girard, and Jonathan Z. Smith on Ritual Killing and Cultural Formation,

Stanford
Hammel, E. A. and Simmons, W. S. (1970), Man Makes Sense: A Reader in
Modern Cultural Anthropology, Boston
Harrnan, G. (1968), Review of Chomsky 1966, Philosophical Review
77:229-235
Harnad, Stephen R., Steklis, Horst D. and Lancaster, Jane (eds.) (1976),
Origins and Evolution of Language and Speech, Annals of the New York
Academy of Sciences 280:1—914
Hartshorne, Charles (1973), Born to Sing: An Interpretation and World Survey
of Bird Song, Bloomington
Harvard Dictionary of Music (1970), ed. W. Apel, Cambridge
Hayes, E. Nelson and Hayes, Tanya (1970), Claude Lévi-Strauss: The Anthro-
pologist as Hero, Cambridge, Mass, and London
Heesterman, J. C. (1957), The Ancient Indian Royal Consecration, 's-Gra-
venhage
Heesterman, J. C. (1962), "Vrätya and Sacrifice," Indo-Iranian Journal 6:1-37
Heesterman, J. C. (1964), "Brahmin, Ritual and Renouncer," Wiener Zeit-
schrift fur die Kunde Sud- und Ostasiens 8:1-31
Heesterman, J. C. (1983), "Other Folk's Fire" in: AGNI 11:76-94
Heesterman, J. C. (1984), "The Ritualist's Problem," in: Joshi, S. D., ed.
167-179
Heesterman, J. C. (1985), The Inner Conflict of Tradition: Essays in Indian
Ritual, Kingship and Society, Chicago
Hegel, G. W. F. (1985), Lectures on the Philosophy of Religion, III: The
Consummate Religion, transi. R. F. Brown, P. C. Hodgson and J. M.
Stewart with the assistance of H. S. Harris, Berkeley
Heidegger, Martin (1953), Sein und Zeit, Tubingen
Heinemann, Fritz (1929), Neue Wege der Philosophie: GeistILebenlExistenz.
Eine Einführung in die Philosophie der Gegenwart, Leipzig
Helms, L. V. (1882), Pioneering in the Far East and Journeys to California in
1849 and to the White Sea in 1848, London
Hill, Jane H. (1972), "On the Evolutionary Foundations of Language,"
American Anthropologist 74:308-317
Hillebrandt, A. (1880), Das altindische Neu- und Vollmondsopfer in seiner
einfachsten Form, Strassburg
Hillebrandt, A. (1897), Ritual-Literatur, Vedische Opfer und Zauber, Strass-
burg
Hinde, Robert A. (1982), Ethology. Its Nature and Relations with Other
Sciences, New York
Hockett, Charles F. (1968), The State of the Art, The Hague
Hoogt, J. M. van der (1929), The Vedic Chant studied in its Textual and
Melodic Form, Wageningen
468 Bibliography

Hooykaas, C. (1964), Ägama Tlrtha: Five Studies in Hindu-Balinese Religion,

Amsterdam
Hooykaas, C. ( 1966), Silrya-Sevana: The Way to Cod of a Balinese Siva Priest,
Amsterdam
Hooykaas, C. (1973), Balinese Bauddha Brahmins. Amsterdam
Hooykaas, C. (1976), ''Social Anthropology, A Discipline of Theories and
Hear-Say? (A Propos of Geertz on Bali)" Archipel 11:237-243
Hooykaas, C. (1983a), "Agni Offerings in Java and Bali," in: AGNI 11:382-402
Hooykaas, C. (1983b), "Homa in India and Bali," in: Strickmann, ed.
11:512-591
Howard, Wayne (1977), Sämavedic Chant, New Haven and London
Howard, Wayne (1983), "The Music of Nambudiri Unexpressed Chant
(aniruktagana)," in: AGNI 11:311-342
Hubert, H. and M. Mauss (1899), "Essai sur la nature et la fonction du
sacrifice," L'année sociologique 2:29-138
Hughes, H. Stuart (1968), "Structure and Society," in: Hayes and Hayes
1970.22-46
Huizinga, J. (1919), Herfsttij der Middeleeuwen, Haarlem
Huizinga, J. (1938), Homo ludens: Proeve eener bepaling van het spelelement
der cultuur, Haarlem
Humboldt, Wilhelm von (1836), Über die Verschiedenheit des menschlichen
Sprachbaues und ihren Einßuss auf die geistliche Enfwickelung des Men-
schengeschlechts, Berlin
Husain, Yusuf (1929), L'Inde mystique au moyen age: hindous et musulmans,
Paris
Husserl, Edmund (1952), Ideen zu einer reinen Phänomenologie und phäno-
menologische Philosophie, II. in: Husserliana IV.
Huxley, J. (ed.), (1966) "A Discussion on Ritualization of Behaviour in
Animals and Man," Philosophical Transactions of the Royal Society of
London, Series B, No. 772, 251:247-526
Hymes. Dell led.) (1974), Studies in the History of Linguistics: Traditions and
Paradigms, Bloomington
Ikari, Yasuke (1975), "Ukhä to Mahâvïra," Journal of Indian and Buddhist
Studies 23:1057-1046
Ikari, Yasuke (1983), "Ritual Preparation of the Mahâvïra and UkhäPots," in:
AGNI 11:168-177
Imrnelmann, K. (1962), "Beitrage zureinen vergleichenden Biologie australi-
scher Prachtfinken," Zoologisches Jahrbuch 90:1-196
Ingalls, D. H. H. (1954), "The Comparison of Indian and Western Philoso-
phy," Journal of Oriental Research 22:1-11
Ingalls, D. H. H. (1960), "On the Study of the Past," Journal of the American
Oriental Society 80:191-197
 Bibliography 469

Jacob, François (1981), Le jeu des possibles. Essai sur la diversité du vivant,
Paris
Jain, Jyotirindra (1984), Painted Myths of Creation: Art and Ritual of an Indian
Tribe, New Delhi
Jaini, Padmanabh S. (1979), The Jaina Path of Purification. Berkeley-Los
Angeles-London
Jakobson, R. (1962), "Why 'Mama1 and 'Papa'?" Selected Writings, Jhc
Hague, 1:538-545
Jakobson, Roman and Halle, Morris (1960), Grundlagen der Sprache, Berlin
James, E. O. (1917), Primitive Ritual and Belief An Anthropological Essay,
London
Jay, P. C. (1968), Primates, New York
Jayatilleke, K. N. (1963), Early Buddhist Theory of Knowledge, London
Jest, Corneille (1964/1965), "Les Thakali," Etnographie 58/59:26:49
Jha, G. (transi.) (1933-1936), Shabara-Bhäsya, I-III, Baroda
Jha, G. (1942), Pûrva-Mïmcunsâ in its Sources, Benares
Jha, V. N. (1973), "Sâkalya and Pänini," Bulletin of the Deccan College
Research Institute 33:223-226
Jha V. N. (1975), "Iti-karana in the Rgveda-Padapätha," Bulletin of the
Deccan College Research Institute 35:49-54
Jha, V. N. (1976), "Stages in the Composition of the Rgveda-Padapätha."
Bulletin of the Deccan College Research Institute 35:47-50
Jong, J. W. de (1978), Review of Lamotte 1976, Toung Pao 44:168-173
Jong, J. W. de (I987), "Lamotte and the doctrine of non-self," Cahiers
d'Extrême-Asie 3:151-153
Joos, Georg (1934), Theoretical Physics, London
Joshi, S. D. (ed.) (1984), Amrtadhârâ: Professor R. N. Dandekar Felicitation
Volume, Delhi
Kane, P. V. (1941, 1962), History of Dharmascistra, Vol. II, Pt. II; Vol. V, Pt.
II
Kashikar, C. G. (1968a), A Survey of the Srauta Sûtras, Bombay
Kashikar, C. G. (1968b), "Presidential Address," Vedic Section, All-India
Oriental Conference, XXIVth Session, Varanasi
Kashikar, C. G. (1979), "Agnicayana: The Piling Up of svayamütrnnä,"
Annals of the Bhandarkar Oriental Research Institute 60:215-218
Kiparsky, Paul (1979), Pânini as a Variationist, Cambridge-London-Poona
Kiparsky, Paul (1982), Some Theoretical Problems in Pànini's Grammar,
Poona
Kiparsky, Paul and Staal, Frits (1969), "Syntactic and Semantic Relation* in
Pânini," Foundations of Language 5.83-117
Klimburg-Salter, D. E. ( 1982), The Silk Route and the Diamond Path: Esoteric
Buddhist Art on the Trans-Himalayan Trade Routes, Los Angeles
470 Bibliography

Kosambi, D. D. (1950), "On the Origin of Brahmin Gotras," Journal of the

Royal Asiatic Society, Bombay Branch 26:21-80
Koster, Jan (1983), "Karel, Darwin en Chomsky. Over taal en evolutie,"
Hollands Maandblad 3-21
Kraemer, H. (1921), Een Javaansche primbon uit de zestiende eeuw, Leiden
Kripke, Saul (1982), Wittgenstein on Rules and Private Language, Cambridge
Mass.
Kroodsma, Donald E. (1974), "Song Learning, Dialects, and Dispersal in the
Bewick's wren," Zeitschrift für Tierpsychologie 35:352-380
Kroodsma, Donald E. (1978), "Aspects of Learning in the Ontogeny of Bird
Song: Where, From Whom, When, How Many, Which, and How Accu-
rately?" in: Burghardt and Bekoff, 215-230
Kuhn, Thomas S. (1970), The Structure of Scientific Revolutions, Second
Edition, Chicago
Kunjunni Raja. K. (1963), Indian Theories of Meaning, Madras
Lakoff, George (1969), "Empiricism Without Facts," Foundations of Lan-
guage 5:118-127
Lakoff, Robin (1969), Review (indirectly) of Chomsky 1966, Language
45:343-364
Lamotte, Etienne (1944-1980), Le traité de la grande vertu de sagesse de
Nägärjuna (Mahâprajnâpâramitàsâstra), I-V, Louvain
Lamotte, Etienne (1958), Histoire du bouddhisme indien, Louvain
Lamotte, Etienne (1973), "Der Verfasser des Upadesa und seine Quellen,"
Nachrichten der Akademie der Wissenschaften in Göttingen, I. Philolo-
gisch-historische Klasse, 31-50
Lancaster, J. B. (1968), "Primate Communication Systems and the Emergence
of Human Language," in: Jay, ed. 439-457
Lauenstein, Diether (1943), Das Erwachen der Gottesmystik in Indien,
München
Leach, Edmund (1970), Lévi-Strauss, Glasgow
Leach, Edmund (1985), Review of Singer 1984, American Ethnologist
12:154-156
Leeuw, G. van der (1948), Wegen en Grenzen, Amsterdam
Lévi, Sylvain (1898, 1966), La doctrine du sacrifice dans les Brähmanas, Paris
Lévi-Strauss, Claude (1962), La pensée sauvage, Paris
Lévi-Strauss, Claude (1964), Mythologiques I: Le cru et le cuit, Paris. English
translation 1969, New York and Evanston
Lévi-Strauss, Claude (1966), The Savage Mind, Chicago
Lévi-Strauss, Claude (1968), Tristes tropiques, Paris
Lévi-Strauss, Claude (1971). L'homme nu, Paris
Levy, John and Staal, Frits (1968), The Four Vedas, LP Album, New York
Lieberson, Jonathan (1984), "Interpreting the Interpreter," The New York
Review of Books, March 15, 39-46
 Bibliography 471

Liebich, Bruno (1891), Pänini. Ein Beitrag zur Kenntnis der indischen Litera-
tur und Grammatik, Leipzig
Ligeti, L. (1978), Proceedings of the Csoma de Körös Memorial Symposium,
Budapest
Lorenz, Karl (1951), ''Über die Enstehung auslösender 'Zeremonien'," Die
Vogelwarte 16:9-13
Lorenz, Karl (1963), Das sogenannte Böse. Zur Naturgeschichte der Agres-
sion, Wien. English translation:
Lorenz, Karl (1966), On Aggression, London
Lorenz, Karl (1967), Evolution and Modification of Behavior, Chicago
Lüders, H. (1959), Varuna II: Varuna und das Rta, Göttingen
MacFarland, Pamela and Staal, Frits (1988), "Ritual Language Dismantled,"
in: Le rituel. Actes du Colloque du Centenaire de VEcole Pratique des
Hautes Etudes, Section des Sciences Religieuses
Mâche, François-Bernard (1983), Musique, Mythe, Nature; ou les Dauphins
d'Arion, Paris
Malamoud, Charles (1967), "Village et forêt dans l'idéologie de l'Inde
brahmanique," Archives européennes de sociologie 17:3-20
Malamoud, Charles (1977), Le svädhyäya: récitation personelle du Veda.
Taittirïya-Âranyaka Livre II, Paris
Malamoud, Charles (1983), "Exégèse de rites, exégèse de textes," Purusârtha
7:17-38
Malinowski, B. (1948), Magic, Science, and Religion, New York. Fragment in:
Waardenburg 1973 I. 546-559
Mann, Thomas (1948), Doktor Faustus: Das Leben des deutschen Tonsetzers
Adrian Lev erkühn, erzählt von einem Freunde, Stockholm
Maspero, Henri (1950), Le taoïsme, Paris
Mauss, Marcel (1969), Oeuvres, III, Paris
May, L. C. (1956), "A Survey of Glossolalia and Related Phenomena in
Non-Christian Religions," American Anthropologist 58:75-96
McAllister, David P. (1949), Peyote Music, New York
Merlan, Philip (1953), From Platonism to Neoplatonism, The Hague
Michael, Tara (1986), "Le Srï-Cakra dans la Saundarya-Laharï, in: Padoux,
éd. 127-137
Miller, G. A. (1964), "The Psycholinguists," Encounter 23:29-37
Miller, G. A. and Chomsky, N. (1963), Handbook of Mathematical Psychol-
ogy, II, New York and London
Monod, Jacques (1970), Le hasard et la nécessité. Essai sur la philosophie
naturelle de la biologie moderne, Paris
Montague, Richard (1968), "Pragmatics," reprinted in: Thomason, ed.,
1974:95-118
Montague, Richard (1970), "Pragmatics and Intensional Logic," reprinted in:
Thomason, ed., 1974:119-147
472 Bibliography

Moore, S. F. and Myerhoff, B. G. (eds.). Secular Ritual, Amsterdam

Morris, Desmond (1957), " 'Typical Intensity' and its Relation to the Problem
of Realization," Behavior 11:1-12
Muktananda Paramahansa. Swami (1974), The Play of Consciousness (Chit-
shakti Vilas). California
Muller, F. Max (1867). Chips from a German Workshop I: Essays on the
Science of Religion. London
Munson. T. N. (1962), "Wittgenstein's Phenomenology," Philosophy and
Phenomenological Research 23:37-50
Murti. T. R. V, (1955). The Central Philosophy of Buddhism. London
Mylius, Klaus (1976), "Die vedischen Identifikationen am Beispiel des
Kausftaki-Brähmana," Kilo. Beitrage zur alten Geschichte 58/1:145-166
Mylius, Klaus (1977). "Potr. potra. Charakteristik eines vedischen Opfer-
priesteramtes. " Ethnographisch-Archäologische Zeitschrift 18:219-232
Mylius, Klaus (1982), "Acchâvakïya und Potra: Vergleich zweier vedischer
Opferpriestamter." Alt-Orientalische Forschungen 9:115-131
Mylius. Klaus (1986). "Acchävakiya und potra—ein Vergleich." Sanskrit and
World Culture. Berlin: 471-474
Naess, Arne (1965), Four Modern Philosophers: Carnap, Wittgenstein, Hei-
degger. Sartre. Chicago
Nagatomi. M.. Matilal, B. K.. Masson. J. M. and Dimock, E. (eds.) (1980),
Sanskrit and Indian Studies. Essays in Honor of Daniel H. H. In ga I Is.
Dordrecht
Nakamura. Hajime (1964). "A Critical Survey of Mahäyäna and Esoteric
Buddhism," Àcta Asiatica 6:57-88, 7:36-94
Nakamura, Hajime (1977), 'The Problem of Self in Buddhist Philosophy," in:
Coward and Sivaraman, eds., 99-118
Nakamura, Hajime (1980), Indian Buddhism: A Survey with Bibliographical
Notes. Tokyo
Needham, Joseph (1954-1976), Science and Civilisation in China. I-V:3.
Cambridge
Nettl, B. (1953), "Observations on Meaningless Peyote Song Texts," Journal
of American Folklore 66:161-164
Neugebauer, O. (1957), The Exact Sciences in Antiquity, Providence
Nietzsche, Friedrich (1871). Die Gehurt der Tragödie aus dem Geiste der
Musik, Basel
Nieuwenhuijze, C. A. O. van (1945). Samsu'l-Dfn van Pasai: Bijdrage tot de
kennis der Sumatraansche mystiek. Leiden
Nottebohm. Fernando (1970). "Ontogeny of Bird Song." Science'\67/2:950-
956
Nyberg, H. S. (1953). "Das Studium des Orients und die europäische Kultur."
Zeitschrift der deutschen morgenländischen Gesellschaft 103:9-21
Obeyesekere. Gananath (1981), Medusa1 s Hair: An Essay on Personal Sym-
bols and Religious Experience, Chicago
 B ibliograp hy Al 3

Obeyesekere, Gananath (1984), The Cult of the Goddess Pattini, Chicago

OTlaherty, Wendy Doniger (1981), The Rig Veda. An Anthology, Harmonds-
worth (Penguin Books)
Ojihara, Y. (1978), "Sur une formule Patanjalienne: na cedanîm âcùryah
sûtrâni krtvâ nivartayanti" Indologica Taitrinensia 6:9-234
Oldenberg, H. (1915). Die Lehre der Upanisaden und die Anfänge des
Buddhismus, Gottingen
Otto, Rudolf (1930), Indiens Gnadenreligion und das Christentum, Gotha
Padoux, Andre (1963), Recherches sur la symbolique et l'énergie de la parole
dans certains textes tan triques, Paris
Padoux, A. (1978), "Contributions a l'étude du mantrasâstra." Bulletin de
récole française d'extrême-orient 65:65-85
Padoux, A. (éd.) (1986). Mantras et diagrammes rituels dans r hindouisme,
Paris
Padoux, A. (1988), "MANTRAS—What are they?", in: Alper, ed. 295-318
Paranjpe, V. G. (1922). Le Vârtika de Kâtyàyana. Une étude du style, du
vocabulaire et des postulats philosophiques, Paris
Parpola, Asko (1981), "On the primary meaning and etymology of the sacred
syllable OM,\' Proceedings of the Nordic South Asia Conference Held in
Helsinki, June 10-12.
Parpola. Asko (1983). "Jaiminïya Srauta Sütra on the Agnicayana, with
Bhävaträta's Commentary/' in: AGNI 11:700-736
Parret, Herman (ed.) ( 1976). History of Linguistic Thought and Contemporary
Linguistics. Berlin
Payne, Richard K. (1986), Feeding the Gods: The Shingon Fire Ritual,
Delhi : International Academy of Indian Culture
Penner, Hans H. (1985), "Language, Ritual and Meaning," Numen. Interna-
tional Journal of the History of Religions 32:1-16
Philp, Mark (1986), "Michel Foucault," in: Skinner, ed., 65-81
Piston, W. (1947), Counterpoint, New York
Piston, W. (1962), Harmony, New York
Plotin (1924—1954), Ennéades, texte établit et traduit par E. Bréhier, I-VI, 2,
Paris
Pott, P. H. (1966), Yoga and Yantra: Their Interrelations and Significance for
Indian Archeology, The Hagne
Porphyre (1924), La vie de Plotin et l'ordre de ses écrits, texte établit et traduit
par E. Bréhier in: Ennéades, I, Paris
Preisendanz, K. (1973-1974), Papyri Graeci Magicae. Die griechischen Zau-
berpapyri, herausg. A. Henricks, Stuttgart
Proclus (1873), In P ri mum Euclidis Elementorum librum commentant, ed. G.
Friedlein, Leipzig
Quine, W. V. (1953), From a Logical Point of View, Cambridge, Mass.
Quine, W. V. (1960), Word and Object, Cambridge, Mass.
474 Bibliography

Quine, W..V. (1981), Theories and Things, Cambridge, Mass.

Renou, Louis (1931), Bibliographie védique, Paris
Renou, Louis (1941-1942), "Les connexions entre le rituel et la grammaire en
sanskrit," Journal asiatique 233:105-165, reprinted in: Staal 1973:434-469
Renou, Louis (1942, 1957), Terminologie grammaticale du sanskrit, Paris
Renou, Louisr (1947), Les écoles védiques et la formation du Véda, Paris
Renou, Louis (1949), "La valeur du silence dans le culte védique," Journal of
the American Oriental Society 69:11-18
Renou, Louis (1953), Religions of Ancient India, London
Renou, Louis (1954), Vocabulaire du rituel védique, Paris
Renou, Louis (1956), Histoire de la langue sanskrite, Pans
Renou, Louis (1960), Le destin du Véda dans l'Inde. Etudes védiques et
pâninéennes, VI, Paris
Renou, Louis (1963), "Sur le genre du sûtra dans la littérature sanskrite,"
Journal asiatique 251:165-216
Renou, Louis (1978), LInde fondamentale, ed. C. Malamoud, Paris
Renou, Louis and Filliozat, Jean (1953), L'Inde classique, II, Paris
Renou, Louis and L. Silburn (1954), "Nirukta and ânirukta in Vedic," Sarûpa
Bhâratî, Lakshman Sarup Memorial Volume, Hoshiarpur
Revzin, I. (1974), "From Animal Communication to Human Speech," in:
Cherry, 15-26
Rickman, H. P. (1967), "Wilhem Dilthey," in: Edwards, ed., 2:403-407
Rickrnan, H. P. (1979), Wilhelm Dilthey: Pioneer of the Human Studies,
Berkeley
Rinkes, D. A. (1909), Abdoerraoef van Singkel: Bijdrage tot de kennis van de
mystiek op Sumatra en Java, Heerenveen
Ripley, Sidney Dillon (1982), A Synopsis of the Birds of India and Pakistan,
Bombay
Robin, Leon (1908), La théorie platonicienne des Idées et des Nombres
d'après Aristote, Paris
Robins, R. H. (1976), "Some Continuities and Discontinuities in the History of
Linguistics," in: Parret (ed.) 31-31
Rorty, Richard (1979), Philosophy and the Mirror of Nature, Princeton
Ruegg, D. S. (1969), La théorie du Tathägatagarbha et du Gotra, Paris
Ruegg, D. S. (1973), Le traité du Tathägatagarbha de Bu ston Rin chen grub,
Paris
Ruegg, D. S. (1978), "Mathematical and Linguistic Models in Indian Thought:
The Case of Zero and Sûnyatâ," Wiener Zeitschrift für die Kunde Südasiens
22:171-181
Ruegg, D. S. (1981), "On the Supramundane and the Divine in Buddhism," in:
Katz, N. (ed.), Buddhist and Western Philosophy, New Delhi, 421-424
Ruffié, Jacques (1983), De la biologie à la culture, Paris, I—II
Russell, Bertrand (1953), Mysticism and Logic, London
 Bibliography 475

Ruweî, Nicolas (1966), "Méthodes d'analyse en musicologie," Revue belge de

musicologie 20:65-90. Reprinted in: Ruwet 1972.100-134
Ruwet, Nicholas (1972), Langage, musique, poésie, Paris
Salmon, W. (1969), Review of Chomsky 1966, Journal of Linguistics 5:165-187
Sarma, Neelakanta (1972), Textes sanskrits et tamouls de Thaïlande, Pon-
dichéry
Sarma, E. R. Sreekrishna (1983), "Kausïtaki Brähmana on the Atirätra," in:
AGNÏ 11:676-699
Sartre, Jean-Paul (1943), L'être et le néant, Paris. English translation: Being
and Nothingness, New York 1956
Sartre, Jean-Paul (1960), Critique de la raison dialectique, Paris
Sartre, Jean-Paul (1963), Les mots, Paris
Scharfstein, Ben-Ami (forthcoming), Like the Song of a Bird: On Art and the
Esthetic Universal
Schechner, R. (1986), Review of AGNI, Journal of Asian Studies 45:355-363
Schechner, R. (1987), "A 'Vedic Ritual' in Quotation Marks," Journal of
Asian Studies 46:108-110
Schipper, Kristofer (1982), Le corps taoïste, Paris
Schipper, Kristofer and Frits Staal (1988), "Vedic and Taoist Ritual," in:
Strickmann, ed. 1988
Schipper, Kristofer and Frits Staal (forthcoming), Ritual Structure
Schipper, Kristofer and Wang Hsiu-huei (1986), "Progressive and Regressive
Time Cycles in Taoist Ritual," in: Time, Science and Society in China and
the West, Amherst: 185-205
Schoenberg, Arnold (1970), Fundamentals of Musical Composition, New York
Schopenhauer, Arthur (1818), Die Welt als Wille und Vorstellung, Dresden
Schopf, T. J. M. (ed.) (1972), Models in Paleobiology, New York
Schrieke, B. O. J. (1916), Het boek van Bonang, Leiden
Schwab, J. (1886), Das altindische Thieropfer, Erlangen
Searle, John R. (1969), Speech Acts. An Essay in the Philosophy ofLâfiguage,
Cambridge
Sebeok, Thomas A. (ed.) (1966), Portraits of Linguists, I—II, Bloomington
Sebeok, Thomas A. (ed.) (1977), How Animals Communicate, Bloomington
Seidenberg, À. (1962), "The Ritual Origin of Geometry," Archive for History
of Exact Sciences, 1:488-527
Seidenberg, A. (1972), "On the Area of a Semi-Circle," Archive for History of
Exact Sciences, 9:171-211
Seidenberg, A. (1978), "The Origin of Mathematics," Archive for History of
Exact Sciences, 18:301-342
Seidenberg, A. (1983), "The Geometry of the Vedic Rituals," in: AGNI
11:95-126
Silburn, L. (1955), Instant et cause. Le discontinu dans la pensée philosophi-
que de l'Inde, Paris
476 Bibliography

Simon, R. (1908), Das Puspasütra mit Einleitung und Übersetzung, München

Singer, Milton (1984), Man\s Glassy Essence: Explorations in Semiotic
Anthropology, Bloomington
Skinner, Quentin (1986), The Return of Grand Theory in the Human Sciences,
London
Skorupski, Tadeusz (1983), "Tibetan Homa Rites," in: AGNI 11:403-417
Smart, J. J. C. (1965), 'The Methods of Ethics and the Methods of Science/'
Journal of Philosophy 62:344-349
Smith, Brian K. (1986), "The Unity of Ritual. The Place of the Domestic
Sacrifice in Vedic Ritualism," Indo-lranian Journal 29!79-96
Smith, Jonathan Z. (1978), Map Is Not Territory: Studies in the History of
Religion, Leiden
Smith, Jonathan Z. (1987), see: Hamerton-Kelly (ed.)
Smith, W. Robertson (1889), Lectures on the Religion of the Semites, London
Smith, F. and Miller, G. A. (eds.) (1966), The genesis of Language: A
Psycholinguistic Approach, Cambridge, Mass.
Snellgrove. David (1959), The Hevajra Tantra, I-II, London
Spiegelberg, H. (1965), The Phénoménologie al Movement. A Historical Intro-
duction, Princeton, I-II
Spiro, Melford E. (1970, 1982), Buddhism and Society: A Great Tradition and
its Burmese Vicissitudes, Berkeley
Sreekrishna Sarma, E. R. (ed.) (1968-1976), Kausitaki Brähmana: Text und
Vyäkhyä of V day a, Wiesbaden, I—III
Sreekrishna Sarma, E. R. (1983), "Kausitaki Brähmana on the Atirätra," in:
AGNI 11:676-699
Staal, J. Frits (1961), Namhudiri Veda Recitation, The Hague
Staal, J. Frits (1963a), Euclides en Pänini. Twee methodologische richtlijnen
voor de filosofie, Amsterdam
Staal, J. Frits (1963b), "Sanskrit and Sanskritization," The Journal of Asian
Studies 22:261-275. Reprinted 1972.
Staal, J. Frits (1965a). "Context-sensitive Rules in Pänini," Foundations of
Language 1:63-72
Staal, J. Frits (1965b), "Euclid and Pänini," Philosophy East and West
15:99-116
Staal, Frits (1966), "Analyticity," Foundations of Language 2:67-93
Staal, Frits (1967a), Word Order in Sanskrit and Universal Grammar, Dor-
drecht
Staal, Frits (1967b), "Review of Hans-Georg Gadamer's Wahrheit und Me-
thode: Grundzüge einer philosophischen Hermeneutik," Foundations of
Language 3:203-205
Staal, Frits (1968), "The Twelve Ritual Chants of the Nambudiri Agnistoma,"
Pratidânam: Indian, Iranian and Indo-European Studies Presented to F. B.
J. Kuiper on his Sixtieth Birthday, 409-429
 Bibliography All

Staal, Frits (1969, reprinted 1976), "Sanskrit Philosophy of Language,"

Current Trends in Linguistics 5:499-531
Staal, Frits (1970a), "Performatives and Token-Reflexives," Linguistic Inquiry
3:373-381
Staal, Frits (1970b), "Cambodia: Sanskrit Inscriptions," The New York
Review of Books July 2, 15
Staal, Frits (1971), "What was left of pragmatics in Jerusalem," Language
Science 14 (February): 29-32
Staal, Frits (1973), A Reader on the Sanskrit Grammarians, Cambridge-
London
Staal, Frits (1975a, 19882), Exploring Mysticism. A Methodological Essay,
Berkeley-Los Angeles-London; also Harmondsworth (Penguin Books)
Staal, Frits (1975b), "The Concept of Metalanguage and its Indian Back-
ground , ' ' Journal of Indian Philosophy 3:315-354
Staal, Frits (1978a), "The Ignorant Brahmin of the Agnicayana," Annals of the
Bhandarkar Oriental Research Institute, 59:337-348
Staal, Frits (1978b), "On and Around Yoga," Journal of Indian Philosophy
6:177-187
Staal, Frits (1979a), "Ritual Syntax," Sanskrit and Indian Studies: Essays in
Honor of Daniel H. H. In galls, Dordrecht 119-142
Staal, Frits (1979b), "The Meaninglessness of Ritual," Numen. International
Journal of the History of Religions 26:2-22
Staal, Frits (1979c), "Oriental Ideas on the Origin of Language," Journal of the
American Oriental Society 99:1-14
Staal, Frits (1980), "Rites That Make No Sense," The Communication of
Ideas: 10th International Congress of Anthropological and Ethnological
Sciences, eds. J. S. Yadava and V. Gautam, New Delhi, 3:145-154
Staal, Frits (1982a), The Science of Ritual, Poona
Staal, Frits (1982b), "The Himalayas and the Fall of Religion," in: Klimburg-
Salter, ed., 38-51
Staal, Frits (1983a), AGNI. The Vedic Ritual of the Eire Altar, I—II, Berkeley
Staal, Frits (1983b), "Moon Chants, Space Fillers and Flow of Milk," Surabhi:
Professor E'. R. Sreekrishna Sarma Felicitation Volume, Tirupati, 17-30
Staal, Frits (1984a), "The Search for Meaning: Mathematics, Music and
Ritual," American Journal of Semiotics 2:4.1-57 = "Ritmo nel rito,"
Conoscenza Religiosa 2:197-247
Staal, Frits (1984b), "Ritual, Mantras and the Origin of Language," in: Joshi,
S. D., ed. 403-425
Staal, Frits (1984c), "Abacadabra voor Intellectuelen," in: De brandende
kwestie, Amsterdam 2:151-177
Staal, Frits (1985a), "Mantras and Bird Songs," Journal of the American
Oriental Society 105:549-558
478 Bibliography

Staal, Frits (1985b), "Substitutions de paradigmes et religions d'Asie," Ca-

hiers d'extrême Asie, 1:21-57
Staal, Frits (1986a), "The Sound of Religion," Numen: International Journal
for the History of Religions 33.33-64, 185-224. French translation by
Michèle Garlati: "Le son de la religion," Maîtrise de Linguistique, Univer-
sité de Paris X, Nanterre
Staal, Frits (1986b), "The Fidelity of Oral Tradition and the Origins of
Science," Mededelingen der Koninklijke Nederlandse Akademie van We-
tenschappen, Afd. Letterkunde, n.r. 49/8:251-288. French translation by
Michèle Garlati: "La fidélité de la tradition orale et les origines de la
science," Maîtrise de Linguistique, Université de Paris X, Nanterre
Staal, Frits (1987), "Professor Schechner's Passion for Goats," The Journal of
Asian Studies 46:105-108
Staal, Frits (1988a), Universals: Essays in Indian Logic and Linguistics,
Chicago
Staal, Frits (1988b), Kailas: Center of Asia, Chicago
Staal, Frits (1988c), "Vedic Mantras," in: Understanding Mantras, ed. H. P.
Alper, New York, 48-95
Staal, Frits (1988d), "The Center of Space: Construction and Discovery,"
Inner and Outer Space, New Delhi
Staal, Frits (1988e), "An Outline of Vedic Ritual," in: Strickmann (ed.)
Staal, Frits (1989), Jouer avec le feu. Pratique et théorie du rituel védique,
Paris
Stebbins, G. Ledyard and F. J. Ayala (1985), "The Evolution of Darwinism,"
Scientific American 253:1.72-82
Steever, S. B. (1985), Review of Obeyesekere 1984, Journal of the American
Oriental Society 105:186-187
Steiner, George (1965), Language and Silence, New York
Steinitz, Paul (1978), Bach's Passions, New York
Steinkellner, E. (1978), "Remarks on Tantristic Hermeneutics," in: Ligeti,
ed., 445-458
Steklis, Horst B., Harnad, Stevan R. and Lancaster, Jane (eds) (1976), Origins
and Evolution of Language and Speech, Annals of the New York Academy
of Sciences 280:1-914
Strickmann, Michel (1981-1983), Tantric and Taoist Studies in Honour ofR. A.
Stein, Ï-II, Bruxelles
Strickmann, Michel (1982), "India in the Chinese Looking-Glass," in: Klim-
burg-Salter (ed.) 52-63
Strickmann, Michel (1983), "Homa in East Asia," in: AGNÎ 0:418-455
Strickmann, Michel (ed.) (1988) Classical Asian Rituals and the Theory of
Ritual, Berlin
Stutterheim, W (1931), "The Meaning of the Hindu-Javanese candi," Journal
of the American Oriental Society 51:1-23
 Bibliography 479

Tambiah, S. J. (1968a), "Literacy in a Buddhist Village in North-east Thai-

land/' in: Goody (ed.) 85-131
Tambiah, S. J. (1968b), "The Magical Power of Words," Man 32:175-208
Tambiah, S. J. (1979), "A Performative Approach to Ritual," Proceedings of
the British Academy 65:113-169
Tambiah, S. J. (1984), The Buddhist saints of the forest and the cult of amulets,
Cambridge
Taylor, Charles (1984), Philosophical Papers, Cambridge, I—II
Thieteke, Gerhard (1964), "Zur Phylogenese einiger Lautäusserungen der
europäischen Baumläufer (Certhia brachydactyla Brehm und Certhia fami-
liaris L.)," Zeitschrift für zoologische Systematik und Evolutionsforschung
2:383-413
Thieme, Paul (1931), "Grammatik und Sprache, ein Problem der altindischen
Sprachwissenschaft," Zeitschrift für Indologie und Iranistik 8:23-32
Thieme, Paul (1935), Pänini and the Veda, Allahabad
Thieme, Paul (1957), "Vorzarathustrisches bei den Zarathustriern und bei
Zarathustra," Zeitschrift der deutschen morgenländischen Gesellschaft
107:67-104
Thieme, Paul (1982-1983), "Meaning and Form of the 'Grammar' of Pänini,"
Studien zur Indologie und Iranistik 8/9:3-34
Thomason, R. H. (ed.) (1974), Formal Philosophy: Selected Papers of Richard
Montague, New Haven-London
Thorpe W. H. (1951), "The definition of terms used in animal behaviour,"
Bulletin of Animal Behaviour 9:34-40
Toulmin, Stephen (1982), The Return to Cosmology: Postmodern Science and
the Theology of Nature, Berkeley
Tsuji, N. (1952), On the Relation Between Brâhmanas and Srautasütras,
Tokyo (in Japanese, with English summary)
Turner, Victor (1975), Revelation and Divination in Ndembu Ritual, Ithaca and
London
Varenne, Jean (1962), "Mantra védiques dans le Raurava-Àgama," Journal
asiatique 250:185-189
Von Schroeder, L. (1887), Indiens Literatur und Cultur in historischer En-
twicklung, Leipzig
Waardenburg, Jacques (1974), Classical Approaches to the Study of Religion,
I—II, The Hague-Paris
Wasson, R. Gordon (1968), Soma: Divine Mushroom of Immortality, New
York.
Waismann, F. (1965), "Notes on Talks with Witgenstein," The Philosophical
Review 74:12-16
Wayman, A. (1976), "The Significance of Mantras, from the Veda down to
Buddhist Tantric Practice," Indologica Taurinensia 3:483^497
Weber, Max (1922), Religionssoziologie, Tubingen. English translation 1963.
480 Bibliography

Weir, R. (1970), Language in the Crib, The Hague

Westergaard, N. (1841), Radices linguae sanscritae, Copenhagen
Wheelock, W. T. (1982), "The Problem of Ritual Language: From Information
to Situation, "The Journal of the American Academy of Religion 50:49-71
Wheelock, W. T. (1985), "Patterns of Mantra Use in Vedic Ritual,11 Numen.
International Review for the History of Religions 32:169-193
Wheelock, W. T. (1988), 'The Mantra in Vedic and Tantric Ritual,'1 in Alper,
ed. 96-122
Whitehead, A. N. (1942), Adventures of Ideas, Harmondsworth
Whitehead, A. N. and Russell, B. (1925), Principia Mathematica, I, Cam-
bridge
Whitney, W. D. (1879, 1889), Sanskrit Grammar, Leipzig
Whitney, W. D. (éd., transi.) (1871), "The Taittirfya-Präticäkhya, with its
Commentary, the Tribhâsyaratna,1' Journal of the American Oriental Soci-
ety, 9:1-469
Whitney, W. D. (1872), "Steinthal and the psychological theory of language,"
North American Review, 114. Reprinted in Oriental and Linguistic Studies,
New York 1874
Wittgenstein, Ludwig (1961), Tractatus Logico-Philosophicus, with transi, by
D. F. Pears and B. F. McGuinness, London
Wittgenstein, Ludwig (1958), Philosophical Investigations, with transi, by
G. E. M. Anscombe, Oxford
Wittgenstein, Ludwig (1966), Lectures and Conversations on Aesthetics,
Psychology and Religious Belief, ed. C. Barrett, Oxford
Ziegenbalg, B., see: Caland, ed. 1926
Zimmer, H. (1926), Kunstform und Yoga im indischen Kultbild, Berlin
Zimmer, K. H. (1968), Review of Chomsky 1966, International Journal of
American Linguistics, 34:290-303
 I N D E X O F N A M E S

Aarsieff, H., 47 Atïkeliya, 325-326

Abhidharmakosa, 388 Anselm of Canterbury, 418
Acchäväka, 82, 244-245 Anscombe, G. E. M., 56
Adams, J. C , 10 Anuyàja, 87
Adhvara, 97 Âpasîcunba S raina Su tra, 193, 195,
Adhvaryu, 72-74, 79-83, 86, 91, 97, 242, 355-356, 358, 363, 419
146-149, 179, 201-203, 207, 210, Apel, W., 168, 176-177
212, 215-217, 232, 241, 244, 269, Appolonius Dyscolus, 33
303, 305, 322, 330, 356-357 Apratiratha, 136
Adhyavasâna, 72, 86-87 Aquinas, Thomas, 7, 29
Advaiîa Vedänta, 122 Àranyakas, 315, 352
Agni, passim Aristotle, 7, 8, 17, 20, 26-31, 35,
Agnicayana, passim 352, 443, 445, 456
Agnidh = Âgnïdhra, 79-80 Arnold, Harold, xviii, 303, 358
Agnidïksanîyestj, 96, 99 Arunta, 127
Agnihotra, 66, 77, 122, 355, 457 Asoka, 411
Agniksetra, 92-93, 209, 214 Astädhyäyi, see Pänini
Agnipranayana, 97, 132, 136-137 Àsvalâyana, 231-232
Agnïsotnîya, 75 Äsvaläyana Grhya Sütra, 231
Agnïsomïyapasu, 101 Asvins, 73, 93-94, 203, 230, 378
Agnistoma, 68, 80, 91-94, 96-97, Atharvaveda, 65, 199-200, 210,
99, 101, 102, 121, 180, 305-306, 253, 373
355 Atharvaveda Prütisäkhya, 50
Agnyupasthâna, 196 Atirätra-Agnicayana, passim
Àhâva, 87 Àtithyesti, 74
Âhavana, 229 Àtmch 118
Aitareya Àranyaka, 371 Attinello, Paul, xviii, 275
Aitareya Brähmana, 196 Austin, J. L., 22-23, 161-162,
Âjyabhâga, 87 237-242, 255-256, 342-343,
Alper, H. P., xvii, 228, 274,427 387-388
Ames, Michael M., xvii Avab/irtha, 11, 86, 88, 97
Anattalakkhana Sutta, 412 AvabhrthestL 101-102
Aniruktagâna, 113 Ayala, F. J., 346
482 Index

Ayer, A. J., 15, 25 Brähmanas, 66, 117-120, 124, 141,

155, 163, 192, 202-203, 223,
Bach, Johann Sebastian, 170-171, 234-235, 255-256, 282, 306, 352,
175-176, 185, 221, 224 361, 364-365, 380, 400, 407, 417,
Bake, Arnold A., 301 457
Ballabh, Ram, 350 Brahms, Johannes, 173
Baptista, Luis, 289 Breda, R. P. van, 14-15
Barbour, Ian, 394 Bréhier, Emile, 455-456
Barlow, George W., 112, 261, 388, Brhad Âranyaka Upa nis ad,
434 231-232, 407
Baudelaire, Charles, 187 Broch, Herman, 187
Baudhâyana, 95, 194, 361, 380 Brough, John, 186, 203, 292
Baudhäyana Srauia Sütra, 91-92, Brown, W. Norman, 113
193-194, 199, 353-355 Brückner, Anton, 173
Beck, Leslie, 14 Brunner, Hélène, 228-231
Beethoven, Ludwig van, 168-169, Buddha, 19, 236, 406-416
173, 175-177 Buddhâbhiseka, 317
Bender, Ernest, xvii Buddhaghosa, 319
Benfey, Theodor, 49 Budwig, Nancy, 272
Berlioz, Hector, 173 Buiskool, H. E., 362
Beth, Evert, 64, 452-453 Burkert, Walter, 70, 134, 261,
Bethel, Tom, 290 429-430, 434-435, 449-450
Bhagavad Gïtâ, 122, 133, 398 Burnell, A. C , 301
Bhandarkar, Ramkrishna Gopal, 49 Bursill-Hall, G. L., 35
Bharati, Agehananda, 305
Bhattacharya, Kamaleswar, 407, Caillois, Roger, 133
409, 412-414 Caland, Willem, 82, 101, 104, 108,
Bloch, B., 52 149, 202, 227, 232,276, 301-302,
Bloomfield, Leonard, 11, 151, 379, 306, 353, 355, 380, 417, 434
440 Candrakïrti, 235
Bloomfield, Maurice, 203, 227 Cantor, Georg, 14 .
Boas, Franz, 313 Cardona, George, 362
Boehtlingk, Otto, 380 Carlyle, Thomas, 427
Bokenkamp, Stephen R., 193 Carnap, Rudolf, 15, 20, 25, 27-28,
Bollée, W. B., 394 243
Boole, George, 442-443 Cassirer, Ernst, 125
Bopp, Franz, 36, 47-48, 442, Chändogya Upanisad, 224, 293
444-445 Chaudhuri, N. C , 117
Borges, J. L., 445 Cherumukku Vaidikan Vasudevan
Boulez, Pierre, 189 Akkitiripad, xv, 68, 186
Bowra, Sir Maurice, 276 Cherumukku Vaidikan, see,
Bradley, F. H., 11 Somayajipad, C. V.
Brahman, 72, 79, 225, 356 Chihamba, 152
Brähmanäc'c ha ms in, 82, 244 Chola, 68
 Index 483

Chomsky, Noam, 12, 24-25, 42, Dumont. Louis, 315

47, 52-59, 108-109, 111, 146-148, Dumont, P.-E., 203
246-248, 256-258, 274, 291, 339, Duncan, Isadora, 116
353, 424, 428, 438-439, 447, 450 Dundes, Alan, 333
Church, Alonzo, 27 Durkheim, Emile, 11, 125-128,
Churchill, Winston, 395-396 130, 141, 148, 150-152, 182-183,
Coedès, G., 404 255, 285, 387-388, 398,401, 417
Comte, Auguste, 7 Durt, Hubert, xvii, xviii
Converse, H. S., 391 Dyscolus, Apollonius, 33
Conze, Edward, 399
Craig, William, 64 Egan, Mike, 162
Crane, Jocelyn, 112 Eggeling, Julius, 117-118, 268
Crawford-Seeger, Ruth, 275 Eibl-Eibesfeldt, I., 434-435
Cullmann, Oscar, 447-448 Einoo, S., 196
Curtius, Ernst Robert, 440 Einstein, Albert, 27, 30
Cuvier, Georges, 442 Ekcidasarudra, 330
Eldredge, Niles, 290, 429
Dammann, Ernst, 233, 450
Eliade, Mircea, 123-124, 232, 319,
Dandekar, R. N., xvii, 261
369
Darwin, Charles, 142, 289-290,
Elias, Norbert, 425-426
293,429,431-432
Euclid, 30, 33, 43, 450
Darsapürnamäsa, 101, 180 (see
Euler, Leonhard, 450
also: Full- and New-Moon
Ceremonies)
Davidson, Donald, 429-430, 441 Faddegon, Barend, 224, 227
Debussy, Claude, 169-170, 176, Fehling, Detlex, 434
188 Feyerabend, Paul, 30
Demiéville, Paul, 235-236 Filliozat, Jean, 316, 369, 371
Descartes, R., 9, 20, 25, 29, 47, 49 Findly, E. B., 225
Dhammapada, 398 Firth, Raymond, 143-152
Dharmasätra, 364 Fodor, J. A., 138
Diehl, C. G., 254, 262, 329 Foucault, Michel, 240, 422,
Dîksâ, 72, 86-87, 394 441-447
Dïksanïyesti, 96, 101 Frank, Joseph, 428
Dilthey, Wilhelm, 28-29, 119, Frauwallner, Erich, 413-414
421-425, 427, 429, 431 Frege, Gottlob, 21, 28, 256
Dimock, Jr., E. C , xvi Freud, Sigmund, 135, 259,
Dionysius Thrax, 33 325-326, 442-443
Donatus, 35 Full- and New-Moon Ceremonies,
Drewes, G. W. J., 327 69, 79, 87, 91-92, 97, 180
Dreyfus, Hubert L., 240, 442, Fung Yu-lan, 137-138
446-447
Duhem, Pierre, 456 Gadamer, Hans-Georg, 25, 28-31,
Du Marsais, C. C , 23 119, 197
484 Index

Galle, J. G., 10 Halle, Morris, xviii, 52, 55, 59,

Gardner, Martin, 277 114, 257, 267, 274, 450
Gardner, Robert, 68 Hamerton-Kelly, Robert G., 70,
Garge, D. V., 419 435
Gautam, V., xvi Hammel, E. A. and Simmons, W..
Gäyatra, Gäyatn, 206, 231-232, S., 21, 129
295, 300, 305-309,451 Hanumän, 228
Geertz, Clifford, 162, 313-314, Hara, Minoru, 434
326-345, 389 Harman, G., 47
Geertz, Hildred, 335-336 Harris, Z., 52, 256
Genesis, 292, 344, 436 Hartshorne, Charles, 284, 288
Gennep, A. van, 120, 123, 128, Haydn, Franz Joseph, 172-173
133-134, 141, 148, 182, 283,285 Heerma van Voss, M., xvi
Ghanapätha, 373 Heesterman, J. C , 86, 108, 110,
Ghurye, K. C , 371 119, 136, 150, 154, 184, 202,
Giesey, R. E., 345-346 314-315, 390,400,408,435,
Girard, René, 435 453-454
Goldman, Robert, xviii Hegel, G. W. F., 9, 10, 17-19, 28,
Gombrich, Richard, 318-319, 325 125, 427, 442
Gonda, J., 118, 148, 191-192, 194, Heidegger, Martin, 20, 25-30, 422,
196,204, 217, 283, 303,306, 424-426
353-355, 380, 415, 417 Heinemann, Fritz, 427-428
Goodman, Nelson, 8, 57 Helms, L. V., 331
Goody, Jack, 61-62, 284, 374, 376 Henry, V., 82, 104, 108, 202, 232,
Goris, R., 227, 329 306
Goudriaan, T., 229, 284 Heraclitus, 18 .
Gould, Stephen Jay, 264, 290-291, Herodotus, 8
428-429 Hubert, David, 14
Grader, C. J., 329 Hill, Jane H., 111-1Î2
Graham, A. C , 418 Hillebrandt, A., 88, 104, 149, 362,
Grapard, Allan H., 392 434
Grâvasiuî, 76, 245, 374-375 Hinde, Robert A., 434
Gray, J. E. B., 296 Hindemith, Paul, 174-175
Green, Douglas M., 170, 173 Hintikka, Jaakko, 441
Gren-Eklund, G., 224, 227 Hiuen Tsiang, 411-412
Grhya Sütras, 193-194, 353 Hockett, Charles, 52, 54, 438-439
Grice, Paul, 239, 256 Hölderlin, J. C. F., 29
Giihyasamâja, 235 Hoogt, J. M. van der, 227, 304-305
Gulik, R. H. van, 254 Hooykaas, C , 228-229, 261,
Güntz, Th., 270 329-330, 335-339, 391, 397, 449
Gupta, 68 'Hotâ, 72, 76, 79-82, 91, 94, 97,
Güttinger, Hans Rudolf, 285-286, 146-149, 195, 232, 244, 356,
289 377-378
 Index 485

Howard, Wayne, 178 Jayatilleke, K. N., 20

Hsiu-huei, Wang, 193 Jest, Corneille, 395
HsünTzu, 137-138 Jha, Ganganatha, 230
Hubert, H., 69, 86, 102, 127, 150, Jha, V. N., 300, 370
255-257, 366 Johnson, Samuel, 394
Huizinga, J., 133,425-427 Jones, Sir William, 35
Humboldt, Wilhelm von, 47-50, Jong, J. W. de, 414
151,267 Joshi, S. D., xvii, 358
Hume, David, 7, 20, 35, 52 Joshua, 57
Husain, Yusuf, 328 Joos, Georg, 12, 13
Husserl, Edmund, 14-15, 28-29, Juergensmeyer, Mark, xviii
422, 427
Huxley, Sir Julian, 112, 136 Kafka, Franz, 346
Kälanyäsa, 225-226, 230-232
Kamatari, 392
Ikari, Yasuke, 226, 269 Kamp, Hans, 243
Immelmann, K., 286 Kane, P. V., 196, 234
Indra, 76, 136-137, 180, 205, 230, Kant, Immanuel, 7, 133, 178, 418,
374, 385 431—432v 442
Ingalls, Daniel H. H., 417, 452 Kapuräla, 321-323, 325
Intichiuma, 127, 182 Kashikar, C. G., 149, 380, 430-431
Isard, Stephen, 249 Kätyäyana, 358, 361
ƒ*(/, 72, 74, 77, 79-81, 85, 96, 178, Kausika, 292
. 181-184, 193, 195, 354, 356-357 Kausitaki, 377
Itti Ravi Nambudiri, 68, 209, 301, Kausitaki Brähmana, 119
383 Kauthuma-Rânâyanïya, 211-212,
218, 298-300, 302, 304, 308
Jacob, François, 290, 431 Kautsa, 112, 234, 373-375, 390
Jaiminiya, 209, 298-308, 383 Keith, A. B., 268
JaiminJya Àranyageyagâna, 200, Kes, Willem, 183
209, 212-220, 266, 301, 304-305, Kielhorn, F., 40, 49, 51, 89
311 Kierkegaard, S. A., 29
JaiminJya Ärcikä, 209, 218, 308, Kiparsky, Paul, 24, 44, 55, 151,
311 297, 309, 358
Jaiminiya Grämageyagäna, 209, Koran, 339, 369, 393, 440
211,279,301,304,311 Kosambi, D. D., 292
Jain, Jyotirindra, 385 Koster, Jan, 290
Jaini, Padmanabh S., 223, 407 Kraemer, H., 327
Jakobson, Roman, 52, 255-257, Kramapätha, 80, 372
267, 274-275, 341-342, 424, 428, Kripke, Saul, 56
437, 450 Krishna, 122
James, E. O., 125 Kroodsma, Donald E., 291-292
James, William, 421 Ksiradhärch 217-220, 303, 330
Jatäpätha, 373 Kuhn, Thomas, 10, 431
486 Index

Kunjunni Raja, K., 20 Malamoud, Charles, 263, 307-308,

Kuroda, S. Y., 239 315
Malinowski, Bronislaw, 130-131,
Lakoff, George, 438 143, 151, 195, 449
Lakoff, Robin, 47 Mallarmé, Stéphane, 187, 441
Lamotte, Etienne, 234, 316, Mann, Thomas, 175
408-409, 411-414, 416 Martin, R. M., 243
Lancaster, J. B., 111 Maruts, 180
Landi, Steffano, 168 Marx, Karl, 259
Lao Tzu, x'ix Maspero, Henri, 369
Lauenstein, Diether, 415 Masson, J. M., xvi
Leach, Sir Edmund, 260, 436 Matilal, B. K., xvi
Leaf, Murray, xvii Mauss, Marcel, 69, 86, 102, 127,
Leeuw, G. van der, 451 150, 255-258, 320, 366
Leibniz, G. W., 9, 17, 20, 28, 443 May, L. C , 273
Leverkùhn, Adrian, 175 Mbatjalkatiumcu 127, 182
Lévi, Sylvain, 255 McAllister, David P., 276
Lévi-Strauss, C 110, 133, 135, Mengelberg, Willem, 183
143, 163, 187-189, 255-260, 417, Menil, Adelaide de, xv, 68
424, 428, 442, 450 Michael, Tara, 228
Levy,John, 373 Mill, John Stuart, 7
Lewis, David, 243 Miller, G. A., 138, 248-249
Lewisf Todd, 396 Mîmâmsch 102, 120-122, 155, 190,
Lieberson, Jonathan, 332, 334, 339 192, 194-195, 203, 230, 234, 241,
Liebich, B., 49, 51 364-365, 375, 390,419
Lorenz, Konrad Z., 112, 286-288, Modistae, 35, 439
428, 434-435 Monod, Jacques, 130, 431
Montague, Richard, 243
Morimoto, Rand, xviii
MacFarland, Pamela, xv, xviii, 186 Morris, C. W., 243
Mâche, François-Bernard, 281-282, Morris, Desmond, 286-287
288-289 Mozart, Wolfgang Amadeus, 168,
Macrobius, Ambrosius Thepdosius, 171-173,233,275,450
456 Muhammad, 328
Madhyandincu 309 Muktananda, Swami P., 392
Mahâbhâraia, 411 Müller, F. Max, 231, 258, 379-380,
Mahâbhâsya, see Patanjali 439
Mahâprajnâpâramiîasâstra, 408 Munson. T. N., 14
Maharishi Mahesh Yogi, 242 Murti, T. R. V., 235, 274, 408,
Mahâvedi, 93 413-414
Mahâvîra, 73 Mylius, Klaus, 119, 244-245
Mailrâvarumu 81-82, 97, 244
Maitrâyani Samhitâ, 269 Naess, Arne, 22
Majjhima-nikâva, 20 Nàgârjuna, 408
 Index 487

Nagatomi, M., xvi Patanjali, 40-41, 46, 89. 109, 250,

Naipaul, V. S., 117 360-364, 384, 444 '
Nakamura, Hajime. 319, 390. 411, Pattini, 320-325
413 Payne, Richard, K., 228, 254, 391
Nâradasïksch 299 Peano, G., 28
Nârâyanasvâmï Dïksita, R.. 212 Peirce. C. S., 243, 255-256
Ndembu, 152-153, 161 Penner, Hans H., 102, 255,
Needham, Joseph, 8-9, 349-350, 258-259
370, 392-393 Philp, Mark. 446
Nes ta, 186 Pingree, David, 350
Nettl,B.,276 Piston, W.. 174-176
Neugebauer, O.. 349 Plato, xiv, 7-8, 17, 20. 26. 29. 58,
Newton, Isaac, 28 456
Nietzsche, F., 29, 178, 441 Planck, Max. 15
Nieuwenhuijze, C. A. O. van, 327 Plotinus. 455-457
Nihnavcu xv, 74-75, 85, 93, 186 Plutarch, 346
Nirukta, 234, 375 Popper. Karl. 15
Nottebohm, Fernando, 287, 289, Porphyry, 457
292 Patch 245
Novalis, 114 Pott, P. H., 228
Nyberg. H. S., 415 Pradhânahoma, 87
Prajäpati, 225
Obeyesekere, Gananath, 313, Prapancasâra, 284
320-326, 343 Präs tot à, 97, 244, 299
PrätaraniiYcika* 76, 377
O1 Flaherty, Wendy Doniger, 454 Pratigara, 82, 232, 322
Ojihara, Yutaka. 361 Pratihartcu 244, 299, 307
Oldenberg, Hermann, 365, 413-415 Pratiprasthùîâ, 81-82, 217, 303,
Otto, Rudolf, 415, 357
Prâtisâkhya, 39-43, 50-52, 256,
Padapätha, 37-39, 50, 300, 370, 309, 372-373, 383
372-373,383 Pravargya, 73-75, 85-86. 93, 157
Padoux, André, 224-225, 228, 231, Prayâja, 87
253-254, 274, 304 Prayanïyestj, 73, 86-88, 93, 101
Pancavimsa Brähmana, 119 Preisendanz, K., 276
Pänini, 39-46, 49, 51, 53, 55, Priscian, 35
58-59, 151, 256-257, 297, 300, Proclus, 33-34
309, 352-353, 357, 360-363, 365, Ptolemy, 8
373, 379-382, 384, 444-445, 455 Purânas, 233-234
Paranjpe, V. G., 419 Puroruc. 82, 87, 118
Parpola, Asko, 209, 274, 301 Purusa, 212, 214
Parsons, Talcott, 129 Püsan, 203, 230
Pasubandhcu 101 Puspasùtra, 300, 30V
488 Index

Putnam, Hilary, 54 Sakhyavisarjana, 76-77, 86, 88

Salmon, W., 47
Quine, W. V. O., 12, 15, 23, 26, Sämaveda, 39, 65, 74, 76, 139, 178,
59, 102, 237, 375, 387-388, 441 185, 192, 207-221, 229, 232,
235-236, 242,282, 296-311,
Rabinow, Paul, 240, 442, 446-447 355-356, 383
Radcliffe-Brown, A. R., 152, 161, Samhitâpâtha* 37-39, 51, 372-373
324 Sâmidhertî, 91-92, 95, 106
Rahasya (gäna), 305 Samistayajus, 97
Räma, 233 Sahkara, 224, 363, 375, 419
Reichenbach, Hans, 243 Sânkhâyana Srauta Su tra, 92
Rämäyana, 411 Sarma, Nilakanta, 316, 474
Renou, Louis, 46, 65, 113, 118, Sartre, Jean-Paul, 259, 426,
•148,261, 263, 274, 298,302, 441-442
308, 317, 353, 359, 371, 380 Sassetti, Filippo, 36
Respighi, Ottorino, 177 Sastradoha, 82, 87
Revzin, I., 111 Satapaiha Brähmana, 117, 306
Rhys Davids, C. A. F., 314 Satarudriya - Rudram, 217, 303
Ricardo, David, 442 Satie, Eric, 166-167
Rickman, H. P., 421-422, 427 Satyâsâdha, 358
Rigveda, passim Satyavrata Samâsramin, 227
Rinkes, D. A., 327 Saussure, F. de, 51, 140, 196, 256,
Ripley, Sidney Dillon, 382 258
Rjuvimalâ, 122 Sautrâmanî, 196
Rkprätisäkhya, 373, 380, 383 Savanïyapasu, 101
Robins, R. H., 35 Savitr, 147, 203, 306
Roche, J. C. 280-281 Sâvitrï, 306-307
Rorty, Richard, 441 Scharfstein, Ben-Ami, 285,
Ross, J. R. rHaj"), 109 287-288, 290
Rtvigvarana, 241 Schärpe, E. J., xvi
Rubinstein, Arthur, 132 Schechner, Richard, 94, 250-251
Rudra, 75, 217, 330, 375, 390, see Schenker, H., 174
Rudram Schipper, Kristofer, 69, 140, 150,
Rudram •= Satarudriya, 217, 303 193,284, 328, 343, 369, 449
Ruegg, P. S., 235, 398, 452 Schlegel, August Wilhelm von,
Ruffié, Jacques, 428 47-48
Russell, Bertrand, 11, 23, 28, 30, Schleiermacher, Friedrich, 427
63, 243, 442-443 Schoenberg, Arnold, 165, 172, 176
Ruwet, Nicolas, 173 Schopenhauer, Arthur, 177-178
Ryle, Gilbert, 14, 340 Schubert, Franz, 172, 176
Schrieke, B. O. J , 327
Sabara, 203, 230 Schwab, J., 104, 149-150, 255
Sadasva, 72 Scott, Dana, 243
 Index 489

Seal, Brajendranath, 350 Steinitz, Paul, 170-171

Searle, John R., 161-163, 238-240, Steinkellner, E , 235
256 Steklis, Horst B , 429
Sebeok, Thomas, xvii, 50 Stobhänusamhära, 227, 253
Seidenberg, Abraham, 350 Strabo, 8
Seymour, T. D., 50 Strauss, Richard, 177
Shinran, 415 Stravinski, Igor, 189, 282
Shukla, Kripa Shankar, 350 Strickmann, Michel, 261, 317, 391,
Silburn, L., 113, 263, 317 396,404-405,449
Simon, Richard, 304 Stutterheim, Willem, 392, 404
Singer, Milton, xvii, 255-256, 260 Subrahmanyä, 74, 75, 93
Si va, 230, 233, 261 Subrahmanyähväna, 244
Skorupski, Tadeusz, 261, 391, 449 Svetäsvatara Upanisad, 415
Smart, J. J. C , 23 Svistakrt, 87. 356
Smetana, Bedrieh, 176 Swaminathan, C. R., 194
Smith, Brian, 365 Sweeney, Amin, 335
Smith, Frank, 138 Swellengrebel, J. L., 329
Smith, Jonathan, 143, 148, 346, Tachiki, Scott, xviii
394, 430-431, 438, 440, 449, 451 Taittirïya, 292, 309
Smith, W. Robertson, 125-126 Taittirïya Âranyaka, 225
Snellgrove, David, 399 Taiîîinya Brâhmana, 203, 354
Snow, C. P., 421 Taittirïya Pràtisâkhya, 50-51, 309
Socrates, xiv, 17 Taittirïya Samhitâ, 86, 92, 97, 136,
Soma, 66, 68-69, 73-76, 79, 81-83, 194, 199, 202, 210, 214-215, 217,
85,87,91,93, 116, 124, 159, 178, 269-270, 276, 303, 305, 354, 407
181-184, 193, 196, 202-207, Taittirïya Upanisad, 226 •
244-245, 295, 299, 305-307, 357, Tambiah, S. J., 161-163, 195,
376-377 238-240, 243, 255-256, 260, 274,
Somayajipad, C. V., 68 313-320, 332, 335, 339
Spiegelberg, H., 423, 427 Tänünaptra, 74, 87-88
Spiro, MelfordE., 314, 335 Tao Te Ching, xix
Spinoza, B., 20, 28, 423 Tarski, Alfred, 21, 23,445
Srautakosa, 104, 202 Taylor, Charles, 441
Srauta Sutras, 66, 101, 104-106, Teilhard de Chardin, Pierre, 431
108, 110, 121, 184, 193, 196,249, Thielcke, Gerhard, 286-287, 289
255-257, 3SV358, 360-366, 375, Thieme, Paul, 36, 59, 230, 358,
380,417,450,452 374,438-439
Sreekrishna Sarma, E. R., xvii, Thomason, R. H., 243
118 Thorpe, W. H., 112, 388,434
Stalnaker, R., 243 Thrax, Dionysius, 33
Stebbins, G. Ledyard, 346 Thucydides, 26
Steever, S. B., 325 Tikopia, 143-146, 148
Steiner, George, 187 Tinbergen, N., 112
490 Index

Tóulmin, Stephen, 431-432 Visvakarman, 97

Trabaci, Giovanni Maria, 168 Voltaire, 290
Trubetzkoy, N. S., 52 Von Schroeder, Leopold, 268-272
Tsuji, Naoshiro, 261, 417, 434
Turner, Victor, 120, 123, 152-154, Waardenburg, Jacques, 128-129,
163 131,258
Tyäga. 121-122, 179 Wagner, Richard, 34, 176-177, 187
* Waismann, F., 139
Waley, Arthur, xix
Udavasàna* 11, 86, 88 Wasson, R. Gordon, 319
UdavasümyestL 101 Weber, Albrecht, 149
UdayaniyestL 11, 86, 88, 101 Weber, Max. 125, 128-130, 143,
Udgätä, 72, 75, 81, 207, 209-220, 151, 319, 376, 385,422
224, 232, 244, 279-280, 299, 303, Weir. Ruth, 272-273
* 305, 307, 330, 356 Werblowsky. R. J. Z., xvi
tjhagämu Ühyagäna, 305-306, Wells, R. S.-, 52
308, 311 Westergaard. Niels. 48
Ukhch 71, 73, 95, 269, 270 Wheelock, W. T., 193, 223-225,
Ukthavîrya, 82, 87 238-240, 243, 274
Upanayana, 226-227, 231, 393-394 Whitehead, Alfred N., 9, 10, 28, 30
Upanisads, 120, 125, 315, 352, 364, Whitney, William Dwight, 49-52,
400-401,406-409,412 151,440
Upasad, 74-75, 85-86, 93, 157 Wiebe, Donald, xviii
Upasampadä, 321, 394 Wilde, Oscar, 395
Uttarädhyayana Sütra;401 Wilkins, Sir Charles, 36, 445
Wittgenstein, 14, 21-23, 55-58,
Väjasaneyi, 309 139, 231, 260, 427, 439, 453-454
Vallabhan, Nambudiri, C. V., xv, Yadava, J. S., xvi
186 Yajamäna, 66, 71-77, 79-82, 93,
Vallée-Poussin, Louis de la, 116, 121-122, 153-154, 179, 181,
413-414 183,201-202,210, 215, 244,279,
Varenne, Jean, 225 307, 316
Vasor dhärä, 75, 330 Yajurvedcu 65, 79, 83, 92, 146, 148,
Väsudeva, 233 180, 192, 194, 218, 298, 309-310,
Väyu, 72, 91-92,95, 215 355-356, 407
Vedänta, 67, 353, 363-364, 375, Yoga, 318-319, 369, 398, 390
408,419 Yogasütra, 388
Vena, 210 Yoruba, 110
Venus, 215
Verrier, U. J. J. Le, 10 Ziegenbalg, Bartholomaeus, xiv,
Vijayanagara, 68 227,275-276
Vimalakïrti, 392 Zimmer, H., 228
Visuddhimagga, 319 Zimmer, K. H., 47