GOYEENMENT OE INDIA
DEPARTMEDfT OF ARCHAEOLOG-Y
CENTRAL ARCHAEOLOGICAL LIBRARY
■GijAsr
AJiSuoa _ ,
2>l(^
Call No.
D.a.A. 79.
•' < -V '»»•»
\ 'f *5
-'•W;A .
AMERICAN ORIENTAL SERIES
VOLUME 3
THE PANCHATANTRA RECONSTRUCTED
VOLUME 2
AMERICAN ORIENTAft
VOLUME 3
COMMITTEE ON PUBLICATIONS
E. WASHBURN HOPKINS
Chairman
CHARLES a TORREY
and
EEIANKLIN EDGESTON MAX L. MAEQOLIS
StB cut BdUon ’of ifiA ^oumcd
PUBLISHED BY THE
^Itncritari <|)riEntal Jiodetp
NEW HAVEN, CONNECTICUT, U. 8. A.
19S4
THE PANCHATANTRA RECONSTRUCTED
AN ATTEMPT TO ESTABLISH THE LOST ORIGINAL, SANSKRIT TMST OP THE MOST FAMOUS OF INDIAN STORY- COLLECTIONS ON THE BASIS OF THE PRINCIPAL EXTANT VERSIONS
€rific#I ^jjareftts, ^nfrohatfintt, Sransfaffott
By franklin EDGERTON
Assistant Professor of Sanskrit in the University of Pennsylvania
VOLUME 2
? HUMPHREY MILFORD
OXFORD^ UNIVERSITY PRESS LONDON iDINBURGH GLASGOW TORONTO MELBOURNE CAPE TOWN BOMBAY CALCUTTA
BT ABOLF HOBZHAOSEN. VIENNA, AUSTRIA
k'
I
'■’i V fS’o
'S
m’
f%.
t
.0 i .
fw
r
V
CONTENTS OP VOLUME 11
iNTEOBUOTIOlSr
Page
Chapter I. Purpose and Results of this Book 3
The Paficatantra in world literature 3
The Paficatantra in India 3
Object of this book, contrasted with previous studies 4
What was the original Paficatantra? 4
Interest of this investigation 6
Method employed in the reconstruction 6
Primary results of this investigation 8
Incidental results of this investigation 9
Extent of divergence from Hertel’s results 10
Chapter 11. The Materials*^^ 12
Paficatantra versions uspd' in the reconstruction 12
The Tantrakhyayika (abbreviated T). . . .' 12
Extent to which the Tantrakhyayika preserves the original text . . 13
Secondary additions in the Tantrakhyayika 13
HertePs views of the Tantrakhyayika 14
The Tantrakhyayika has no privileged position among Paficatantra
versions 16
The Southern Paficatantra (abbreviated SP) 17
Extent to which the Southern Paficatantra preserves the original text 18
Secondary additions in the Southern Paficatantra 19
The Jfepalese Paficatantra (abbreviated N) 19
The Hitopade§a (abbreviated H): its origin 20
Greneral plan of the Hitopadefia 21
Extent to which the HitopadeiSa preserves the original text .... 22
Secondary additions in the HitopadeSa 22
The Paficatantra’s position in the ByhatkathS 28
Eifect of language and versification on the Brhatkatha versions . . 23
K§emendra (abbreviated K§) 24
Somadeva (abbreviated So) 26
The so-called textus simplicior *’ (abbreviated Spl) 27
VI
Pagre
General plan of Siraplicior
Extent to which Simplicior preserves the original text 29
Secondary additions in Simplicior 30
Peirp-abhadra (abbreviated Pn) . 30
General plan of Parpabhadra: his two main sources, TantrakhySyika
and Simplicior 31
The “ Ur-Tan trakhyayika,” source of the “ Ur-Simplicior ” and the
Tantrakhyayika . * 36
Piirnabhadra’s other source or sources 37
Value of Pflrpabhadra for the reconstruction 38
Extent to which Purnabhadra preserves the original text 39
Secondary additions in PQr^iabhadra 39
The Pahlavi translation (abbreviated Pa) 40
Immediate offshoots of the Pahlavi 41
Offshoots of the Pahlavi thru the Arabic ‘ 43
Use made of the Pahlavi versions in the present work 44
Extent to which the Pahlavi preserves the original text 46
Secondary additions in the Pahlavi 46
Table showing interrelations of older Paficatanfera versions 48
Chapter III. Methods employed in the Reconstruction 49
Purpose of this chapter . . * 49
Three ways of proving laecondary interrelationship 49
Versions which are not secondarily interrelated 62
How to determine original matter? 62
All versions point to a definite literary archetype 53
1. Features common to ^1 versions must he original 65
2. Omission of fctnres in Hitopadela and the Byha^atha versions
not significant 66
3: Very minor features common to a smaller number of independent
versions are not necessmtly original 56
4. More important features common to several independent versions; probability of originality tends to vary with importance and
closeness of correspondence 57
6. Entire stories common to several independent versions at the same
place are almost certainly original . . i 68
6. Summary of methods by which originality is determined .... 60
7. Features occurring only in a single stream of tradition ..... 60 Our methods are verified inductively and pragmatically, and are not
based cn mere abstract consideratioDs 62
Critique of HertePs method . . ; 64
Chapter IV. Secondary Interrelationships of Various Versions 68
Common archetype of the Old Syriac and the Arabic 68
Common archetype of Soraadeva and Ksemendra 69
The ‘"Ur-SP,” archetype of 8P, K, and H 69
The “ Ur-N, ” the pcondary archetype of N and H 69
K^emendra used a TantrSIkhylyika manuscript 70
Vll
Page
The “ Ur-Simplicior, ” source of our Simplicior, and one of the main
sources of PCLriiabhadra 70
Duplications in Pflrriabhadra, due to his use of two sources. ... 71
The “ Ur-Tan trSkhyayika,” archetype of Tantrakhyayika and the
“ Ur-Simplicior ” 72
Secondary stories inserted in “ Ur-Tan trSkhyayika” and found only
in its descendants ; . 73
1. The Blue Jackal. 74
2. Jackal outwits Camel and Lion 76
3. Weaver Somilaka 76
4. Talking Cave 77
6. Potter as Warrior 78
6. The Clever Hansa 79
7. Other stories which may possibly have been found in the
Ur-Tan trakhyayika 79
Verbal correspondences between TantrakhySyika and Simplicior and
Piirnabhadra 79
Clearly secondary correspondences in detail between Tantrakhyayika
and Simplicior (and Pdri^-abhadra) 80
1. Reconstruction 1 §§ 18 — 22, including vss 4, 6 . , . . , . . 80
2. Reconstruction I §§ 29, SO 84
3. Reconstruction III vs 99 ... 85
4. Reconstruction in § 64 86
6. Reconstruction III §§ 71, 72 ... 87
6. Reconstruction II § 233 87
7. Minor and miscellaneous agreements of T and Spl 88
Chapter V. Critique of HertePs Views of Interrelationship of Versions , 89
General remarks on Hertel’s views of the Paflcatantra versions , . 89
Points in HertePs genealogical table of versions which this chapter
will try to disprove . 90
HertePs proofs are insufficient even if they were individually sound 91
These theories are not only unproved but unprovable 92
The supposed archetype “ t ” 92
What is meant by this “t”? 92
1. prat^&yUo^ T A 149 ; ” Reconstruction II § 62 . 93
2. The verso T H. 87 ; Reconstruction II vs 63 , 94
3. hlhqjam.a^^ T p. 60, 1. 9; Reconstruction I§ 570 ....... 96
4. The laree-oracle, T p. 67, 1. 16fif.; Reconstruction I § 647 . . 97
6. The crocodile and the ape, T “ A 286; ” Reconstruction IV § 36 98
6. The verse T IH. 125; Reconstruction III vs 107 98
7. The verse T L 174; Reconstruction I vs 163 98
Summary and conclusion regarding “ t 99
The supposed archetype “ K ” . * . . 101
What is meant by the archetype K . 101
1. The ape and the crocodile, Book IV, frame; particularly T “A 286, ” Reconstruction IV § 36 . . 102
VUl
Page
% The verse T 11. 90 j Reconstruction n vs 55 105
8. The verse T II. 25; Reconstruction U vs 15 106
4- Hnskt or unhuskt sesame? Story II. 2 106
5. Other evidence for “ K ” in Hertel, Tantr. Einl. p. 31 , . . , 108
6. The verse T L 19; Reconstruction I vs 21 109
7. The verse T II. 61; Reconstruction II vs 36 110
8. End of Book IV Ill
9. The verse SP HI. 32; Reconstruction HI vs 44, and preceding
prose Ill
Summary and conclusion as to the “archetype K” 116
The supposed archetype “N-W” 117
What is meant by the supposed archetype “N-W”? 117
1. The sesame story again 117
2. Story of Brahman and Rogues, III. 5 118
Summary and conclusion 119
Relations of TantrSkhyayika a and p, and of the mss. of T .... 120
Hertel’s view that T a is more original than p 120
The present writer's views 120
Alleged interpolations in T p from a “ K ” codex 121
Minor variations in the language of T a and p 123
Supposed “attempted corrections”, in TP, of T« readings . . . 124
The manuscripts of TantrUchyayika 125
Summary and conclusion 126
Chapter VI. Examples of Method of Reconstruction: Original and
Unoriginal Agreements. . . ‘ 128
Purpose of ihis chapter 128
Reconstruction of Book I, %% 34 — 48, and vss 7 — 23 129
Original and unoriginal agreements 151
Unoriginal agreements between H and Pa 151
between H and Jif. 162
between SP and Jn 152
between P? and K? 152
between T and SP 152
Chapter VIL Examples of Method of Reconstruction, continued: Establish- ment of Original by Agreements of Other Texts than Tantra-
khylyika 154
Purpose of this chapter 154
Agreements of Ur-SP, Br, Jn, and Pa, against T 156
of Ur-SP, Jn, and Pa 159
of Ur-SP, Jn, and So or K§ 162
of Ur-SP, Pa, and So or K§, against T (and Jn) . . . 163
of Pa, Jn, and So or 164
of Ur-SP and both Jn versions, against T . , . . . . 166
of Ur-SP and Spl, against T (and Pn) 167
of Ur-SP and Pi^, against T (and Spl) 168
of Ur-SP and Pa 169
IX
Pago
Agreements of Ur-SP and So or 171
of Pa and Jn 171
of Pa and So or K§ 174
of Ju and So or 174
Other unoriginal features in Tantrakhyayika 175
Insertions in Tantrakhylyika 177
Chapter VIII. The Original Work as Eevealed by the Reconstruction. 181
Purpose of this chapter 181
Kame of the original work 181
Meaning of the name . . . . . 181
Date of the original work 182
Authorship of the original work 182
Home of the original work 183
Language of the original work 184
Character of the original work as a political textbook 186
Story-contents of the original: stories included by me but excluded
or doubted by Hertel . 186
Conspectus op Stories of the Original 189
Conspectus op Text-units op the Original 192
Chapter IX. Critical Notes on the Text of the Tantrakhylyika .... 259
Purpose of this chapter , . . . 269
Emendations in the text of Tantrakhylyika 269
Unfortunate emendations made by Hertel in the text of TantrS-
khyayika 260
Unfortunate choices made by Hertel between variant manuscript
readings in the text of Tantrakhyayika 263
Translation
Kathamukha or Introductory Section 271
First Book: The Separation of Friends, or, The Lion and the Bull , . 274
Story 1 : Ape and Wedge 277
Story 2: Jackal and Drum 284
Story 3 a: Monk and SMndler . , . , 288
Story 3 b : Earns and Jackal 288'
Story Sc: Cuckold Weaver and Bawd 289
Story 4: Crows and Serpent 294
Story 5: Heron and Crab 294
Story 6: Lion and Hare 296
Story 7 : Louse and Flea 602
Story 8: Lion’s Retainers and Camel #08
Story 9: Strandbirds and Sea 812
Story 10: Oeese and Tortoise #18
Story 11: Forethot, Ready- wit, and Come- what- will , #14
Story 12: Ape, Glow-worm, and Bird . #20
EdgeriiOii, Pafic&bantra IL . ^
X
Page
Story 18: Evil-wit and Honest-wit . . , . 822
Story H: Herons, Snake, and Mongoose 328
Story 15: Iron-eating Mice 325
Second Book: Tke Winning of Friends, or, The Dove, Crow, Mouse,
Tortoise, and Deer 329
Story 1: Mouse and Tw-o Monks 338
Story 2: Huskt for Hnskt Sesame. 339
Story 3: Too Greedy Jackal 340
Story 4: Deer’s Former Captivity 352
Tbird Book; War and Peace, or, The Crows and the Owls 358
Story 1; Ass in Panther’s Skin 364
Story 2; Birds Elect Bang 364
Story 8: Elephant, Hares, and Moon 365
Story 4: Cat, Partridge, and Hare 369
Story 5: Brahman and Eognes 372
Story 6: Old Man, Young Wife, and Thief 376
Story 7: Brahman, Thief, and Ogre 377
Story 8: Cuckold Carpenter 378
Story 9: Monse*Maiden 380
Story 10; Progs Bide Serpent 386
Pourth Book: The Loss of One’s Gettings, or, The Ape and the Crocodile 393
Story 1: Ass without Heart and Ears 398
Fifth Book: Hasty Action, or, The Brahman and the Mongoose . . , 401
Story 1; Brahman who Bnilt Air- castles 401
Story 2: Barker who BuHed the Monks 404
Addenda et Corrigenda 406
Eclgerton, Pancatontra. XL
1
CHAPTER I
PURPOSE AND RESULTS OF THIS BOOK
The Pancatantra in world literature. — No other work of Hindu literature has played so important a part in the literature of the world as the Sanskrit story-collection called the Pancatantra. Indeed, the statement has been made^ that no book except the Bible has enjoyed such an extensive circulation in the world as a whole. This may be — I think it probably is — an exaggeration. Yet perhaps it is easier to underestimate than to overestimate the spread of the Pancatantra. In Professor Johannes Herts’s book on the subject^ there are recorded over two hundred different versions known to exist in more than fifty languages; and about three-fourths of these languages are extra-Indian. As early as the eleventh century the work reacht Europe, and before 1600 it existed in Greek, Latin, Spanish, Italian, German, English^ Old Slavonic, Czech, ^ and perhaps other Slavonic lan- guages. Its range has extended from Java to Iceland.
The Pahcatantra in India. — Nor has this famous work been without honor in its own country. No other collection of stories has been so popular thruout the length and breadtli of India, It has been workt over again and again, expanded, abstracted, turned into verse, retold in prose, translated into medieval and modem vernaculars, and retranslated into Sanskrit. And most of the stories contained in it have “ gone down into the folk* lore of the story-loving Hindus, whence they reappear in tike
^ According to Winternitz, DLZ, SI (1910), 2698; not, however, with hh endorsement.
® Das Puiicaiavira^ ^eim GeschichU nnd smne Verhrmimig\ ILeipzig and Berlin, 1914. (Abbreviated! “ Hartel, See the Indices to this
I, p. 451f.
® In several of the languages named, a number of different versions existed at that early date.
4 Chapter I; Purpose and results of this book
collections of or^l tales gathered by onodern students of folk- stories.^
Object of this book, contrasted with previous studies.— It is not my purpose at present to trace the history of the Panca- tantra or its stories, as they appear in successive works of literature or in folklore. This eithe;r has been done, or is being done, by others.^' The task I am undertaking is rather the re- verse: to follow back the streams of PaScatantra tradition in the hope of finding their source. For my present purpose, the contents of the versions of the Pancatantra are of interest only in so far as they may throw light on the ultimate source of them all.
What was the original Pancatantra? — Even a superficial exa- mination of the existing Pancatantra versions indicates with tolerable certainty that they all go back to a book of fables and stories consisting of five books or sections and a brief introduction. Tbe, introduction provides tbe frame ” or setting, and at the same time suggests what must have been to tbe author’s mind the key-note of .the whole work: it was supposed to be a kind of Fiirstempiegel or Mirror for Magistrates^ teach- ing worldly wisdom to princes, by entertaining examples, as well as by cleverly phrased precepts, The precepts are princi- pally found ill: the vrerses which are. abnudp.% scattered thru most parts; of the work. The example consist in the stories themselves, which are jhJld mainly in prose. Each of the five sections or books ” forms a dramatic unit in itself, and all five are, as I said, set into the Introduction as a frame. In the Introduction a wise brahman undertakes to enlighten three ignorant princes. He does so by narrating to them, one after another, the five books of tbe Pancatantra. Each of tlie five books contains not only a primary story,* which we call the “frame-story,” but also at least one,, and usually several, ^‘emboxt” stories; that is, stories represented as told by one character in the frame-story to another. Sometimes tliere is a double “ emboxment a character in an “ emboxt ” story tells
* See W, hfonsnan Brown, Tbe Pancatantra in Modern Indian Folklore,” JAOS, 39. 1 flP. This subject is not included in HertePs Pancatant)’(L, mentioned in note 3 above.
^ See notes 2 and 4 above.
Wliat was tlie original raiicatantra?
5
a story to another character, (In some of the late versions of the Pahcatantra this process was carried even further, so that we have a sort of “Chinese nest” of stories.) Most of the stories are beast-fables, that is their principal actors are animals deckt out with human properties : but a number of them have only human characters, while some have both men and animals, and even — tho rarely — gods and other supernatural beings. The stories are in general very well told and of a high artistic qua- lity. Unevennesses and inconsistencies appear not infrequently in all of the existing versions, to be sure. But I hope to be able to show that most of them (not quite all) are secondary, and due to the fact that the tales tend to deteriorate with re- telling. Most of the stories remain true to the key-note of the hook, its Machiavellian character; they are generally unmoral, and at times positively immoral, in the political lessons they inculcate. The story-teller and the political strategist are com- bined in the personality of the author, and on the whole combined very successfully. Sometimes one gets the upper liand, sometimes the other. There are passages which become tiresomely tech- nical in their expatiations on policy. More numerous, it seems to me (and fortunately so, from our point of view), are the passages in which the author as a master of narrative forgets his protest practical purpose and loses himself in the joy of telling a rattling good story.® In general, however, the two things are very skilfully united, so that a story which is clever in itself, as a story, also becomes an apt illustration of a poli- tical maxim.
Interest of this investigation. — Such, very briefly, seems to have been the original Pancatantra. If tlie genuine and pri*^ mitive text of it were known to us ; or if we were in possession of a text which could be called a reasonably close approximation to it; then this book would be unnecessary, or at least less necessary. Unfortunately we have neither of tiiese things; cer- tainly not tlie original Pancatantra, and in my opimon--» opinion which I hope to prove in the course of this
® On this point I do not agree with Herfeel, who tbinka th4t dee contained no storj that did not teach a definite political aequently rejects all stories in which he cannot find any. I shall return to this suhjeet later^ see page 77, riot0’2j page 1§6.
6
Chapter I: Purpose and results of this book
reasonably close approximation to it. If this be true, and if there is any possibility of reconstructing the lost original with reasonable accuracy and confidence, the task would seem worth the pains. If any study in literary genetics has interest or value^ surely it must be worth while to recreate the original form of a work that has enjoyed such enormous popularity in so many different times and lands.
Method employed in. the reconstruction. — My method may be briefly described as follows. I first selected the yersions of the Pancatantra which, on the basis of previous studies (especially Hertel’s), could be assumed to contain all, or at least practi- cally all, the evidence that could be used in reconstructing the original Pancatantra.'^ All other known versions can be practically excluded from consideration, since they are known to be almost or quite completely dependent on one or another of tliese versions; hence whatever they have of -the original may in general be assumed to come from one of these older and more original versions.®
Next, I undertook a very minute comparison of all the ma- terials found in each of these versions in so far as they cor- respond in meaning to materials found in any of the others. For this purpose I divided the texts into the . smallest possible units, each unit consisting^ as a rnte, in the ease of the San- skrit versions, of a single Stanka or prose sentence, — sometimes of a part pf a sentence.® I treated the text of each version
These are: Tantrlkhjlxika, Somthern Paficatantra, Nepalese Pailcatantra, Hitopadeia (in ^eater part a Pafiaat^intra version), the poetic versions found in Somadeva's Kathlsarits%ara and in K^emendra’s Bi’hatkathaniafijarl, the **textas simplidor,’' Por^abhadra, and the principal offshoots of the Pahlavi tramlaiaon.
* Possibly an exception might he made of some of the offshoots of the “ textna simplicior,” of wliich text we have no critical edition. But I believe that there is little chance of serious vitiation of the final result on account of this. See page S8. I hare used all the information available to me (espe- cially in BfertePs hook, Dom regarding the numerous later ver-
sions of the PaUoatantra. A few bits of interesting evidence bearing on minor points of the reconstruction ha^e been extracted from them, and will be presented at the proper places. In general they do not affect the result, but merely tend to confirm conclusions which were reacht without their aid.
® A start towards such a subdivision was furnisht by Hertel in the table printed in the Einleitung to his translation of the TantrlkhySiyika, pages
Method employed in the reconstruction
7
critically, noting variant readings of different manuscripts and editions in so far as these are available »
Confronting these text-units, as found in the different ver- sions, with each other, I studied the relationship of the ver- sions. When a sentence or verse was found in identical or practically identical language, and in the same position, in all the prose Sanskrit recensions, and when its general sense was found in the poetic and translated recensions, I assumed that this sentence or verse was a literal inheritance from the original. I found that such obvious correspondences are sufficiently nupierous to establish, as it seems to me, beyond the possi- bility of doubt the fact that all these recensions do in truth go back to the single literary archetype assumed, Otherwise it would seem impossible to explain so many verbal identities, not only in verses, but also in prose.
However, in the large majority of cases I was not so for- tunate as to find such general and absolute agreement. Here it was necessary, by a careful examination of the cumulative evidence of all the parallel text-units, to discover the relation- ship of the versions to the original and to each other, in order rightly to interpret their variations.^^ Unless and until this could be done with an approach to certainty, no reconstruction could be made, with any confidence, of passages in which the existing versions disagree, or which are totally lacking in some of them; for otherwise we could not answer the question, which version is more apt to be original in any given case?
100 ff. My own comparisons included a number of texts not included in this table; and ray subdivisions of the text are ranch more minute. For instance, Hertel does not divide the prose text of the “ eraboxt ’’ stories at all. He does furnish the correspondences of ailindividnal stanj&ae that occur in the versions included in his table, I found Hertel’s table very mseful as a starting-point. It goes without saying, however, that I did not aasnra© without careful verification any of the correspondences stated in it. In fact it. contains tluite a number of errors, and a more considerable number of omissions, especially in regard to the Pahlavi versions.
For examples, see Chapter VI.
Here again I found myself to no small degree anticipated by Hertel? but also, I found that in many important respects the evidence seemed to disprove some of Ms most cherisbt theories. I shall make clear below the extent to which I agree with his views as to the genealogy of the Paflea- tantra versions.
8
Chapter I; Purpose and results of this book
Primary results of this investigation. — I must postpone for a time a more detailed statement of the way in which this problem was approacht. (See Chapter III, pages 49 ff.) I wish now to state briefly just what I think has been accomplislit in regard to the primary object of the investigation, the constitution of the text of the original Pahcatantra. The Sanskrit text here publisht and translated can, in my opinion, he regarded as a close approximation to that original. It is surely, I think, very much closer to it than any existing version. More specifically, it seems to me that the following facts regarding it can be demonstrated — if not beyond the possibility of doubt, at least with an approach to certainty as great as one can often hope to attain in a matter of literary genetics. The grounds on which these propositions are based will, of course, be furnisht later.
1. Every story contained in my reconstruction can be attri- buted with great confidence — in my opinion, with virtual cer- tainty— to the original Pancatantra.
2. The original — again with virtual certainty — contained no other stories than these.
3. Every stanza contained in my reconstruction occurred in the original, with the possible exception of those which I en- close in parentheses in text and translation (thirty out of four hundred and twenty-two stanjzas). .
4. It is very possible that the original contained some verses which are not inckded in my reconstruction. I believe that there were not veiy many such.
5. As to the pro8& passagos, which for the most part con- stitute the stories proper; every sentence of my reconstruction represents at least the general sense of a corresponding sen- tenee of the original, except that;
(a) Such sentences, phrases, words, or parts of words as I enclose fa parentheses cannot vrith certainty be attri- buted to the original; that is, they may perhaps be se- condary insertions. They constitute, roughly, perhaps five to eight percent of the total prose.
(b) Such sentences, phrases, or words as I enclose be- tween daggers may fail to reproduce even the general idea of the original, altho the evidence shows that the original
Primjiry results of this inTe8tig*ation
9
had something where they stand. That is, the ^^ersions are so seriously discordant that they force us to resort to guess-work as to which retains the general sense of the original. Such cases are negligibly few.
6. I believe that there was very little, if any, prose matter ill the origiilal of which I have failed to include in my re- construction at least the general sense.
7. Furthermore^ in the case of all Sanskrit words or parts of words which I print in Roman type, as distinguisht from italics, and outside of parentheses, I believe we can be vir- tually, if not absolutely, certain that we have preserved the exact language of the original Pailcatanti'a. This is the case with most of the stanzas, and a not inconsiderable part of the prose. We occasionally find entire prose sentences which I be- lieve reproduce the original, word for word and letter for letter. More frequent are sentences of which this is only ap- proximately true, and still more frequent are sentences whicli contain a few words, or only a word or two, that were cer- tainly in the original exactly as they stand; while there arc many sentences of which even this can not be said. In the case of the verses, on the other hand, only a minority are in such a state that we cannot predicate originality of the greatest part of their language. In the case of both prose and verses I print in italics, in the text, all matter of which I do not feel virtually certain that it literally reproduces the original.
8. The order of the original — not only the stories, but the individual verses and prose sentences — was, with a very few possible exceptions, exactly as it is in my reconstruction. As to the order of the stories there are no exceptions. Attention is called in my Critical Apparatus to the few cases in which doubt exists as to the relative order, in the original, of verses and prose sections. The somewhat more frequent^ but less significant, uncertainties regarding the exact order of individual words in a sentence are not always specifically mentioned by me, because they are both obvious, and of minor importance.
Incidental results of this investigation. — One incidental result of this investigation is the fact to which I have already alluded, that many flaws in existing versions, even in the best of
10
Chapter I; Purpose and results of this hook
them, are now shown to be unoriginal. In other words, the original Pahcatantra turns out to have been a finer work, artistically^ than any of its descendants. This statement holds good, as a general proposition^ of the relationship between the original and at least the oldet existing versions— those which I have used in my work. When they depart from the original, they almost always make it worse. There are exceptions, but they are not numerous. — More important by-products of the work are the considerable number of cases in which light is thrown on problems regarding the text or interpretation o£ individual versions, as well as on their general interrelation- ships. In many cases the evidence of other versions tells us which of several variant manuscript readings should be adopted in a particular version. In some cases uncertainties as to the meaning of a passage are liquidated by reference to the other versions. And I hope to haA^e furnisht a more correct picture of the relative positions of the soA^eral extant versions than has been furnisht previously (see my genealogical table of the versions, page 48, and Chapters IV and V of this Intro- duction).
Extent of divergence from Hertel’s results. — Students of the Paneatantra will be particularly interested to knoAv the extent to which my results tend to confirm or disprove the opinions of Professor Johannes Hertel, to whose long-continued activi- ties in this field we owe so much, particularly as to the re- lations of the several versions to each other and to the ori- ginal. It seems, therefore, worth while to summarize as follows the extent to which my own views, based on tlie studies con- tained in this book, differ from Hertel’s. For a more detailed statement, see Chapter V below.
1. There are four independent streams of Paneatantra tra- dition. (For the list, , see page 52.) Hertel believes that there are only two, Tantrakhyayika, and archetype of all other
versions (and in part of one subrecension of Tantrakhyayika).
for iastanoe my article oa “ Evil-wit, No-wit and Honest-wit,” 40. §71 ff,, in which I explain the previously misunderstood verse Tantrflkhylyika I vs 167 (Reconstruction I vs 158) by reference to the j)arailel versions.
Extent of divergence from Hertel’s results
11
2. Positive agreement between versions belonging to any two of these constitutes ^rima facie evidence of the reading of the original Pancatantra.
3. Hertel assumes that all existing versions go back to a corrupt archetype, which he calls “t”. This I think is pure imagination.
4. Hertel assumes an intermediate archetype “K ”, to which all versions except Tantrakhyayika go back, and from which even one subrecension of Tantrakhyayika was contaminated. I think this “ K ” is a myth. The versions in question do not go back to any secondary archetype. They are not especially closely related — no more closely than any one of them is re- lated to Tantrakhyayika (thru the original Pancatantra).
5. Hertel also assumes another intermediate archetype “ H-W to which the Southern Pancatantra (and its relatives, the Ne- palese Pancatantra and the Hitopadesa), the Pahlavi, and the Simplicior go back. This also, I think, is a myth. These ver- sions are not connected in any close or secondary way.
6. The manuscripts of the subrecension of the Tantrakhyayika which Hertel calls ^ are not, certainly not to any considerable extent, interpolated, as compared with the other subrecension, a. On the contrary, a is fragmentary, and when it fails to re- produce something found in (3, it is generally, if not invariably, a which has lost something, not ^ which has inserted it The suhrecension ^ is as pure a Taiitr^kliyayika version as a, and on the whole a better representative of the original. No TantrS- khyayika text, however, has anything like the privileged posi- tion among Pancatantra versions which Hertel claims for the Tantrakhy^yika as a whole.
Other, less important, points on which I differ from Hertel will be brought out later. Most of the other statements found or implied in his genealogical table (‘‘ Stammbaum”) of Panca- tantra versions are borne out by my results.
CHAPTER II
THE MATERIALS
Pancataatra versions used in the reconstruction. — In tliis chapter I shall give a summary account of the texts which have formed the basis of my work, and their interrelationships a.s I conceive them, with an estimate of the value of each of them for my purpose. I shall reserve for later chapters lengthy discussions of such of my statements as may need them.
As already stated in footnote 7 on page 6 (c/. also foot- note 8, same page), the versions which I have principally used are : TantrSkhyayika, Southern Pancatantra, Nepalese Panca- tantra, Hitopadeia, the versions found, in Somadeva’s Katha- sarifsagara and Ksemendra’s Pjhatkathamanjari, the so-called “textus simplicior,” Purpabhadra, and the principal offshoots of the Pahlavi translation.
The TAJsramadmTiKiA
The TambrSkhy&yika (p.bhreviated T),^ — This is a recension of which the only mauuseiipts known come from Kashmir and a^e written in the &rada alphabet. It was discovered by Hartel in the early years of the twentieth century. It exists in two subrecensions, called by Hertel a and p, each of which contains ope or more stories, and (at least in the case of 3) a more considerable number of verses and prose ‘sentences, which the other lacks. Except for tihis, however,, the text found in both recensions is practically identical;, ,the different readings ip the manuscripts are comparativialy few and un-
Editiou : TantrXkliySyika. Die Ulieete Faesung ,des Fafioatantra . . . heraua- gepban Ton Jobaanes Hertel. Berlin, 1910. (Abh. kgl.Ges. d. Wisa zn GBttingen phil.-hist. Kl., rr. F. Bd. XII, no. 2.)— Trajialation: TaatjAbySyika. Die alteste Fassung des Paficatantra, ana dem Sanskrit Ubergetet mit Einleitung and Anmerknngen von Johannes Hertel. 2 Vols. Leipzig and Berlin, 1909.
The Tantrakhyayika
13
important. Herters edition combines the two, and quotes the variant readings of both in the critical apparatus; it tends to prefer the readings of a to those of ^ in case of a disagreement, because the editor believes that a is the more original recension. , My own opinion is rather the reverse. In any case, however, the readings of all the manuscripts quoted by Hertel must be considered in a critical study of the text. It is not safe to neglect any of them.
Extent to which the Tantrakhyayika preserves the original text. — The Tantrakhyayika gives us, on the whole, more of the original text than any other recension. I estimate that it contains the general sense, at least, of ninety-five percent of the original text, both prose and verses. And the exact lan- guag’O of the original appears to have been preserved intact more extensively in the Tantrakhyayika than in any other version. These statements are more nearly true of the 0 sub- recension than of the a; the a subrecension has omitted one entire story and a number of individual sentences and verses winch P has preserved from the original; whereas the reverse is very seldom the ease (in particular, p has all the stories of the original, and a has no original verses that are lacking in g). Yet there are, in the aggregate, a not inconsiderable number of clear omissi9ns in the TantiTikhyayika— that is, in all manu- scripts alike. To some extent these may he due merely to im- perfect textual tradition. For there are some obvious and indu- bitable lacunae in the text as we have it,— -some passages in which it is clear that the author or redactor of the Tantra- khyayika wrote something that has been lost from our manu- scripts (all of wliich are late and more or less corrupt). There are, however, also cases in which the omission of something original appears to go back * to the redactor of the Tantra- kbyayika, or even to an archetype of it, a still older but also secondary version. There are likewise many eases in which the TantrUkhyayika’s text has more or less seriously altered, without entirely omitting, a section of the original.
Secondary . additions in the Tantrakhyayika.— The inMehldes to the original found in Tantrakhyayika consist mainly of iii- sertions and expansions rather than omissions or subslitutions. Both of its subrecensions contain throe stories which did not
14
Chapter II: The materials
belong to the oi-iginal; and, in addition, a alone contains one other, and p alone fire others (but three of these five may really have been found in a, since the a manuscripts happen to have long lacunae at the points where p has these stories),®
Moreover, both recensions contain a quite considerable number of verses and prose passages which are certainly or probably unoriginal. This is, more true of ^ than of a; a contains few insertions (only a single stanza, for instance, except those pertaining to the interpolated story a III. 5) which are not found also in (3.
Hertel’s views of the Tantrakhya3rika, — Altho my object in tliis chapter is to give mainly a summary of my own deductions from my investigations, rather than to engage in controversy, I feel that it would be unfair to the discoverer, and first editor and translator, of the Tantrakhyayika if I failed to mention at this point the extent to which my views of this version differ from his. Wlien he first discovered the Tantrakhyayika, Hertel hailed it as the genuine, original ^‘Urtext” of the Pancatantra itself, — the very thing which it is the object of my present in- vestigation to reconstruct. This opinion was decidedly untenable, and Hertel has withdrawn materially from it. His present, much more modest opinion he has stated as follows:^ “The enormbiM adv^antage which the Taiitrakhyayika furnishes us lies in the faet that it is tl.ie only version which contains the unab- breviated and hot imteniioually altered language of the author, which no ottter Indian Patcataatra version has preserved,
^ The stoi-ies of the TantrEfchj^yika are: I. 8 (Blue Jackal), 1. 13
(Jackal Outwite Claxpel aad luion), II. 4 (Weaver Somilaka) ; in a alone, a III. 6 (Treackeroua BawiI); in p alone, HI. 7 (Kang ^m), p IH. 11 (Fox and Talking Gave), HI. 11 of edition (Old Hansa), IV. 1 (Punisht Onion-thief), p IV. 3 (Potter as Wamor). There are laounae in « at the places where P has the drat, third, and fourth of the five last named. All but one (King &vi) of th we nine atoiiefi occur somewhera in some one or other of the other recensions included in my study. Keverthaless I think they can all be shown pretty conclusively to he secondary. Hertel also regards them as secondary. He likewise holds several other atone® found in both recensions, and one story (Old Man, Young Wife, = and Thi^) found only in p (Appendix, p IH. 6), to be certainly or possibly secondary. I shall show later that there seem to be good grounds for oonaidering tiiem original.
^ ZDMG, 69. 118 (year 1915); Uus is the latest statement on the subject from Hertel which T have seen.
Hertel’s views of the Tautrakhyayika
15
while the Palilavi translation distorts it by numerous misunder- standings.” This is qualified elsewhere by the admission that in addition to the “ unabbreviated . , . language of the author ” it contains also numerous additions and interpolations from later hands.'^ But even thus qualified, the statement seems to me misleading in two respects.
First, I think that many of the alterations (which are after all rather numerous in the aggregate, if proportionally few; they certainly mount into the hundreds) made by the Tantra- khyayika in the text of the original were probably just as ‘‘ intentional ” as the alterations made in other versions. Surely the insertions, which Hertel himself admits were numerous, must have been “intentional” alterations; and if the redactor of the Tantrakhyayika “ intentionally ” changed the text in one way, why should he not have done so in another? In fact I think it can be proved that he or his archetype did, almost surely “ intentionally,” make many changes^ — including both omissions and substitutions — in the original author’s words.
Secondly, I think it is a very serious exaggeration to describe the advantag’e which the TantrakhyEyika has over the other versions in this respect as “ enormous ” {iingelieuer). All the Sanskrit versions which I have used in this work contain some of the original author’s words. The mainly prosaic recensions (Southern Pancatantra, Hitopadesa, “textus simplicior,” Piirna- bhadra) show, by the extent to winch they agree verbally with the Tantrakhyayika and with each other, that to a not incon- siderable extent (tho, I grant, not to the same extent as Tantra- khyayika) they too “ contain the unabbreviated and not [in- tentionally] altered language of the author.” The same was true of the Sanskrit original of the Pahlavi. And when these other versions differ from the TantrakhyEyika, it is not by any means safe to assume that the Tantrakhyayika is more original than they. Especially is this true of the Southeim Pancatantra. To be sure, the Southern Pancatantra abbreviates the text to a considerable extent. But it is equally true—and this is what Hertel seems to overlook — that it contains a
^ Hertel actually admits more interpolations in the text of TantrakhyEyikA than I should; at least, he regards as insertions, certain or probable, several stories which I consider genuine.
16
Chapter 11; The materials
very large proportion of the original text in unabbreviated, or only slightly abbreviated, form. In a great many sentences it agrees with other versions, especially the Tantra-
kliyayika. And it has one great advantage over the Tantralchya- yika^ that it has almost no interpolations. Nearly everything which it contains is taken from the original, at least in general sense, and largely in exact language. — 1 shall point out in dealing with the various other versions, especially the two Jain versions Simplicior ” and • Purnabhadra), that Hertel underestimates their value, also, as representatives of the original.
The Tantrakhyayika has no privileged position among Panca- tantra versions. — In short, the difference between the Tantra- khyayika and the other versions, in their relations to the original, is a difference of degree , and not a difference of kind. AH are to a considerable extent original. All are to a not in- considerable extent unoriginal. On the whole^ the Tantrakhya- yika contains more of the original than any other. But it would not be true to say that a greater proportion of the text of the Tantrakhyayika is original than of any other. In this respect it is surpast by the Southern Pahcatantra, which has much less unomginal material than the Tantrakhyayika, and probably less than any other version,^ except the greatly abbreviated and versified Somadeva. And, I "^fouid lay special emphasis on the words on the whole, italicized above. In spite of all his resarvationa, Hertel tends to assume much too lightly that the language of the Tantrakhyayika is the language of the original Paficatantra. In my opinion this can never be assumed without confirmation from some other version. And there are, all in all, a good many cases in which not only is such confirmation lacking, but on the contrary the other versions prove quite oondusively that the Tantr^khyllyika^s language is unoriginal. See Cliapter VII below, where I have collected fully two hundred such cas-es.®
* It mi^ht be equalled in this respect by the Sanskrit original of the PahiaTi, if we had It
® Over-confidence in Hertel’s opinion has misled many scholars, including myself in the past, in this respect. Thus in AJF. 36. 58 I drew the same distinction that Hertel draws between the TantrSkhySyika and all other veirions, stating that the latter were all ‘‘ deliberately and radically recon-
Tantrakhyayika — Southern Paficatantra
17
On the relation of the Tantrakhyayika to the Jain versions, see below page 36 ff.
The Southern Pa^oatantra and Related Versions
The Southern Pancatantra (abbreviated SP)J— As the name implies, this version is characteristic of Southern India. Its numerous manuscripts are groupt by its editor, Hertel, in five subrecensions^ which he calls a, o, and t He considers a the best and most original subrecension, on the whole; and in this he is clearly right. The readings of the a manuscripts, as quoted by him, regularly (tho not invariably) tend to agree more closely with other versions than those of the ^ manu- scripts. The other three subrecensions contain many secondary insertions and are in general inferior. The readings of the subrecensions a and ^ often differ considerably, — more than those of the Tantrakhyayika a and p, for instance. In view of the general superiority of a, it is unfortunate that Hertel in his edition chose to ignore a in constituting the text which he prints, using (3 exclusively, even in the many cases where ^ is corrupt and a gives us the true reading. This means that anyone who wishes to make any scientific use of the Southern Pancatantra must go to the great trouble of searching thru the wilderness of Hertel’s critical apparatus for the readings
structed”, so as to be “ really quite new works.” So also Tkomas, JJM8, 1910, p. 971: The differences which mark off the other redactions [than Tantr.] are of an order practically precluding textual comparison ^ they belong to the higher criticism, involving omissions and insertions of whole stories ... in fact recasting of a drastic character.” I now realise Hxat such views must be abandoned. Both Thomas and I, like many others, were too easily imprest by the extreme confidence of Hertel’s statements. Thomas frankly stated in the same article (p. 970) that he had not undertaken a real verification of Hertel’s theories, since that would demand an amount of time comparable to that spent upon it by Dr. Hertel himself.” Having now spent such an amount of time upon it, I feel better able to distinguish the sound from the unsound in Hertel’s work.
The edilio princeps^ by M. Haberlandt (Sitzungsberichte of the Vienna Academy, phil.-hist KL, Bd. 107, p, 397ff,) is now superseded by the following: Das siidliche Paficatantra. Sanskrittext der Rezension p mit den Lesarten der besten Hss. der Rezension a. Herausgegeben von Johannes Hertel. Leipzig, 19U6. (Abh. d. phil.-hist. KL d. kgl. sachs. Ges. d. Wiss., Bd. 24, no. 6.) No translation into a European language has yet appeared.
Edgerton, Paficatantra. XI.
2
18
Chapter 11 s The materials
of the a manuscripts on everj single word, — a ^vearisome and gratuitous labor which Hertel ought to have spared the users of his book.^
Extent to which the Southern Pahcatantra preserves the original text. — As Hertel 1ms repeatedly stated, the Southern Paiica' tantra gives us a text which is, at least to some degree, an abstract. The abbreviation of the original is, however, not so drastic as one might suppose from reading Hertel ’s statements. Every original story is preserved. The general sense of the narrative is faithfully followed; as a rule. Seldom is an essen- tial feature omitted or obscured by abbreviation. More than this: a largo number of individual sentences are taken over from the original, either verbatim, or with only slight changes. I estimate that more than three-quarters of the bulk of the prose found in the original is found, at least as to general
® HertePs reason for this procedure was a passionate opposition, amounting almost to a mania, to what he calls “ eclecticism.” According to him, the a manuscripts of the Southern Pahcatantra are not complete enuf to make it possible to print their text in its entirety; and so, rather than “contami- nate ” the p text with the readings of other subrecensions, he chose to print the “pure” text of p (with quantities of corruptions which are simply un- in terpretable). These couBiderations do not seem to me valid. It is not “ eclecticism ” to print the best text available of an individual recension, such, as SP, using aU manuscripts of that recension, whatever their inter- relationship. A suhrecension, so-called, is not an indfipendent version; it is mmrely a conveniMt grouping of manuscripts. All the suhrecensions (if the word is properly used) represent ultimately one and the same text. There is no scientific interest or tAIuo in the stupid scribal blunders of SPP, which distort so much of the printed text of the Southern Paficatantra; and there is very lifelde Interest in the still more numerous variations of P which are grammatically and Siemanrically possible, but shown by agreements of the « manuscripts with other versions to be secondary. What we should have desired of H^tel is the best approach possible to the true “ Urtext ” of the Southern Pafioatantra.^ — That Hertel made this error of judgment, to the great inconTenienoe , of all users of his edition, is all the more surprizing in view of the contrary system which he (very rightly) adopted in editing the Tantrakhyiyika In that case, altho he regards Tantrakhyayika « as more original than P, he does not hesitate to reject its readings in favor of those of P when the latter are (in his eyes) evidently required by the sense, nor to fill the extensive lacunae of the a mss. by the text of p. This is just as much “ eolectitcUm ” as it would have been to print the text of Southern Pahcatantra « so far as available, supplementing it by P; and no more so.
Southern Pahcatantra — Nepalese Pahcatantra 19
sense and to a considerable extent as to exact language, in the Southern PaScatantra.-^ The proportion of original verses preserved is only slightly less (more than two-thirds). The compression of SP should not obscure the fact that it does, after all^ preserve very much of the original, and often more accurately than the TantrUkbyayil^a.
Secondary additions in the Southern Pahcataatra.— The South- ern Pancatantra contains very few interpolations. There is one interpolated stoiy (I. 12, Shepherded and Lovers). There are a very few insertions or expansions in the prose narrative, and apparently a few inserted verges. Nearly tiie whole of the text may be regarded as representing the contents of tiie original Pancatantra.
The Nepalese Pancatantra (abbreviated N). — In 1905 Hertel received a copy of part of a Nepalese manuscript apparently intending to furnish the vers^s^ only, of a Pancatantra recension nearly allied to the Southern Pancatantra. Later he received another copy containing the remaining portions of presumably the same manuscript. This Nepalese version contains neaidy (tho not quite) all the verses contained in the a subrecension of the Southern Pancatantra. It also contains one single prose sentence found in the latter. Evidently this was included hy the redactor under the impression that it was a verse, This circumstance incidentally shows — what we should assume a priori — that this recension was prepared on the basis of a
® It must be remembered that Hertel’s printed text will not sbow this to anyibmg like tiie extent that the « manuscripts show it.
We are compelled to regard, proTiaionally, as suA ters^
aa appear only in the Soni^ern Paioatantra and the related Hepalie# t®5d and Hitoplbd®^ ft is probable tliat most of at were not found' in the o-rtg^nai, aii' otherwise the dianeea ‘tre that mM WotM
preserve a traoe of However, “iasiiis oan of 'Ooira® aot 'bO
cex'tain, and in view of the general rarii^ of id. ii ''te h/
means unlikely that some of these verses may be inhoi^tod froah ih®
The niiot tliat most of the verS'es are only looa^y '»et i® th^#' and that it is easy both to insert and to omit makes it more dffdoalt
to be sure of the seoondary charaoter of v^ea th;aO of prose text-units which are found in only one stream of tradition-.
Hdlted by pLertel! Imtroduciion and Books I—EI in the “ Anmerkungen (p. 117 E) to his edition of the Southern Falloatantraj p. SXVll of Introdnotion to his edition of
; 'p '
20
Chapter 11: The materials
complete PaScatantra text containing^ as usual, both prose and verses. Since the Nepalese text contains not a single verse or sentence that is not found in the Southern Pancatantra (a); it is safe to assume that its original was a text very similar to that. SincO; however, it frequently happens that the Nepalese text has readings which are different from those of the Southern Pancatantra (all manuscripts), and since neither is consistently superior to the other, but each often has readings shown by the other Pancatantra version^ to be more original than the other: therefore we may agree with Hertel in thinking that the Southern Pancatantra and the complete text on which the Nepalese is based were not identical, nor directly derived one from the other, but that they are closely related offshoots of the same archetype (which I would propose to call the Ur- SP," that is the archetype of the Southern Pancatantra). We shall presently see that the archetype of the Nepalese text (called by me “ Ur-N ”) was the same as that of the Hitopade^a.
The Hitopade^a (abbreviated B) : its origin.^® — This is a version connected especially with Bengal, where it is very popular, and where it presumably originated. At any rate it has sup- planted all other Pancatantra versions in popular favor there. The aiithor gives his own* name as Nlirayana^ and tells us that he used ^^the Pameataaitra and another work in, composing the Hitopade^a- He profeably liwed between 800 and 1373 A. n.j it has not been , possible, to determine the date more exactly (Hertel, Pcs^c., p. 39). The version of the Pancatantra
nepaAtedly edited, but a sstilsfactory critical edition is yet to be made. Ifor my present inresrigation I baTe used tbe two best of those accessible to me (SohlegeFs, unfortanatelyy' was not accessible), namely: (1) HitopadeSa by Edited by Peter Peterson. Bombay, 1887, (Bombay Sanskrit
no. XXXIIL)’ — (2) Handbooks for the Study of Sanskrit. Edited by Max MtUer, M. A. I: The Pirst Book of the Hitopadesa . . , London, 1864. H; The Second^ Tted and; Ifourth Books of the Hitopadela . . . London, 1865. —Mtllleris edition does not pretend to be critical or scholarly, being pro- leesedly a reader Ibr beginneimi Bevertheless it seems to me, on the whole, that the text is as good as Peterson’s: Each contains many original features that are db-anged in tibe other, so that they are both valuable for our purposes. Peterson’s odation dahns to be critical; Hertel speaks slightingly (perhaps too slightingly) of its reliability.— Numerous translations of the Hitopadeia 4aTe been uaAde in most modem European languages. Bee Hertel, .Pa^c., 43 A literal, interlinear translation is furnisht in Mdller’s edition.
Nepalese Pafieatintra— Hitopadesa
21
whicli lie used was, as Hertel has indicated, apparently the same one (called by me “ Ur-N ”) which served as a basis for the Nepalese verse- text mentioned above; that is, a near relative of the Southern Pancatantra. This is shown by the following* facts. (1) Books I and II of the Pancatantra are transposed in the Nepalese text and. the Hitopadesa, and in no other ver- sions. (2) The Hitopadei^a, like the Nepalese text, contains most of the verses of the Southern Pancatantra (except those wliich occur in parts of the work omitted by it), and its read- ings tend strongly to agree with those of the Nepalese when the latter differs from the Southern Pancatantra. The Hito- padeSa also contains a few verses of the Southern Fanea- tantra which the Nepalese, perhaps by accident, omits. It contains practically no original Pancatantra verses that are not found in the Southern Pancatantra. (3) The prose text of the Hitopadesa, in so far as it belongs to the Pancatantra tradition, tends to agree closely with that of the Southern Pancatantra.
General plan of the Hitopadesa. — As already indicated, the Hitopadesa is a combination of Pancatantra materials witli those of some other, unnamed work (or works?). Its general plan appears to have been largely original with its author. To he sure, the transposition of Pancatantra Books I and II goes hack, as we have seen, to its immediate Pancatantra archetype. And the frame-work of these two books is mainly preserved in Hitopadesa Books II and L But the rest of the work is quite new in plan. Instead of five books, the Hitopade4^a has only four. Its third book has as its frame a story whidh fe only a remote reflex of PaScatantra Book IIL The frame of its fourth book is wholly new, tiho evidently intended as a companion-piece to Book III and suggc^d by the title of the original Paficatantra^s third booki^® Book IV of Ae PaScatantra is wholly omitted ; the stories of Book V, including the story, are included as emboxt stories in Hitopade&i Books III
Patio. Book III is eiatitied War and Peao®” and of a war between the crows and tbs owls. Hit. Book HI is nailed and teUs the story of a war between two other specie! of birda, hahsas” and the peacocks; its Book IV is called Peace aad tells hew peace was made between the same two parti ea.
22
Chapter 11: The materials
and IV. Several of the emhoxt stories of Pancatantra Book I are transferred to the Hitopade^a^s new Book IV; those of Pancatantra Book III are impartially divided between Hito- padeSa Books III and IV; not a few stories of the first three books of the Pancatantra are omitted altogether, and various stories not found in the Pancatantra are inserted in all four books of the Hitopade^a, presumably from the unnamed “ other work ” referred to hy NarEyajia.
Extent to wMcli the HitopadeiSa preserves the original text.— In spite of this extensive rearrangement of its materials, the Hitopade^ia is of considerable value for the reconstruction of the original Pancatantra. It preserves most of the frame-stories of Books I and 11, and over half of the emhoxt stories of the entire Pancatantra. More important is this fact: in so far as it uses a Pancatantra archetype at all, it tends to follow it rather closely, not only in general sense, but in exact language, aldxo there are stories in which, hy exception, it departs widely. I estimate that it contains at least the general sense of not far from two-fifths of the prose, and nearly one-third of the verses, of the original Pancatantra. If the first two hooks of the Pancatantra be considered separately, the proportion of tiieir materials preserved in the Hitopadesa would be higher (perhaps' one-M£ of the prose and two-fiiA^ of the verses). Since ife PaScalafifea ^^hetype was closely allied to the Southern Pancatanka, it wilt be found that it tends to agree in general witli the readings of that text. But it forms a valuably check on them, not infrequently shows superior readings, agreeing with other versions against the Southern Paheatantra, To a considerable extent it replaces for us the lost prose of the archetype of the Nepalese verse-text. It even contains, tiho rarely, sections of tlie original which are entirely O^nitted in all our manuscripts of the Southern Panca- tantra, '■
Seeondaxy additions in the Hitopadesa. — We have spoken already of the numerous new stories found in the Hitopadesa. Aside from these, there occur, in fixe stories and parts of stories takM from the Paficatantra, a considerable number of iijLserted verses, and some expansions of the prose narrative. The latter are, however, not numerous.
Hitopadesa— Brhatkatha versious
23
The Bl.aiATKATIll VeESIOHS (SoMAEEVA ANI> KtjiEMEEEUA)
The Pancatantra’s position in the Brhathatha. — The studies of F. Lacote^^ in the existing descendants of the great story- collection, in Prakrit verse, called the Brhatkatha and attri- buted to Gunadhya, have made it practically certain that the original text of that -work contained no version of the Panca- tantra. But, according to Laedte— and his arguments seem strong, tho not perhaps absolutely compelling, on this point too— a version of it was contained in a later recast, and ex- pansion, of the Bphatkatha, made at an uncertain date apparently in northwest India, — perhaps in Kashmir. Laeote believes that this recast, too', and consequently the Pancatantra version con- tained in it, was composed in Prakrit verse, in tho dialect called PaisS-ci. This northwestern Bvhatkatha, like its archetype, the original work, is lost to us. It is known only thru two later versions: Somadeva’s KathasaritsHgara (or, as it was perhaps called originally, BpliatkathUsaritsagara ; see Speyer, Studus about the KatMsaritsagaraj Amsterdam, 1908), and Kfemendra’s BrhatkathamanjarL Both of these w^orks are in Sanskrit verse, and both were composed in Kashmir, probably in the eleventh century a, d. The evidence of these two works seems to prove that the Pancatantra version contained in their common ori- ginal was very radically abbreviated. Apparently it omitted the Introduction and at least one story of the original (I. 3). Certainly it aimed to tell the tales as briefly as possible, and contained few, if any, expansions, while omitting many features of the original which seemed to its author unessential. Especially the verses of the original suffered in the abbreviation. Very few of tihem survived.^^ The reason for this te clear; most of the verses are moralizing, proverbial stanzas, and are not a real part of tho narrative at all.
Effect of language and versification on the Bvhatkatha Torsions. — If Lacdte is right in supposing that Somadeva and Kiemendra
Particularly in his sur ta Faditi,
Only about one-fiftb of all the yeraes of tb© original bave seryed in Somadeya and Kfemendra together (counting those yrhWt oow in one but not in the other). And a nnrnber of these are catdh^vwefS of stories, not the ordinary proyerbial stanzas.
24
Chapter II: The materials
go back to an original, tbe northwestern Bi'hatkatlia, which was composed in the Pai^aci Prakrit, then it follows that the Sanskrit of these two versions is a retranslation of a trans- lation. This would lead us to expect that little, if any, of the exact language of the original could be preserved in them. Add to this consideration their poetic form, and their drastic abbreviation, and it would seem hard to believe that they could give us many words just as the original had them. Nevertheless we find in the aggregate quite a good many such, altho few in comparison with the mainly prosaic Sanskrit recensions. The preservation— or restoration — of some words of the original Sanskrit after two translations can he explained by the fact that the first translation was into a Prakritic dialect, that is a dialect closely related to Sanskrit, which preserved the hulk of the Sanskrit vocabulary, with only the usual pho- netic and morphological changes in the words. Hence it is not, after all, surprising that some of these words were retranslated into the same Sanskrit words that were found in the original. So it happens that these versions are of some help in determin- ing even the exact language of the original. There are, howe- ver, few, if any, entire sentences or verses of the original that are preserved intact in them.^®
Ksemendra (abbreviated E:s).^’^—K§emendra's text is the most drastically abbreviated of ail those which I have used. It carries the abbreviation much farther than its supposed archetype, the lost northwestern BrhatkathS, apparently did, — at least much farther &m Somadeva does. Nevertheless it contains
Ouio or two ea»es in wMcb this is approximately the case in K^emendra auiy be dne to its borrowings from, the TantrEkhySyika; see below.
The Pailcatantra section of K^eraendra has been edited by itself: Der dem FaMcatmira in Kslmnendras BrihatkathdmaiijaA. Einleitung, Text, Cfbersetzong und Anmerkungen ron Leo von Madkowski, Dr. iur, et pbiL Leipisig, ie9t. Most of Madkowski’s text is based upon a single imperfeet manuscript. The editor emends freely, sometimes judiciously, but often unsucoesafuliy. On the whole more useful, because more complete and based on more manuscripts (whose variants are quoted), is tbe text found in the following edition of ELsemendra’s complete work: The BrihatkatUmanjart of Kahe^mmdra. Edited by Mahdmahopadydya Papdit fevadatta ... and KMhiuUh. Pdudurang Parab. Bombay, 1901. (KavyamSla 69.) Paficatantra on pp, 661 ff. I have collated the text of the Paftcatantra in both these editions.
K?emondra
25
five stories which were not found in the original.^® All of these interpolated stories are found in TantrakhyUyika one of them in no other version used by me, and another nowhere'* else at the same place, while none of the five occurs outside of Tantrakhyayika and the Jain versions (which latter, as we shall see, used the same secondary archetype as Tantr.). These facts seem to justify us in believing with Hertel that if K^emendra’s principal archetype was the northwestern Brhat- katha, he must have used also a manuscript of Tantrakhya- yika. Por this reason other agreements between Ksemendra and Tantrakhyayika cannot bo considered as evidence bearing on the original. As a matter of fact K§emendra’s text is so mangled by abbreviation that he gives us comparatively little help in reconstructing even the general sense of the original; and he seldom preserves any of the original words, from what- ever source. He includes, to be sure, all the stories of the original except the Introduction and I, 3, being thus more complete than Somadeva; but as the stories lacking in Soma- deva may have been taken by Ksemendra from the Tantra- khyayika, we cannot assume that they occurred in the supposed northwestern Brhatkatha. And in spite of this relative complete- ness of his materials, the major part of the prose narrative of the original (I estimate, fully fifty-five percent) and nearly all the original verses (close to ninety percent) are omitted without trace in Ksemendra. In short, the stories are cut to the bone (to the great detriment of the result, artistically speaking). Yet, since Ksemendra contains some matter that Somadeva lacks, we cannot entirely neglect him; tho we must remember the possibility that such matter may have been taken from the Tantrakhyayika.
Secondary additions in Ksemendra, except the stories men- tioned above, are practically non-existent.
Somadeva (abbreviated So).^^ — In Somadeva’s KathasaritsSgara the five books of the Pancatantra are found separated from one
These are I. 7 (Blue Jackal), 1. 12 (Jackal outwits Camel and Lion),
III. 11 (Old Haiisa), IV. 1 ( Punish t'Onion-thief), and IV. 3 (Potter as Warrior.)
IV. 1 occurs elsewhere only in Tantr., and III. 11 only in Tantr. in the same place (in PUrpabhadra in Book I).
There are two editions of Somad era’s complete work. (1) KaiM B(wU S^gara, Die MUrchensammhmg des Somadeva. Herausgegeben von Hermann
26
Chapter U; The materials
another by extraneous materials. In this respect HerteP*^ believes that Somadeva follows his original, the northwestern Brhatkatha. *^His work is characterized by a graceful and attractive style; his stories are well-told, and while no words are wasted, they are seldom cut down so as to spoil the artistic workmanship of the narrative. In both of these respects he contrasts favorably with Ksemendra. Somadeva lacks five stories o£ the original, besides the Introduction. To what extent these omissions go back to his supposed archetype, the northwestern B]*hatkatha, cannot be determined with confidence.^^ On the other hand he preserves considerably moi'e than Ksemendra does of the bulk of the narrative. He contains at least traces of about three-fifths of the original prose. Of the original verses, of course, he gives
Brockhaxis. Leipzig: (Part I, Books 1 — 5) 1839, (Part II, Books 6—8) 1862, (Part III, Books 9 — 18) 1866. (The last two = Ahliandltmgen fur die Kunde des Movgenlandea'XX, 5 and IV, 6.) The Paiicataiitra is found on pages lllflp. of Part ni. (2) The Kafkdsaritsdgara of Somadevahhatta, Edited by Pa^idit LurgHpras^d and Eksinith Papdnrang Parab. Bombay, 1889. (Pafiea- tantra, according to Hertel, Tafw, p. 32, on pages 365 ff.) 2nd ed., Bombay, 1903. (Paflcatantra on pages 309 ff.) I have compared tbruouf the texts of both Broekhaus and DurgSprasad and Parab (2nd ed.) for the Paflcatantra section. The variants are few and usually unimportant. — The entire work of Somadeva has been translated into English; The Kathd Sarit Bdgara or Ocean ^ & B^ewm of translated; .. . . by 0. H. Taif^ey, M. A. 2 vols.
Calcutta, T860 aiad 1884 48—62, 64— 75, 84—87,
90—91 ofVoL2»
^ Bee his mono'graph Mm kUmdi^Jm Nan^mhuchf Ber. ti, d. Verb. d. kgl. sHchs. Ces, d,1ViSs., phB.-hist. KL, 1912, Bd. 64, Heft 1.
We have aeea that BCfomendra also lacks the Introduction and 1. 3 (Three Self-chused Mishaps), which therefore may be presumed to have been lacking in ilie northwestern Bybatkatha. Besides these Somadeva omits I. 4 (Crows and Serj^ent), II. 4 (Beer’s Former Captivity), and the two emboxt stories of Book V (Brahman builds Aircastles, and Barber who killed the Monks). Of these IL 4 is properly only an unessential incident in the frame- story of Book IJ, and may have been lost in the process of shortening; many of the original are lost in the Brhatkatha versions.
This same story was dropt, obviously for the same reason as here suggested, by a late descendant of Pttrvabhadra; see Hertel, Pafic,^ P-117. I. 4 is par- ticularly interesting because it forms the frame for 1. 6 in the original; Somadeva preserves I. 6 but not I. 4, and is therefore exceptionally awkward in the way he fits I. 5 into the frame. Hertel (Tantr, Einleitung zur tJber- setzung p. 42) assumes— too hastily, I think— that this omission goes back to Somadeva’s original. It nuig do so, but there is no possibility of telling.
Somatlova— Jain versions
27
US very much less (traces of a sixth to a fifth). In general he shows extraordinary fidelity to the sense of the original, in so far as he preserves it at all. There are few changes, and almost no insertions. Every story in Somadeva is (in my opinion) original, and almost every phrase gives us at least the sense of something original. For this reason, in spite of his brevity, he is very useful for the reconstruction. Moreover, there is no reason to suspect his text of being contaminated with an extraneous version, as K§emendra^s is.
The Jain Vmisions (“ Sdcplioiob ” and PtJR^jTABnADKA)
The so-called “ textus simplioior ” (abbreviated Spl).^^ — The name textus simplieior ” goes back to Kosegarten, the first editor of this version, and is kept for want of a better, since its author’s name is unknown and the titles given in the ma- nuscripts (PancakhylLnaka, or PancakbyHna, “also called Pan- catantra”) are not sufficiently distinctive (the former is applied also to PtLri^abhadra’s text). On the whole I agree with Hertel’s opinion that the author was probably a Jain, tlio not all his arguments (summarized Pane, p. 72 f.) seem to me effective, and the sum total of them is perhaps not absolutely compelling. His date is put by Hertel between 900 and 1199 (the latter being the date of Purnabhadra, who used this text — or rather,
T should say, its archetype). This version became very popular in western and central India, and, with other versions which are based on it largely or wholly, it has virtually crowded out all other Pancatantra recensions in those regions. I regret to say that the materials at my disposal for determining the text of Simplieior (as I shall call it for short) were less satis-
The imperfect ediHo by Kosefarieo (Bqjan, 1848), has been
supplanted by that publisfat in the Bombay Sanskrit Series under the title FancKatantra (BSS I, Bombay 1868, edited by G. Bthler, contains Books It" and V; BSS III, 1868, also by Bahler, Books II and III; BSS IT, edited by F. Kielhorn, Introduction and Book I). This was not intehdtdLto be a critical scholarly edition, but merely a school textbook for bofinneci;., it was apparently based on a sing-le manuscript (see Kielhorn’s statej^en^l quoted by Hertel, ZDJfG. 66. 298f.), and Hertel soapects that the autliors corrected this manuscript from Kose^arten’s edition. No other odition can be used in a critical way at all; rarious prints by Hindu editor® be of little or no value. For translations see Hertel, Fmc,, p, and p Wi,
28
Chapter II; The materials
factory than the materials for any other recension. In addi- tion to the editions referred to in note 22, I had only such scattered information about the readings of various manuscripts as is given in various places by Hertel, especially in the “Pa- rallel Specimens ” in Harvard Oriental Series 13, According to Hertel, the manuscripts fall into two groups or subrecensions, which he calls the H-class and the tj-class. To the latter be- longs the- ms, used by Buhler-Kielhorn, to the former those principally used by Kosegarten. “ Of the two classes^ each at times excels the other in the greater originality of an occa- sional passage,” It is therefore certain that the text of Sim- plicior studied and quoted by me is imperfect. A really critical edition of it would improve the readings in many places. But whether these improvements in the text of Simplicior would often have any important bearing on the reconstruction of the original, T doubt. For, in the first place, the Simplicior happens to be of less importance in reconstructing the original than, perhaps, any other text used by me. And, in the second place, all its manuscripts appear to be sufficiently close to each other in their readings so that we may assume, on the theory of chances, that the coincidence of a serious divergence in their readings^ witia a passage in which Simplicior is of senoit^s im- portance for the reconstruction, would be a rare one. This thesis I have tested on the Parallel Specimens in H08. 13, and find that it holds good. Not a single word of the original, as I reconstructed it without the use of any Simplicior text but Kielhom-Biihler, had to be changed because of the read- ings of Simplicior manuscripts there quoted.
General plan of Simplicior. — Like the Hitopade^ia, this text handles the original rather freely. It keeps the five books of the original, but makes considerable alterations in their con- tents. To begin with, it makes all five of more nearly equal length. In the original, Books IV and V are very short. Sim- plicior makes them about as long as the others. It transfers
^ Hertel, 12, p. IS. TMa statement seems to me to be proved quite
co-nelusively by tke Parallel Specimens, IIOS. 18. As to the further statement, op. cU. p, 14, that ** the text of the H-class seems to me, on the whole, to he the more original one ”, I have no means of verifying it. It hardly seems demonstrated by the small amount of material at my disposal.
“ Textus Simplicior
29
to Book IV several of the stories of Book III, and inserts several new stories in Book IV. And most of its Book V is new. Moreover, it makes Story V. 2 of the original (The Bar- ber who killed the Monks) the frame-story of Book V, and emboxes within it the frame-story of the original Book V (Brahman and Mongoose), altering it at the same time. It also makes radical changes in the frame-stories of Books III and IV, so that they resemble the originals only in a general way. The same is true of some of the emhoxt stories of Simplicior. And it adds a number of new stories in the first three books, as well as in the last two.*— On the immediate archetype of Simplicior, and its relation to the Tantrakhyayika, see below, pages 31 ff., 36 f.
Extent to which Simplicior preserves the original text. — In spite of these extensive alterations, Simplicior retains to a con- siderable extent not only the general sense of the original, but even its exact language. It must be used with caution, but can by no means be neglected in the reconstruction, Hertel says:^^ As for the single stories, he [the^ author of Simplicior] not only altered their wording throughout, but also their pur- port.” It seems to me that this is a serious exaggeration. In many individual prose sentences (not to mention stanzas) it preserves nearly, if not quite, the exact language of the ori- ginal. Many of the stories are told in a manner substantially as close to the original as in the other versions. All that I should wish to say, as a general characterization, is that on the whole Simplicior is less faithful to the general sense of the original than any of the other versions previously dealt wnth, and that it is on the whole less faithful to the precise language of the original than any of the other mainly prosaic recen- sions. I find that it is much less faithful in preserving the verses of the original than the prose (as to its genei'al sense, at least). This is curious, since it is by no means averse to stanzas; it inserts an enormous number of unoriginal stanzas. Yet it gives us only about one-tliird of the stanzas of the ori- ginal, while it has at least the general sense of probably two- thirds of the original prose. It is noteworthy that its fideli^
2** nos, 12, p. 11,
30
Chapter II: The materials
to the original decreases as the work progresses. Its innovations becorne more markt in the third, fourth, and fifth books. It preserves the sense of probably four-fifths, or very nearly as much, of the original prose of Books I and II,* while in tlie last three hooks the proportion sinks to not much more than one-half. Infidelities to the original consist partly in omissions, but more often, as regards the prose, in substitutions. Many of tliese substitutions are undoubtedly deliberate, tlio usually unsuccessful, attempts to improve the story. But many others are doubtless due to mere carelessness or indifference.
Of the stories which I believe to he original, Simplicior contains all but three and it contains a remote variant of one of these in a different position.
Secondajy additions in Simplicior. — These have been perhaps sufficiently described already. Most striking is the enormous number of inserted verses, despite the fact that Simplicior leaves out approximately two-tliirds of the verses of the original. How many of these were composed by the author of Simplicior, or his immediate archetype, it is hard to say; undoubtedly many, and probably most of them were taken from other sources, not belonging to the Paucatantra tradition. Insertions in the prose text of the stories are also not rare, and some- times very lengthy. Tih^ exceed m importancq^ those that are found in any other, vetsion me, except P^n^hhadra,
which used Simplicior as a source.
Purmabhadra (abbreviated Pnb^^We are on much surer ground regarding the text of this, the second Jainistic recen-
^ It iB, therefore, agpain an exagg^eraiion when Hertel says {Pa^c. p. 70):
** die Jaina-Rezensionen Itxlrzen ihre Yorlage bzw. Vorlagen niclit, sondern erweitem sie,’’ Tiiis is doubtless true as a general proposition, but certainly not as sn absolute rule. It is, however, true, as Hertel says (1. c.), that Simplicior goes back to an approximately complete version of the work, not to an abbrevfafion such as the SouthOm PaUcatantra.
^ These are IL 4 (Deer’s Former Captivity, really only an incident in the h-ame-story of Book H, cf. page 2C, note 21), III. 7 (Brahman, Thief and Ogre), and HI. 10 (Progs ride Serpent), A remote variant of the last- named appears as Simplicior IV. 1.
Edition: Tim Pamehaiwnira in the Recension, mlled FanchMj^onaka ... of,. , Piirnabhadra. Critically edited ... by Dr. Johannes Hertel. Cambridge, 1908. (Harvard Oriental Series 11.) An introduction and critical apparatus
Simplicior — Ptlrnabliadra
31
sion of the Pancatantra, which has been shown by Hertel's researches to have been composed probably in the year 1199 a.b, by the Jain monk Pur^iabhadra. The text of this version seems to be in very satisfactory shape; there is little doubt that as printed by Hertel it comes very close to the manuscript of the author. The differences in the oldest manuscripts are, in Hertel’s opinion, insignificant.
General plan of Pnrnabhadra: his two main sources, Tantra- kliya3rika and Simplicior. — It is quite clear that the most of Pur^abhadra’s text presents the aspect of a mosaic of the texts of the Tantrnkhyayika and Simplicior-— or of texts closely re- sembling these two ^ as we have them. This much is sufficiently indicated hy a glance at Hertel’s Parallel Specimens in H(}S, vol. 13; for they are quite typical of the most of the work. It is perhaps even more strikingly proved by the fact, which I shall show below (page Ilf.), that in a number of places tbe mosaic-work is done so unskillfully that we find in POrna- bhadra two different versions of the same passage, one copied from the TantrakhyUyika and the other from Simplicior (or from a closely similar source in each case). It appears that PQr^abhadra kept before him copies of these two main sour- ces, and for the most part literally followed one or the other, as seemed best to him. As to general plan, TantrakhyEyika and Simplicior differ little in Books I and 11. Their principal differences appear in Books III, IV, and V, and in these I think that pGr^ahliadra uniformly followed the general plan of his Simplicior archetype, which I call the “ Ur-Simplicior.” This “ Ur-Simplicior ” differed from our Simplicior text in one important respect. We have seen that the frame^story of Book III is wholly changed in our Simplicior, and tihat a number of tbe emboxt stories of Book III are transposed to Book IV. In the “ XJr-Simplicior,” which PUrnabhadra follows, apparently only part of this alteration had taken place. The first part of the frame is altered, and the fir^t emboxt story (As$ in Pan- ther’s, or Tiger’s, Skin) transposed to Book IV. But the later
to ttis Tolume appwod in HOS. 12 (19^2^ ajad a ©ompanloa ^
parallel spoeimens in HOS. 13 (1912). A ae«an tsrmnalatiom FaUcatan^am (tesdus omalwr)^ by Riebard Scbmidt, appearei at (undated; publisht 1901).
32
Chapter II: The materials
part of the frame — the consultation of the owHdng with his ministers — is retained substantially as in the original; and sto- ries 6, 8, and 9 of the original Book III remain in Book III, and are not transposed to Book IV, as they are in our Sim- plicior. That this is the case, and that Purnabhadra’s superior originality as compared with our Simplicior is not due to liis following the Tantrakhyayika or any other version, seems to me to be made pix)bable by the following facts. First, Purna- bhadra agrees mainly with our Simplicior thruout Book IV, and differs from it most strikingly in the omission of just these three stories which originally belonged to Book III. Secondly, and much more compellingly: in the entire text of the stories III. 6 (Old Man, Young Wife and Thief), III. 8 (Cuckold Carpenter), and in the latter part of III. 9 (Mouse Maiden), Puri^ahhadra agrees almost word for word with the text of Simplicior. (See my Critical Apparatus for the evidence.) It is obvious that he must have got these entire stories (ex- cept the first part of III. 9, in which he follows Tantrakhyayika) from a Simplicior manuscript. But he places the stories, not in the place to which all our manuscripts of Simplicior have transposed tliem, in Book IV, but in their original place, in Book III, where all other versions including Tantrakhyayika have them. It ^ seems to me hardly likely that he would have done this if he had used our text of Simplicior. Had he done so, he would probably have given these Tories either in the position in which Simplicior has them, or in the wording in which Tantrakhyayika has them. I can scarcely think that he would have followed the order of Tantrakhyayika, hut gone to the fourth book of a version of Simplicior and extracted from it the language of the corresponding stories found tbere.®^
I differ in this reg^ard from Hertel, who believes that PUr^ahhadra used manuscripts of both of the subrecensions of Simplicior, “H’’ and “cr”, but not an older Simplicior text to which both go back. The former propo- sition he bases on the fact that at times POrvabhadra agrees with each of the two subrecensions, in turn, in superior readings. This would be adequately explained by the supposition which I make, that he used a text much older and naore original than either subrecension. The second proposition, which denies my assumption, he bases (HOS. 12, p. 14) on the circumstance that “ in some places either the H-class or the o-class is more original than Pariiabhadra’s text.” He does not quote the passages which he has in mind.
Plan and sources of Pilnjabhadra
33
Except to this extent, PQrpabhadra agrees quite closelj with our Simplicior in Books III, IV, and V. In Books I and II
But I would suggest that such cases are doubtless due to aecomdary and independent variations made by Punpabhadra himself. Of such there is no lack. Or, some of them may be due to Par^iablmdra’s use of another version than Simplicior — whether Tautrakbyayika, or some other. Prom such out* side sources, which we know he used, be may at times have borrowed readings that are secondary in comparison with eitlier Simplicior subrecen- sion, or both.
It may be of interest to note here that there are some later Hindu versions of the Paficatantra, based mainly on Simplicior or Pur^ahhadra or both, which are closer to the original Pahcatantra than either of them in one respect, at least, namely, that the story of the Ass in the Panther’s (or Tiger’s) Skin appears in its original place, as the first emboxt story of Book UI, and is not transferred to Book IV as in both Simplicior and Pdrpabhadra. (Some of these versions repeat the story in Book IV, where Simplicior and Ptlrnabhadra have it.) Among these versions are: the manu- script (Hmrtel, Pa^., p..l04), Eatnasundara’s KathSkallola {of^cU*
p, VaccharSja’s PaficSkhyana Caupal (op. oi^. p. 199 fi.), and Hegha-
vijaya’s FaScMiy^oddhara (op. <M, p. 105 C). This naight seem to suggest that they used a stall older form of the Simplicier than the one need by Ptrpabhadra, and that in the Simplicior used by them even the^r#i part of Book III was retained essentially in its original form, tlnfortunately the data furniaht by Hertel (which are all that I have to judge by) are not sufficient to make it possible to decide this (Question definitely. But such in- formation as he furnishes is not favorable to that assumption. On the contrary, it seems to indicate that these lat© recensions got their version of the story of the Ass in the Panther’s Skin directly or indirectly from a different recension, not belonging to the Simplicior tradition at all. In one case this different recension was certainly the TantrSkhySyikaj and it was perhaps the same in the case of the o&ers- hTamelyi.the tet of thfr particular story as found in the manuscript ^ B' ^ is by Cartel, I'6. 517 £
Bow it hafpens.Hnst ^1? p,arll«»la|p story ,i» 'told In very different in the sevend (M%«d Afpamtus). BO'tably iite dsin
versions (SlmpEdlor and Ptr!^bhadra)| the ap^einf very el^osely with 'saA other, are Morent^from Tanirikhy'i^ia^ But ih# “1^
agrees so closely with the Taatrtfchyiyiica ,(in i^ito of v^bsl ‘fSrlatoi#} that there can he no doubt that it got its text from the m
suggesih. (The other Sanskrit texts are sufi&^en^y different to proven ^at they could not have been concerned.) It will be obvious to imy<»e Ww cares to examine the text of E,” in oomphrlsorn wi^ the readlnp 'Oif, versions quoted in my Critical Apparatus, that itaprimiMy
are Simplicior and Pfin^abhadra, int^olated thla parfbular TantrSkbytyika manusenript, directly or indirectly. On the of
the story in “ E,” aee the next paragraph but one.
PaiieiitttiitJm. II. ^
34
Chapter II: The -materials
he tends perhaps rather to agree with the general plan of Tantrakhyayika than with our Simplicior (but the differences
As to the other late versions referred to, the only one whose version of this story is fornisht by Hertel is that of Meghavijaya (partial text and complete analysis in ZBMO. 57, 6S9fP.). According to Hertel, Meghavijaya used as his source a version which depended on Vaccharaja, and the latter in turn was dependent on Ratnasundara. If this is the case, Ratnasundara’s version of the story of the Ass in the Panther’s (or Tiger’s) Skin would presumably decide the c[uestion of the ultimate origin of the story as found in these three recensions. Hertel does not quote either Ratnasundara’s or VaccharSja’s text of the story; and Meghavijaya’s text is a drastic abbreviation, consisting of only a few lines. It is not enuf like any of the older versions to make it possible to decide its origin. It does, indeed speak of a tiger’s {’o^&ghror) skiij, rather than a panther’s agreeing to that extent
with Simplicior and Pilr:^;iabhadra ; but in this respect its prose story may have been influeust by the catch-verse; and, as I am about to show, this would not decide the question.
The catch-verse in these four late versions needs more careful con- sideration. In the ms. it reads:
suciraih hi caran nityaih ^reyah sasyam abuddhiman vyaghracarmapratichanno vSkkrte rasabbo hatah.
In Meghavijaya it reads exactly in the same way but for the following variations, all of which, there is reason to believe, are secondary, and some of which are obvious corruptions: b, ire$ihmh iasyami sa (!) huddhimam.\ c, d, (I). The verse as given by Ratnasundara and Vaccha-
rSja is not quoted in full by Hertel, but he tells us {Pa^. p. 201) that they are like Meghav^aya in having the corruptions sasymi (or Sa^) m bvtddhmidsfi^ and vydkk^te (or It appears that we may safely assume
that all four of these recensions have the catch-verse essentially as in “E.”
Let us examine the catch-verse in Ihe older Sanskrit recensions. The TantrakhySyika has this form:
suciraih hi caran nityaiii grl§me sasyam abuddhiman dvXpicarmapratIchanho vl[kk|tad rUsabho hatah.
The Southern Paficatantra agrees except for °parichanno in c and vagdo^ad in d. The Nepalese text and the Hitopade§a agree with Southern Pailcatantra but also read (Hit, Mtlller k^etre) for g-HpriCy and Sasyam (N corrupt)
for saPf and gardahho for r&sahho. The Jain versions (Simplicjor and PUr^iabhadra), however, have a wholly different first half verse:
suguptaih rak§yamap,o ’pi darSayan damiiam vapufi.
In the second half verse they agree with Tantrakhyayika except that they read tyS>gl)T(P for dfcfjpi®, and >i}dkkrU, Consistently with the former change, they speak in the following prose story of a tiger’s skin, not a panther’s skin. All the other versions make it a panther’s {dm>pi-) skin,
Plan and sources of Pflrriabhadra
36
in these hooks afe not very great, and possibly the Simplicior text which he used may have been more like Tantrakhyayika,
except Southern Paficatantra and Hitopade^a, the former of which once, and the latter reg'ularly, also make it a tig^er’s skin in the prose story, tho readings in the catch-verse (did they take dvlpl- in the sense of
“tiger,” a sense attributed to it in Hindu lexicons? SP in the prose elsewhere uses dvlpi-\).
From these facts it seems clear that: (1) The ms. “ E,'*’ whose prose text follows TantrSfchyayika exclusively (and~*NB.— always has iMpP, not vydghraP)^ has a contaminated form of the verse, in which the first half agrees with the older versions including Tantr. (except that it agrees with Nep* and Hit. in ireyalj, for an interesting but probably secondary
agreement, since Southern Paflc., the nearest relative of Nep. and Hit,, agrees with Tantr., indicating that Nep. and Hit. go back to a version which bad this reading) ,* but in the second half “ E ” agrees with Simplicior and Purnabhadra. We must remember that the catch-verse to this favorite and widespread fable was doubtless a familiar proverb, and that slight variations in it may mean only that a particular redactor had heard a different version quoted orally. So the variations in the second half may be ex- plained;— a.nd even the inconsistency (\>ydglvra : d/aUpi) between the verse and the prose fable has a parallel in the Hitopadesa. But the difference in the first half is too markt to be accidental. This first half must certainly have been drawn by “ E ” ’s source from a text close to the TantrSkhylyika —and not from the Jain versions. That is, the first half verse was doubtless taken from the same source from which “ E ” drew the prose text of the fable. (The agreement with Nep. and Hit. in reading ^reya^ is, as I said, doubtless an accidental coincidence; “ E ” ’s prose text, at least, shows no relation at all to the SP-Nep.-Hit. group of versions.) The second half it may have contaminated from the Jain versions which were its principal sources.
(2) But the more important point is this. From HertePs statements, Katnasundara, Vaecharaja, and Meghav^aya present practically the same form of the catch-verse — in both halves^ — that “ E ” does. It seems not overbold to guess that fhey have a common source. And if they have a common source for the catch- verse, it would not be strange if they had a common source for the prose text too. But, as we have seen, the pros© text of E ” unquestionably goes back, directly or indirectly, to an inteis polation from the TantrlkhySyika. This is evidently the reason for th© position of the story in “ E,” as 8tory 1 of Book HI, instead of in Book IV* where Simplicior and Punciabhadra have it. Since Ratnaaundara etc. have the story in the same position, may we not provisionally guess that the same circumstance has the same explanation, and that these versiosae too go back directly or indirectly to the TantrEkhylyika in this story? Of course, this can only be a provisional hypothesis. But at least there is at present no reason for supposing that these versions point t© a form of the
36
Chapter II: The materials
of. the preceding paragraph), while including most of the interpolated stories of both Tantrakhyayika and Simplicior and a goodly number of others.
The “ Ur-Tantrakhyayika,” source of the ** tJr-Simplicior ” and the Tantrakhyayika. — These two principal sources of Pur^ia- bhadra appear to go back directly to a common arclietype, which I call for convenience the “ Ur-Tan trakhyayika.” It differed from the original Pancatantra in having at least three interpolated stories, and an uncertain number of minor expan- sions and additions of both prose and verses. Whether it also’ contained omissions is necessarily uncertain, since even when such are found in common in Tantrakhyayika and the Jain versions, we cannot he sure that they have not occurred in- dependently. In any case they were few in number. — That the Tantrakhyayika and the Ur-Simplicior ” are sister-versions, and that neither was derived directly from the other, seems
“ Ur-Simplicior ” in which the transposition of the story to Book IV had not yet taken place.
On pag^e 189 of Hertel’s PaikcLtantra he mentions another point in which Ratnasundara agrees with Tantrakhyayika j but he there expresses the opinion that the agreement is not due to borrowing, and states that he baa found no ia:aces of the n^e of TantriCkbyfyilca by Ratnasnndara. Tins opinion deserTes weighty and mo more hesitant regarding the suggestion
made above. Yet it cah of courso not be regarded as final. Only the text of Ratnasundara’s story can decide the matter. It is unfortunate that Hertel failed to present it
** TantrSkbyEyika 1. 8 (Blue Jackal), I. 13 (Jackal outwits Camel and Tiion), H. 4 (Weaver Somilaka). These occur only in Tantr., Simpl,, PQni. and (the first two) in Ksemendra, which doubtless borrowed them from the Tantrakhyayika (see page 25), There are good reasons for denying that any of them belonged to the original Pailcatantra. I believe that the “ Ur- TantrltkhySyika also contained HI. 11 of Tantr. |3 (Appendix 3 in edition; P’ox and Talking Gave), IV. 3 of Tantr. p (Appendix 4 in edition: Potter as Warrior), and perhaps III. 11 (Old Hansa). None of these are found in Tantr^khySyika a; hut this does not prove them late, since a omitted also the original story of the Old Man, Young Wife and Thief (P III. 6, edition Appendix 2). The first two are found in the same place in the Jain versions, the last in Pdnpahhadra in a different place. None of the three occur in any other vision except (the last two) in Ksemendra. — It is very possible that the “ Ur-Tan trikhyayika ” contained still other secondary stories^ the lack of any particular story in either our Tantralthyayika or one . or both Jain versions may he due to omission.
Sources of Pflrijabhadra
37
indicated by the fact that each preserves features of the original which the other lacks. This might, to he sure, be explained by the hypothesis that one or the other is a contaminated version, like Pur^iabhadra. That is just what Hertel does assume in his “ genealogical table ” of Pancatantra versions; namely, he regards Simplicior as a contamination of Tantra* khyayika with another recension. I see no basis for this opinion and consider it most improbable. Hertel has, in fact, made no attempt to prove it, so far as I have been able to dis- cover.
Purnabhadra’s other source or sources. — But PUr^abhadra seems to have used still other Pancatantra versions, or at least one other, not closely related to either Tantrakhylyika or Simplicior. For we find that PGirg-abhadra lias a number of features o£ the original in common with other versions— the Southern Pancatantra, the Pahlavi, or the Brhatkatlia versions — which are lacking, or are replaced by different features, in both TantrEkhySyika and Simplicior. In some such cases we even find TantrEkhy^yika and Simplicior agreeing in a secon- dary trait, against Pilri;iahhadra and other versions. We may , assume in such cases that TantrakhyS-yika and Simplicior found these secondary alterations in their common archetype, the “ Ur-Tantrakhyayika.'’ If so, apparently Purnabhadra must have derived his more original readings from a different source. What was that source, or were there several such? We can only vaguely guess. There seems to he no sufficient reason to suppose tliat Plir^iahhadra used any of tlie other versions which we now possess, such as the Southern Pancatantra or the Brhatkatha versions; nor their immediate archetypes, such m the Sanskrit original of tihe Pahlavi. For his occasional agree- ments with them are not favorable to such an assumption. They are usually features which seem to have pertained to the original Paficatantra. In a few cases they may be merely due to some accident (c. g. the occasional independent insertion of a stanza familiar to different redactors as a gofltgeltas Wort,” or a similar twist which happens to have been fiven independently to a prose passage). When one text hte used another, or when both go back to a secondary archetyfe, it is usually quite easy to detect the fact, from uniimstaiabie
38
Chapter II: The materials
evidence. {Cf. p. 49 ff.) Such evidence consists in extensive and markt agreements in secondary matters, that is in features which clearly depart from the original Pancatantra. Evidence of this kind exists to establish the interdependence of Tantra- khyayika^ Simplicior, and Pur^iabhadra, and of the Southern Pancatantra, Nepalese Pancatantra, and Hitopadesa; and the dependence of Ksemendra on Tantrakhyayika. We do not find evidence of such relations between Purnabhadra and any known version except Simplicior and Tantrakhyayika. We must there- fore provisionally assume that Purnabhadra had no closer relations to any other known version. But since he shows a number of original features at points where Tantrakhyayika and Simplicior agree on unoriginal ones, it seems to follow that he probably used some independent offshoot of the original which is inaccessible to us. He may even have used more than one such, for aught we can tell. But it seems not humanly probable that he used many more than the three versions which we have now assumed as his sources, — simply because to do so would have given him more trouble than a Hindu , redactor is likely to have taken.
Value of Purnabhadra for the reconstruction. — While Purna- bhadra was, therefore, a contaminated version, this does not mean that his text cannot be used for the reconstruction. On the contrary, it is extremely useful. To be sure, we need to remember his dependence on Tantrakhyayika and Simplicior, which means that agreements between these texts prove nothing for the originaL On the other hand, however, we have seen that there is reason to believe that he used not our Simplicior, but an older Ur-Simplicior; ” so that we can improve on our text of Simplicior hy reference to Purnahhadt^a. The same seems to be true, only in a less degree, of his relations to Tantra- khyayika; the Tantrakhyayika text which he used was at least better than our Tantrakhyayika manuscripts in many details, so that Hertel occasionally emends Tantrakhyayika’s text on the basis of Pcirnabhadra^s readings (and might with profit have done so more frequently, I think). But it is when Purnabhadra agrees with other versions against Tantrakhyayika and, Simplicior that his value is greatest. For in such cases the general presumption is that he has used his third,* to us
Pur;iabliadra’e value far reconstruction
39
unknown, source; and that such agreements establish the text of the original Pancatantra.
Extent to wHcli Ptlrnabhadra preserves the origin^ text — I
estimate that Ptlrnabhadra preserves — from one source or an- other— at least the general sense of not far from ninety per- cent of the prose text of the original, and seventy percent of the verses. The reason for the much poorer preservation of the verses is that Purnabhadra follows Simplicior to such a considerable extent; Simplicior^ as we have seen, preserves only a minority of the original verses. The exact language of the original is preserved in PHr^abhadra perhaps more exten- sively than in any other version except Tantrakhyayika; but this is largely due to the fact that Puri^abhadra follows Tantra- khyayika so extensively. However, it should be remembered that even in sections where PUr^^abhadra appears to depend on Tantrikhyayika, his text is often superior to our Tantra- khyayika manuscripts, presumably because he used a much older and more perfect manuscript than any that we have.— Every story of the original is preserved in PHruabhadra ; and all are in the order of the original except Story III. 1, which is transposed to Book IV following Simplicior, and the stories of Book V, which are also arranged as in Simplicior,
Secondary additions in Purnabhadra. — These are more nume- rous and extensive than in any other version used by me. They include, to begin with, nearly all the inserted stories found in both Tantrakhyayika and Simplicior, and a conside- rable number of others that are found in neither of these, his two principal sources. They also include very many, and fre- quently very long, additions and expansions, both prose and verses. Many of these are taken from PUri^abhadra’s several sources; but not a few seem to be original with him. Ptr^a- bhadra’s text is not only synthetic but rationalming. Hm aim is to improve on his sources. When he notes a feature which he thinks needs improvement, his general tendency is not to leave it out, but to add something which will satisfy his tense of what is fitting. An interesting instance is the way he hanil« Tantrlkhyayika’s allusion to the tala of the ‘‘Butter-blkd Brahman;” see page 177.
40
Chapter 11: The materials
The Pahlavi and its Desoenbants
The Pahlavi translation (abbreviated Pa),^ — A Persian jjhy- sician named Burzoe (also spelled Burzuyeh, and in other ways), living* under the patronage of King Chosrau Anosliarwan (these names are also variously spelled; his dates are given as 531 — 579 a. b.), made a translation into Pahlavi of a number of Indian stories of various provenance, the chief of ■which was a version of the Pancatantra. He seems to have given to his entire work the name ^^Karataka and Damanaka” (to use the Sanskrit forms of the names)^ after the two jackals who play such an important role in the first book of the PaSca- tantra. We need not concern ourselves with the parts of the work which were drawn from other sources^ such as the Maha- bhfirata. It appears that, for some reason or other, BurzQe’s translation did not include the Introduction to the Pahcatantra. Otherwise it included the entire Pancatantra except for three stories that seem to have been omitted (II. 4, Deer’s Former Captivity; III. 1, Ass in Panther’s Skin; and V. 2, Barber who killed the Monks), It transposed the story of the Three Fish (I. 11 of the original), making it the seventh story of Book L It also contained one story not found in the original, namely the Traaeheroas Bawd (I. 3 c of the Pahlavi).*^ Other- wise the Pancatantra is in a way which shows that
the Sanskrit text which the translator used was an extremely ancient one (which is indeed indicated by the date of the translation), and was very close to tlie original in most details as well as in the generaP sense of the stories. (I refer to it as tlie ‘^IJr-Pa.”) It suffered, of course, in the translation. Hertel is very severe on the translator, whom he accuses of
This story appears in TantrlkliySyika a, as III. 5, in a different place from the Pablayi, and quite differently told. It is undoubtedly a secondary interpolation made independently in both places; nevertheless the Pahlavi translator may well have found it at the place where he has it in the Sanskrit version which he used. This is not disproved by HertePs argument ZDMQ. 69. 116 for the Sanskrit catch-verse to Story I. 3 may easily have been so rewritten as to include a reference to this as well as to the other “ selhstverschuldete UnfEIle The secondary character of the story is proved not by this, but by the fact that all Sanskrit versions agree in not having the story at this point.
The Pahlavi
41
rank ignorance of Sanskrit, We must remember, however, that we do not possess the Pahlavi itself, but only secondary and tertiary offshoots. It is true tkat they present the original text often in a very distorted form. But it is certain that many of the distortions are due to later retranslators. This can be seen by comparing the Old Syriac with the Arabic and its descendants; frequently one or the other comes quite close to the original Sanskrit while its rivals are very remote and secondary. If we had even the original Pahlavi, not to mention the Sanskrit on which it was based, I think we should probably have a closer approach to the original Pancatantra than we now possess (allowing, of course, for the change of language). Only the order, especially of the verses, and to some extent of the prose sentences and paragraphs of the ori- ginal, seems to have become confused even in the Pahlavi (tho in this respect too its descendants have made tlie con- fusion considerably greater). It may be added that the same is true of every Sanskrit version we have, tho usually not to a like degree; and that therefore there is no reason to doubt that at least a part of this confusion in order goes back to Ur-Pa, the Sanskrit archetype of the PahlaAi.
Immediate offshoots of the Pahlavi, — Unhappily the Falilavi translation is lost, along with its Sanskrit original. We haA^e to rely for our knowledge of this extremely important stream of Paiicatantra tradition on its offshoots. Probably the most important of these is the Old Syriac (abbreviated Sy), made by a certain Bud, apparently about 670 a, Tho known
First edited and translated by G. Bickell, with an introduction by Theodor Benfey (Leipzig, i876). This translation was a very careditabl© work in its day, and occasionally is useful even now as a check on the following, which has in general superseded it: Kalila utid Dknna* J^ruch tmd Von Friedrich Schulthess. Berlin, 1911. The translation of Schulthess has valuable critical and comparative notes, with additions by Hertel, and with marginal references to the TantrEkhylyika (and occasionally other Sanskrit versions) added by the same scholar. It is thus made convenient for refers ring to the Sanskrit, Unfortunately Schulthess has been too much iufiuen^t, occasionally, by the impression derived from Hertel, that tho Tantr^hylyika is the original Paficatantra, An instance in which this impression has led him into a false emendation of his text, as it seems to me, is shown in his handling of vs 72 of Kapitel d (our reconstruction III vs 99); see my Oritical
42
Chapter II: The materials
only from copies of a single corrupt and fragmentary manu- script, it contains nearly the whole of the Pancatantra text as found in the Pahlavi (there are only two or three lacunae of consequence, due to defects in the unique manuscript).
The Pahlavi was also translated into Arabic by 'Abdallah ibn al-Moqaffa' about 750 a. d., under the title “ Kalilah and Dimnah.” According to information kindly furnisht me by Professor M. Sprengling of the University of Chicago, we learn from Arabic tradition that at least one — possibly several — other translations of the Pahlavi into Arabic were made; these are not recorded in HertePs Pancatantra,^^ The work became very popular in Arabic literature and there are now in existence numerous manuscripts and a number of printed texts of it. These differ very widely from one another. Equally wide differences are found in the numerous translations and retranslations from the Arabic to which reference will be made presently. It is not yet known to what extent these differences are due to editing or to secondary changes in Abdallah’s text and in translations thereof, and to what extent they may be due to the influence of different translations from the Pahlavi. It is presumed that most of the Arabic manuscripts and editions, and tiie translations therefrom, represent on the whole various revisions of Abdallah’s work. For our purposes this difficult problem is of little importance. For we can be certain that all Arabic texts and offshoots, in so far as they contain matter that represents the original Pancatantra, obtained that matter directly or indirectly from the Pahlavi* translation; and it makes little difference to us whether they derived it from Abdallah’s translation or from some other Arabic rendering of the Pahlavi. I use the term “Arabic” to denote collectively all Arabic texts and descendants so far as they are accessible
ApiparatBs on this versa. Here Bickell seems to me to have been, nearer 1h.e truth. And this in not an isolated instance.
^ Professor 8prenglin|^ refers for his authority to Hadji Khalfa’s Bihlio- I gra^lM Dictionary under “ Oalila et Dimna,” and to an-H^dim’s FihrUt, 1 p. 306, L Uf, Hadji Khalfe names as a second translator of the work from 5 Pahlavi into Arabic ^Abdallah ibn HilSI [elsewhere called ibn *Ali] al-Ahwdzi, and dates his work a.h. 166 = a. ». 781/2. Little is known of this mati, and his alleg-ed work is not definitely known to exist now.
The Arabic and its offshoots
43
to me (see below), without meaning to imply any theory as to their relation to Abdallah’s translation or any others. Under- stood in this sense, the Arabic is a more complete represen- tatiye of the Pahlavi than the Old Syriac. Nevertheless^ the Old Syriac contains some details which are omitted in all texts and translations derived from the Arabic that are known to me.^^
Qifshoots of the Pahlavi thru the Arabic, — The Old Syriac version of the Pahlavi has left no known descendants. But Arabic versions were translated and retranslated repeatedly in very early times. In default of a critical edition and trans- lation of any Arabic version itself, these early offshoots are of great importance in establishing the sense of the Pahlavi.
I shall make no attempt to enumerate them; they are fully described in the eleventh chapter of Hertel’s Fafieatantra. Here I shall mention only a few of the more important ones, chiefly such as I have used in the work of reconstruction.
Perhaps the oldest is a second Syriac version made in the tenth or eleventh century, which has been made accessible in an English translation by Keith-Falconer (Kalilah and JDimnah or the Fables of Bidpai^ Cambridge, 1885). In the eleventh century a Greek version entitled 2'i:e©avtV/)i; xal IxvTjXaTT;; was made by one Symeon Seth; from it were made Latin, German, and Slavonic versions. In the twelfth century one Nasrallah translated the Arabic into Persian; his work served as a basis for a later and better-known Persian version, the Anwari Suhaili (called in English ^’Lights of Canopus”), which has
The first edition of any Arabic text was that by Sylvestre de Sacy, Oalila et Dimna ou /aides de Bidpai, Paris, 1816. This is said to be a com- posite and imperfect text, containing a contamination of several subrooensiona.
It has been translated into English (Knatchbull, Oxford, 1819; reprinted at Cairo, 1905; a very loose and poor rendering), German (Wolff, Stuttgart, 1837; 2nd ed. 1839; a good rendering; also Holmboe and Hansen, Christiania, 1832), Prench, Danish, and Kussian. It is said by Arabists that the best text yet printed is that of L. Cheikho (Beyrouth, 1905), which is based on a single old manuscript; but this text is also imperfect, and needs to be supplemented by others. Another well-known edition is that of Khalil al- Jazidji, which is not rated highly by Arabic scholars. A critical edition of j the Arabic, based on a thoro study of all available materials, is now being j undertaken by Professor Sprengling of the University of Chicago.
44
Chapter II: The materials
been repeatedly translated into many languages of Europe and Asia (English by Eastwick, Hertford 1854, and by Wollaston, 1877, 2nd ed. 1894).. The Arabic was rendered into Spanish by an unknown author about 1251; this is a very valuable version, which rests on an Arabic text closely related to that used by Rabbi Joel in his Hebrew rendering. This latter was composed in the twelfth century, and has been edited with a French translation by J. Derenbourg, Paris 1881; Bibl. de r&e. des hautes dt. 49 (this volume also contains an edition, by Derenbourg, but no translation, of a later Hebrew trans- lation from the Arabic, made by Jacob ben Eleazer in the thirteenth century). Our text of Joel is unhappily fragmentary; the entire first book is lost. We have however the complete text of a Latin rendering of Joel, made by John of Capua between 1263 and 1278, which was printed twice about 1480 and exists also in manuscripts of about the same age. One of the early printed texts has been reprinted with valuable notes by J. Derenbourg (Bibl de l’6c. des hautes dt. 72, Paris,
1887) . The Latin of John of Capua became famous in the Middle Ages, and was rendered into Spanish, into German {Buck der Beispiele der alien Weism^ by Anthonius von Pfor or Pforr, pnblisht about 1480; an extremely popular work in medieval Europe), and into Italian (by one Doni, printed 1562). Thm Italian version was the basis of . the earliest Eng- lish descendant of the Pancatantra, by Sir -Thomas North {The MeraM Pkilosophie of Doni^ London, 1570; reprinted 1601; mi latdy reprinted again by Joseph Jacobs, London,
1888) .
Use made of the PaMavi versions in the present work.~Gene-
rally speaking a clear agreement in sense between any des- cendant of the Pahlavi and any of the Sanskrit versions raises a Btrong presumption that we are dealing with a feature of the original PaB^catant^aj since there is no evidence of. any secondary
Hertel toeatious on]y the edition of Clifford G. Alien, Macon (France), .1906. Aceorain,^ to Solaiinde an earlier edition by Gayangos appeared at Madrid in ISOO.. The edition used by me is that of Antonio G. Solalinde: CalUa y.Dhmm Fdbnlas, Aniigua i^ersion 0(utdUtma, Madrid, 19175 h is according to the editor, primarily on the editions of Allen and of Alemany (Madrid, 1915).
Use of the Pahlavi in this work
45
agreements between the Pahlayi and any Sanskrit version.®^ The number of purely accidental coincidences must in tlie nature of things be limited. In default of the Pahlavi text, tlie‘ ideal desi- deratum for use in such comparisons would he careful colla- tions of both the Old Syriac and the Arabic texts. Schulthess^s edition of the Old Syriac, supplemented hy his notes and hy BickelPs edition, gives us all the matexual that can he hoped for on that subject. Unfortunately we are not so well off as to the Arabic, Of course no single Arabic version can be used alone. However, my friend and former associate, Dr. W. N. Brown, has prepared a rendering of Books II and IV of the Pancatantra in their Arabic guise which I believe approaches our requirements. It is primarily a rendering of Olieikho’s text (see page 43, note 33), but with indications in the notes of all possibly important variants in certain other Arabic editions (especially Khalil’s) and in the principal offshoots of the Arabic. It thus contains, we may be fairly sure, all evidence for the reconstruction which could probably be extracted from any of the known Arabic texts and descendants thereof. Brown’s rendering of the Arabic for Pafic. Book II has appeared in JA08. 42. 215 — 250. His Book IV is not yet publislit, but he has kindly allowed me to use it and quote from it in manu- script. For the other three books (Pancatantra I, III, and V) I have been forced to rely almost exclusively on older and less scientific translations, since my knowledge of Arabic is not sufficient to make possible an independent use of Arabic editions. I have relied principally on the Old Spanish (ed. Solalinde), the Younger Syriac as translated by Keitli-Falconer, the Latin of John of Capua and its original, Joel’s Hebrew (so far as extant), and Wolff’s German translation (2nd ed.) of the Arabic *as edited by De Sacy. Occasionally I have used Symeon Seth’s Greek (which is less valuable for comparative purposes because much freer than the versions named above), and the AnwHri Suhaill in Eastwick’s English translation.
Extent to which the Pahlavi preserves the original text. — In estimating the value of the Pahlavi’s evidence as to the original
See Chapter V for HertePs attempts to proye such, and my reasons for disagreemg 'with him, Cf. also page 49 fP. on general methods of fixing the original.
46
Chapter II: The materials
text, we must bear in mind the allowances that have to be made for translation and retranslation and re-retranslation. From the Pahla-vi versions alone we cannot often hope to infer the precise language of the original Sanskrit. The most we can hope, in general, is that they will show us that something approximately similar to a particular verse or prose sentence was contained in tlieir Sanskrit archetype. They show us that, to an extent which we must acknowledge with deep gratitude. I find evidence that at least some parts of fully eighty percent of the original prose sentences^ and that more than seventy percent of the original verses, were found in the Pahlavi. (The percentages in either the Syriac or the Arabic alone would be somewhat lower; they would be lower in the Syriac than in the Arabic.) The reason for the smaller percentage of verses px'eserved is doubtless in part the greater difficulty of the language of the Sanskrit verses, which made successful trans- lation harder; and in part the fact that the sententious verses could more easily drop out without leaving an appreciable gap. The accuracy and completeness of the translation varies greatly in different parts of the work, as well as in the different versions. Often it is so close that it could pass for an almost word-for-word rendering of the original Sanskrit, as indicated by the extant Sanskrit versions. On the whole I can say that I am honestly surprised at the frequency of such cases, in one Pahlavi version or another.
I have already mentioned the fact that the Pahlavi omits only three emboxt stories of the original, besides the Intro- ductioiL All other stories are preserved in both Old Syriac and Arabic, except that a defect in the manuscript of the Syriac leaves us, quite accidentally, without its version of Stoi^y 1. 2 (Jackal and Drum).
Sdoondaiy addUdons in the Pahlavi. — These are few in the sections paralleling the Pancatantra. In this respect the Pahlavi rivals the Southern Pancatantra as a faithful reflex of the ori- ginal, and far surpass^ TantrK.khyayika and the Jain versions. It is distinctly surpast only by Somadeva. have seen that it includes only one unoriginal story (L 3 c, Treacherous Bawd). It includes also a small number of verses (that is, of passages which obviously represent sententious verses of the Sanskrit;
Secondary additions in Pahlavi
47
for the Pahlavi renderings are of course in prose) which at least appear in no Sanskrit version, and most of which were therefore prohablj not in the original Pancatantra. It doubtless contained likewise a number of prose insertions and expansions. But it is harder to judge of this point, because most of the existing Pahlavi versions show a strong tendency to expand on their own account. Expansions common to the Old Syriac and the Arabic are not very numerous; and it is only these which we can with confidence attribute to the Pahlavi.
TABLE
SHOWING INTERRELATIONS OP OLDER PANGHATANTRA VERSIONS
48
Indicates hypothetical versions. Italics indicate translations into other languages than Sanskrit.
CHAPTER in
METHODS EMPLOYED IN THE RECONSTRUCTION
Purpose of tMs chapter. — In this chapter I shall present a statement of the methods which I hare workt out for estab- lishing' the text of the original Pancatantra, positively and negatively, together with a brief statement of the reasons why wc may be confident that there really taas an original Paiicatantra, — that we are not eliasing a will-o'-the-wisp. De- tailed illustrations will be furnisht in later chapters. Since nothing can be decided finally about the original until we are sure what versions are secondarily interrelated^ I shall first take up the methods by which w^e may hope to decide that question.
Three ways of proving secondary interrelationship. — By ''se- condary interrelationship ” between two versions, I mean de- scent, in whole or in part, from a common archetype later than the original Paneatantra, and secondary in comparison with it. There are not more than three ways in which such descent can he proved, in my opinion; and of these I regard only the first two as entirely conclusive. A combination of the first two is desirable; and it is indeed a fact that these two generally go together, more or less, tho either may he in individual instances more important than tho other. The three methods are:
1. Proof that the versions in question agree in showing a not inconsiderable number of important and striking features which cannot reasonably be supposed to have belonged to the original Paiicatantra, nor to have been added indepen- dently in the same place in the several versions where they occur. Secondarily inserted stories are the best, and almost the only conclusive, sort of evidence that can be considered under tliis head. For in the case of a stan5?;a, or a minor motive or
Edgerton, Paacatantra, II. 4
50 Chapter IIIj Methods employed in the reconstruction
feature in a story, appearing in several versions, it is easier as a rule to suppose either that it belonged to the original, or that it was added independently in more than one version. It is much harder to suppose that two redactors should, by mere cliance and independently of each other, have added tlie same story at the same place in the text, unless indeed the original text contained a definite reference to the story in question. In actual fact no such case occurs in the Pahcatantra. There is no instance, in my opinion, of the insertion of a se- condary story at the same place (this qualification is important) in independent versions. At the same time it is usually easy to find grounds for doubting the originality of stories that have been secondarily inserted. — By this method I think it is possible to prove the interrelationship of e, g, Tantrakhyayika and the Jain versions, and of Tantrakhyayika and Ksemendra, which have a number of secondary stories in common, occur- ring at tlie same points in the text.
2. Proof of constant and far-reaching agreements in minor verbal details between the versions in question. Such agree- mentS; to prove the point, must be so regular as to be over- whdming in their force, and must include a goodly number of passages in which comparison with other versions warrants us in assuming that they do mot go back to the original Panca- tantra. By litis method I think we can prove the secondary connections of, e. g.^ the Southern Pancatantra, Nepalese Panca- tantra, and Hitopadeia; also of TantrSfkhyayika and Pur^a- bhadra.
3, Less reliable is the third method of proof, namely, proof tliat tlie versions in question are parts of some larger whole, and that said larger whole is of common origin. This is the case, among the versions used by me, only with the Pahlavi and the Brhatkatha versions. As pointed out above, the Old Syriac and the Arabic versions are offshoots of the Pablavi, which included not only a translation of a Pancatantra version but a considerable amount of other material. Since the Old Syriac and the Arabic agree in presenting this other material, which is not found connected with the Pancatantra in any other version, we should perhaps be justified on this ground alone in assuming that the Pancatantra versions found in them
• Ways of proving secondary interrelationship 61
are closely and secondarily connected. Of eonrsej ihe same can be proved by both of the other methods mentioned above. The case is different with the Brhatkatha versions, Somadeva and Ksemendra. Here this third method is the only way by which we can prove their interrelationship. It seems clear that the Kathasaritsagara of Somadeva and tlie Erhatkathtoahjarr of Ksemendra both go back as a whole to a common original (see Lacote's work cited on page 23, note 14). Therefore it seems fair, a priori^ to assume that materials common to both works were probably drawn, at least primarily, from tlmt source (in spite of the fact that K^mendra evidently used also anotlier Pahcatantra version, see page 25). But for this fact, however, it seems to me diat there would be no sufficient reason to assume such relations between the Pancatantra sec- tions of Somadeva and Ksemendra. On the one hand, they contain no secondary stories in common (indeed, Somadeva contains no secondary stoiues at all). And on the other hand, they do not. strikingly agree in verbal details. It may be a^umed that this is due to the facts that both of them are drastically abbreviated, and that both have cast their materials in poetic guise. In spite of these facts, however, both of them have managed to retain many verbal correspondences from the orh ginal; and it is curious that even in these inherited traits they seldom agree closely with each other; rather, each preserves at different times different original features. The only striking agreements between Somadeva and Ksemendra are their com- mon omission of the Introduction and of Story I, 3. But common omissions constitute merely negative agreements and prove no- thing as to ultimate relationship ; it is easy to suppose that they occurred independently. For these reasons, I retain a lingering suspicion that after all Somadeva and Ksemendra may not impossibly have got their Pancatantra versions from different sources. That is, I tliink it is at present impossible to prove absolutely that they got these sections from tlie same common source from which they undoubtedly got most of the other materials in their works; tho the presumption remains tlmt they did. Nothing is shown by the position occupied in the KatliasaritsEgara and the Brhatkathamafijarl by the Pancatantra sections of eacli ; for both Somadeva and Ksemendra rearranged
3715
52
Chapter III; Metho^ls employed in the reconstruction
their materials so extensively that there is little correspondence in the order o£ the major sections or books of their respective works, (This is, however, not true of the internal order of tlie Pancatantra sections of the two works, which in both cases follow strictly the order of the original Pancatantra.)
Versions which are not secondarily interrelated. — Unless ver- sions can be shown by one of these three methods, and pre- ferably by the first two combined, to be related, I believe that it is safe to consider them independent offshoots of the original Pancatantra. By applying these tests, I think that it is possible to establish four independent streams of Pancatantra tradition. These are:
1, TantrakhyEyika, Simplicior, and Ptiniahhadra. To this group belongs also Ksemendra in part, since it apparently used Tantrakhyayika. On the other hand, Purnabhadra made partial use of at least one different stream, not secondarily related to any of the others; so that we have traces of at least a fifth stream, which however nowhere appears in a pure and un- contaminated form in the texts which we have.
2, Southern Pancatantra, Nepalese Pancatantra, and Hito- pade^a.
3, The Brhatkatha versions, namely Somadeva and Ksemendra. But only Somadeva is a pure representative of this stream; Ksemendra is contaminated from Tantrakhyayika. Tiiereforo Ksemendra is significant when agreeing with 2 and 4,. but not with 1.
4, The Pahlavi versions.
How to determine original matter? — My readers will by this time be asking, how can one tell whether a given feature — especially one occurring in more than one of the older versions— belongs to the original or not? Or how can one gauge varying degrees of probability in this respect? I have workt out a method for this operation, which is doubtless not infallible, but which in my opinion yields results that are as sure as our materials permit, and sure enuf to justify their publication. It is not easy to make it clear in a few words; I shall develop it as succinctly as possible in the following pages. Illustrations of its workings in detail will be furnisht later.
All versions point to one archetype
53
All versioas point to a definite literary archetype.— In tlio first place the question might be raised (altho, so far as I know, it has not been responsibly raised in print), whether there ever was any “ original Paneatantra,” in the sense of a single defi- nite composition from which all the versions descended. It might bo suggested that we are dealing simply with a nebulous mass of popular fables and stories, with its edges never clearly de- fined; a treasure-store upon which various literary redactors drew, each taking portions, and thus forming, as it wore, various overlapping tho not identical Paneatantra “ schools.” ^ Nothing is more certain, to my mind, titan the impossibility of such a view. A glance at tlie table showing the conspectus of stories of the original, Chapter VIII, is perhaps enuf to show this. From tliat table it appears that, disregarding tho Hitopadesa (which is only partly based on the Paneatantra and has extensively rearranged the stories), all the versions agree in showing nearly all the stories which I take to bo original ; and, what is much more important, they have them in the same order, almost without exception. The frame stories of the five hooks are tho same except that the Jain versions use a different story as the
^ The Vedic schools have been suggested to me orally as a j^ossiblo analogy, by a scholar whose judgment I value highly. Hut this analogy seems to me a very poor one. The Vedic schools grew np around the ritual*, all the literary collections of the Veda owe their origin, form, and content to the Vedic ritual. The words spoken at this ritual were originally a quite ancillary matter, and naturally, therefore, a nebulous and indefinite one. The words actually varied constantly from time to time and from place to place, and their various forms boro only a vague and indefinite relation to each other. Out of that nebulous mass, as the thing gradually began to get crystallised, naturally there dovelopt quit© a number of more or less variant forms of tho spoken ritual, which resembled each other only to an extent comparable to that to which tho various temporal and local forms of the pragmatic ritual resembled each other. That is, there was a profound general similarity; after all, the ritual was essentially the same all over; but there was an in- definite number of minor variations, each of which, generally speaking, had as good a right to be called “ original ” as perhaps any other. — But until some reason can be shown for such a process of development in the case of the PaheaUntra, it seems to me wo can hardly pass from one to tho other as if the cases were analogous. That they certainly are not, it seems to me. What ritual, or other outside consideration, could possibly have been re- sponsible for the comparatine fixation of the Paficatantra which must surely be admitted to be indicated as a condition precedent to all our versions?
54
Chapter III; Methods employed in the reconstruction
frame for the fifth book. Of the thirty-two emboxt stories, twenty-three are found in all the versions. Of the remaining nine, one (IV. 1) is lacking only in the Nepalese verse-text (that is, the single verse which it contained was omitted by the extractor of the verses); two others (I. 4 and V. 1) are lacking only in Somadeva; one (HI. 1) only in Pahlavi; two (III, 7 and 10) only in Simplicior; one (I. 3) only in Somadeva and Ksemendra; one (V 2) only in Somadeva and Pahlavi; and the ninth (11. 4) in Somadeva, Pahlavi, and Simpli- eior.® All the stories are found at the same point in the text of all recensions (so far as found in them at all), except that (1) Pahlavi has placed I. 11 before I. 7; (2) the Jain versions have transferred III, 1 to Book IV and rearranged the stories of Book V; (3) Simplicior has transferred to Book IV some of the other stories of Book III (c/, on this, however, page 31 f. above). It is hardly plausible to suppose that so many redactors should have drawn on a loose stock of fables and, by mere accident^ have come so close to selecting the same fables. But it is next to impossible that, having once selected the fables, they should have arranged them all in practically the same order, — unless it were possible to show some reason in the nature of things, or some external determining cause, why precisely this order and no other should hav6 been selected; and that seems not to he possible, The fact that some of the versions have inserted secondarily quite a number of other stories does not detract from the force of this argument.
Even more compelling, however, is the striking verbal agreement between the versions thruout so much of their extent. Not only do they all, as a rule, tell the same stories in the same way. Their very language is to a considerable extent identical; to an extent which would, I think, be literally inconceivable except upon the assumption that they go back to the single definite literary archetype assumed. Take for example the passage, I §§ 34 — 48 and vss 7—23, quoted with readings of all versions in Chapter VI below. This passage includes fifteen consecutive prose sections and seventeen con-
® Our ins. of the Old Syriac happens to have a long lacuna where Story I. 2 was found; since the story occurs in the Arabic, this lack need not be counted as a real omission.
How to determine original material
56
secatire verses from the frame-story of Book I. Be it noted that the character of this particular passage is most unfavorable to its preservation intact. It contains no action whatever, no dramatic ’■elements which would arrest the attention or impress the memoiy. Yet I think one who reads the variants of the several versions can hardly help agreeing, not only that they all, except Somadeva and K^emendra, have preserved the sense of nearly all of it; but also that the extent of their verbal coincidences is such as would be quite inconceivable unless we assume that they all copied from texts which ulti- mately went back to one definite literary archetype. Even Somadeva and Ksemendra show some traces of it (c/. for instance Somadeva on I vs 9); in the dramatic portions, where a story is being told, they are much closer to the rest. It is true that the verbal correspondences found in this particular passage are more perfect than is often the case for such a considerable stretch of the text. But on the other hand^ the correspondences in general sense^ at least, are often, and especially in the dramatic and narrative portions, even more complete; that is, there are fewer omissions in some of the versions. Enuf said: we cannot but assume the actuality of our goal, the original Paiicatantra. This being admitted, the question remains how to reach that goal?
1. Features common to all versions must be original.— It seems that we have the right to assume, as a starting-point, that such features as are common to all the versions considered in this work — which includes all the older versions— -and occur at the same point, belong to the original. Otherwise, we should have to assume either a chance coincidence (surely scarcely possible in so many versions), or that all of them go back to a secondary archetype more recent tlmn the original Pancatantra. There is, in my opinion, no reason whatever to make such an assumption. (See below, Chapter V, for my reasons for not accepting an assumption of this sort made by Bertel) At any rate, we can only treat the common original of all existing versions as, for practical purposes, the original Pahea- tantra. We can hardly hope to get at one that is more original.
2. Omission, of features in Hitopadesa and the Brhatkathi, versions not significant.— Secondly, the omission in certain
56 Chapter III; Methods employed in the reconstruction
versions of features common to all the other versions does not
seriously diminish the virtual certainty that these features are
original. For instance, it is obvious on the face of it that the
Hitopade^a has rearranged its Pancatanti'a materials so com- pletely that the omission, in it, of a particular story or other feature cannot even tend to make us doubt the originality of that story or feature, if it is found in all the others. In the case of the Bpliatkathg versions, Somadeva and Ksemendra, we must be more cautious; but something of the same sort is true of them. They preserve, to he sure, most of the stories, and follow the general drift of the text. But it is obvious, so obvious that anyone who knows them cannot help regarding it as axiomatic, that they have abbreviated the text most drastically. Particularly in the non-narrative poi’tions, such as the sample referred to above and quoted in Chapter VI below (I §§ 34—48 and vss 7 — 23), they are extremely scanty. Therefore, if we fail to find a trace of an individual sentence or verse in Somadeva or Ksemendra, or both, it is evident that this is no reason for serious suspicion that it is unoriginal If it is found in Tantrakhyayika, Southern Pailcatantra, the Jain versions, and Pahlavi, and (if a verse) in the Nepalese Pancatantra, all in the same position, it would be a hardened sceptic indeed who would refuse to believe in its originality, Chance could surely not account for the independent insertion, at the same place, of mamj identical features in so many versions; and I have been unable to find the slightest reason for suspecting that all these versions go back to a secondary archetype.
S. Very minar features common to a smaller number of in- dependlent versions are not necessarily original, — V^hen it comes to agreemenlB between a smaller number of versions, we must go more slowly. When such agreements conceim only small details, it often becomes conceivable that they may bo the resuh of chance, even tho they occur in two or three independent versions. A slight change in the prose narrative may occur to more than one redactor at different times. A proverbial stanza, known to many people as a geflUgeltes Wort,’^ may be inserted independently at the same point in the narrative, if its meaning happens to fit the context. Such stanzas are often current in
Minor eorrespoiidences
57
several more or less variant forms ) a redactor may liave found a stanza in a certain form in his original, but because he happened to he familiar with. the same stanza in another form, he may have changed it.^ A redactor of another, in- dependent version may do the same thing; then wo have an agreement, which however means nothing as to the original. The general habits of individual recensions, as well as their general interrelationships, must he carefully considered in such matters. For instance, the Southern Pancatantra in its most original form, the Brhatkatha versions, and the Pahlavi are all versions which contain few interpolations or expansions. Hence if we find a feature recorded in the Southern Pancatantra, Somadeva, or Pahlavi, and also, in the same place, in some unrelated version, this raises a strong presumption that the feature is original; a stronger presumption than, for instance, would be the case with Simplicior or Piirnabhadra, both of which expand freely. Again, if the common feature occurs not only in the Southern Pancatantra but also in the Nepalese Pancatantra or the Hitopadesa, the presumption becomes still stronger; for this indicates that it probably goes back at least to the common archetype of those versions, the “ Ur-SP.'’
4. More important features common to several independent versions : probability of originality tends to vary witk importance and closeness of correspondence. — The more striking and im- ]_>ortant the feature in question is, the greater is the likelihood that agreements between different versions indicate originality-— * always barring the possibility of secondary interrelationship, which must he shown by one of the methods outlined above (page 49 ff.). Some features (for instances, see Chapter YII) may occur in two versions only, and yet it may be more reasonable to assume that the others have omitted them, tiian that the two versions inserted them independently. These are the two alternatives that arc always before us in such a It is by no means always easy to choose between them. Thw is no rule of thumb, no definite line that can he drawn; we can not define tlie exact point at which a variation becomes
3 For examples (at least possible ones) of the last two processes, see the “ unoriginal agreements ’’ cited in Chapter VI.
58
Chapter III: Methods employed in the reconstruction
so important, so peculiar, that it is harder to suppose its in- dependent occurrence than its inheritance front the original. And, as indicated in the preceding paragraph, no single instance can be considered absolutely alone. It must be considered in the light of all other similar instances that occur; and in the h’ght of the general habits of the rersions containing it.
6. Entire stories common to several independent versions at the same place are almost certainly original.— When it comes to entire stories occurring at the same place in different ver- sions, it seems to me that the case is different, and much simpler. Independent insertion of the same story at the same place in versions which knew nothing of each other, or of a common secondary archetype, seems to me a priori so impro- bable that we might almost reject its possibility— unless indeed there were in the original text a clear reference to the story in question. And if the stories are told in the several versions not only at the same point, but also in lainguage that shows clear verbal correspondences, then it seems to me that all possibility of doubt is liquidated. In that case the versions must have taken the story from the same source. And that source can only have been a Pancatantra version — whether the original, or a secondary archetype. Otherwise— if they drew on. an outside source— what human probability is there that they would have happened to insert the same story, told in <he same language (in part at least), at exactly the same point in. the text? Seldom indeed is the appropriateness of an emboxt story to its context so compelling and exclusive that we could see any reason why, on the theory of chances, a redactor should have inserted that story precisely here, rather than in any of numerous other places.* — But if the story in question
* Wtat happens when the same story is inserted independently in different versions can easily be seen from the instances in which it has occurred. Namely: (1) The stories are told in very diffevmt term,, with a markt lack of the verbal correspondences that tend to characterize the stones taken from the same archetype; and (2) They are found at widely different places. Examples are the stories of the Treacherous Bawd (Pahlavi I. nli 6, Southern PaBoatantra ? I. 28, Nirmala Pathaka’s
0 a MarEthl V. 9; see Hertel, ZDUG. 69. 115, and Pafic. p. 285); and the Blue Jackal (Tantr. L 8, K?emendra I. 7, Simpllcior I. 10, Purnabhadra I. 11 Lin all these secondarily related], and HitopadeSa m. 6 Pet., HI. 7 MU.;
Correspondences of entire stories
59
was taken from a secondary archetype, my experience leads me to be confident that it would not stand alone. There would be many other features in the versions concerned which would show the same common origin — whether entire stories inserted, or other less important insertions or variations. As I have pointed *out above (p^iges 49 ff.), and as I shall illustrate in detail below (Chapter IV), such is regularly the ease with secondarily related versions. Their secondary relations strike one so forcibly that it is hardly possible to be in any doubt about the matter.^
While such a jmori considerations may be allowed weight, they have not been solely responsible for the conclusion which I have reacht on this point, and of which I feel more than usually confident. That conclusion is that stories which occur at the same place in more than one independent version belong to the origmal. Specifically, this naeans that stories occurring in the same place in versions belonging to any two of the four groups mentioned on page 52 must be origmal, viz, (1) Tantrlkhyayika or Simplicior or PQrnabhadra; (2) Southern Pancatantra or Nepalese Paiicatantra or Hitopadesa^ (3) Somadeva (or K^emendra, except that agreement between Ksem. and Tantr. and the Jain versions must be ignored); and (4) Pahlavi. There is a strong a priori presumption that smaller agreements between two or more members of these different groups also represent the original; but in the case of entire stories this presumption amounts to virtual certainty. In actual fact, every story which I attribute to the original is found at the same place in at least three of these four streams of tradition, with two exceptions (11. 4— really only an incident in Book II’s frame story, c/. note 21, page 20— only in Tantr., ^ SP, Ksem., and Ptiru.; and V. 2, only in these same versions and Simplicior and Hitopadesa [not in the same place in the
also in numerous later and secondary versions, Hertel, Paiic,,
That the latter story occurs in a different place in Hitopadesa is of obw® not significant, since HitopadeSa otlierwise transposes the stories. What is significant is that the story is utterly different in Hitopadela; its correspon- dence to the others is extremely remote.
® Except as to Somadeva and K^emendra, which are so seriousJy abhrevxated that the ordinary tests cannot be applied to them with such success; p. 61.
60
Chapter III: Methods employed in the recoustructioii
Jain versions and Hit.]). — On the otlier hand, unoriginal are a number of stories found only in Tantr. and the Jain versions, or Tantr. and Ksemendra; and one story found only in SP, Nepalese, and Hitopade^a. In the case of the stories common to Tantr., Jain versions and Ksem., there are internal reasons for thinking them spurious in most of the cases (c/. p^ige 74 ff. below); and their omission in all streams of tradition except one is pretty sure evidence in itself. Especially noteworthy is their omission in SP; for SP is remarkably faithful in pre- serving all iviportcmt details of the original (it compresses, but does not omit much), and in particular it has preserved, I think, every story of the original, a distinction which it shares only with Tantrakhyayika and Purnabhadra.
6. Summary of methods by which originality is determined. — What is true with yirtual certainty of entire stories is true with varying degrees of probability of smaller text units, down to individual words. If they occur in more than one of the four independent streams of tradition (page 52), the a priori presumption is that they are original. The strength of this presumption is greatest with larger sections, less with brief phrases, and least with single words. The presumption is strengthened by lack of any positive agreement among the remaining, discordant versions. If we find two alternative and irreconcilable agreements, each supported by two or more independent versions, it is evident that we are dealing, in one case or the other, with a chance coincidence; for both cannot go back to the original. In such cases we can only conjecture, with more or less plausibility, what ilie original had. But conflicts of this sort occur, I believe, only in the case of Individual words, or at most very brief phrases; and even these are comparatively rare.
7. features occurring only in a single stream of tradition.—
Agreements between versions whicli are known to be even partially interrelated can never have conclusive force. For instance, an agreement between Tantrakhyayika, Simplicior, Pdr^iabhadra, and Ksemendra never has more force than the reading of a single version, because these versions are all to some extent interdependent. On the other hand, when the disagreements of the other streams of tradition are purely
Features in only one stream of tradition
61
negative; that is, when the others simply omit a minor feature found in one stream, instead of containing a discordant reading; then it is often impossible to be certain that the feature in question is unoriginal. For it is often quite conceivable that the feature has been omitted independently in the archetypes of as many as three streams of tradition. We must remember on such occasions that the “ Ur-SP ” and tlie Brhatkatha archetypes abbreviate more or less on principle; and that we liave only secondary and corrupt descendants of the Pahlavi archetype. Accordingly, when we find a minor feature well attested as belonging to (especially) the Tantrakhyayika-Sim- plicior-Purnabhadra(-Ks0mendra) archetype, and when there is no reason a priori to think that the feature is secondary (that is, when it is not inconsistent with something which we can establish on other grounds as pertaining to the original), then it seems to me that there is enuf chance of its being original to warrant putting it in the text — but always in paren- theses^ by which I indicate that the words in question may be seoondaiy insertions.. — This applies to minor ” features prima- rily; for the more important and striking a feature is, the less likely is it that it would have been omitted in three different archetypes, particularly in the Southern Pancatantra, which omits little of importance. A fortiori^ this principle can hardly apply to entire stories at all, in my opinion. So few original stories are omitted in any version (none whatever, I believe, in the Southern Pancatantra or Tantrakhyayika or Purjgiabhadra), that it would be surprising to find the same story omitted independently in three archetypes. But furthermore: the insertion of a story is almost sure to result in changes in the surrounding material, introducing in the context features which are indicated as secondary by the positive agreement of the other versions against those intruding features.
In regard to the moralizing verses which are so abundant in the Pancatantra, it is usually very easy either to insert them or to omit them witliout altering the context at all— or at most only by adding or omitting an uktam ca or the like. Con- sequently all redactors seem to have done both, either de- liberately or accidentally. In general 1 deal wnth the verses as with the prose, inserting in parentheses those whose originality
62
Chapter III: Methods employed in the reconstruction
is not certain, particularly those occurring in Tantrakhyayika and the Jain versions but nowhere else. With this exception, I make it a rule not to include, even in parentheses, verses of which no traces are found in any but a single stream of tradition. There is more justification for making an exception of agreements between Tantr. and the Jain versions in the case of verses than in the case of prose. For the Brhatkatha versions omit almost all the verses; lienee the onaission of verses in them means little. And both Pahlavi and Ur-SP reproduce the verses less perfectly than the prose.
I freely admit that it is not only theoretically |)ossible, but even likely, that I have by this method omitted a few stanzas which belonged to the original, but were lost in all versions except, say, Pahlavi, or the Ur-SP. I can only say in defense that it seems to me that I have come much closer to the original as a whole by this method than by any other which could have been adopted; say, by including all the verses found only in Ur-SP. Verses found only in the Pahlavi could not, of course, he included without guessing at the Sanskrit originals.
As to prose features, I think there is every reason to believe that the general sense of practically everytliing found in the original is included in my reconstruction, if not as a part of the certain text, then at least in parentheses as a possible but uncertain element in the oiuginal.
Our methods are verified inductively and pragmatically, and are not based on mere abstract considerations. — These conclusions, I say, are not based wholly, nor even primarily, on the a priori considerations advanst above. They have been workt out slowly and painfully, from a study of all the materials, I have care- fully tested all the other possibilities that I have been able to conceive; for I am well aware of the ease with which one may deceive himself by theoretical reasoning. I can lionestly say that no other theory seems to me possible, in the light of all the evidence. I hope and believe that anyone who open- miiidedly studies my text and Critical Apparatus will agree with me. For those who have not the time or inclination to do this, I offer below (Chapters VI and VII) some examples which illustrate my conclusions. It must be remembered, however,
Methods verified indticfcively
6S
that any sucli selection must in tlie nature of things be regarded as illustrative, rather than as final proof. To prove the point definitely the 'whole must be considered.®
® Wintarnitjs, DLZ. 81 (1910), 2760, was guidedi by very g'ood instiaet when he said: Jedenfjills scheint mir die t!rb'ereinstinimtiiigi,zwisch!ein awei Oder mehreren der alien Rezensionen das stlrkst© Indizium fidr den Zustand des Gmudwerkes zn seiii.” He has in mind here entire stories 5 bnt tiie same could be said of smaller text-units. Only instead of “ d^ alien Re- zensionen ” he mi^ht better have said “ der gegenseitig unabliEngii^n Re- zensionen ” — which is doubtless what he really had in mind ; this would answer Hertars question in reply, ZDUG. 69, 118, « warnm nnr alien f Und wo ist die Grenze zwischen alt und jung?*’ (Qf. below, p, 67, note 7.) The qualification that such correspondences, to be compelling, must be found at the same place in the several versions, was clearly in Winternitz^s mind, as is indicated by his following sentences. He was, to be sure, unfortunate in one of the instances be quoted; the story of the Treacherous Bawd is noi found at the same place in Pahlavi and TantrSfchyllyika «, as of oours© Heriel was not slow to point out in his reply. But Wintemita was absolutely right in asserting,, against Eertel, the originality of the story of the Old Man, Young Wife, and Thief "(Reconstruction IIL 6). This story occurs in all the visions except the a recension of TantrSkhylyika and lb© Hito- ^adefe*., — and In the same place in aU except Simplicior, which tramspoees It to the fourth book along with several other stories of the original Book m. Hertel’s arguments (most recently in ZBMG, 69. 1171) against the originality of this story seem to me lacking in all force. They are as follows :
(a) The story is inserted in a most extraordinarily awkward way in the frame-story of Tantrlkhylyika p, — True; but this merely shows the oorruptne^^ of the TantrakhySyika tradition. See my reconstructed text and Critical Apparatus, HI || 165, 166, from which it is evident at a glance that Tp has transposed to a position f>efore the ©mboxt story th^ two sections, which all other versions (SP, Pp, BfhatkathS versions) have in their premier place after the story. 1 say, in their proper place; because they wakev^ good sense here, and where T^ has them they make nonsense, or very »«r it. It is just this transference in Tp that has produced the awkwardness of which Hertel complains. The trouble with Hertel here, as in many othw cases, is that he cannot bring himself to conceive th^ other versions may be more original than Tantrakhyayika. — Furthermore, however, @ve«i if th# “awkwardness” were original, and not secondarily produced in Tantr, alone, I agree with Winternitz (U c,) that it would by no means disprove the originality of the story. There are not a few cases in which features which seem to us decidedly awkward are nevertheless surely original,
(b) Hertel asserts that the sup^josedly secondary insertion of this story in Pahlavi is responsible for the fact that the frame-story is there disarranged, so that the last owl-minister does not speak, — This is a typical example of
64 Chapter III t Methods employed in the reconstruction
Critique of Hertel’s method.— I find myself here again differing from Hertel on an important matter of principle. He seems to
the way in which Hertel jumps at conclusions which happen to support his views. A very moderate amount of comparative study of the texts would have shown him how groundless this allegation is. In the first place, there were in the original fwe owl-ministers, each of whom was consulted in turn hy the king. Pahlavi mentions the consultation of only three. The one who falls out at this place is, therefore, not the only one whom Pahlavi drops; nor is he “ der letzte,” for the last of the owl-ministers, PrakSra- kariia, speaks very ranch later in the original (Reconstruction III § 191; Tantr. “ A 281 This latter passage is omitted in Pahlavi too. Does Hertel connect Dm omission with the alleged insertion of the story of the Old Man, Young Wife and Thief, which occurred several pages earlier?— But it is easy to demonstrate that the earlier omission of an owl-minister, which occurs just before this story in Pahlavi, has nothing to do with the story in any way. Consult III § 155 of my reconstruction, with Critical Apparatus, In this § 155 the original introduced the third owl-minister, Diptaksa. The section is omitted in Pahlavi, except that apparently some of the words contained in it are confused with the preceding vs 62 of the original (in the speech of the second minister, Krurak§a). It is clear from this that the omission of the third (not “last’’) owl-minister is due to the fact that the Pahlavi runs together his speech with that of the second; and this occurs. before the story in question, and at a point whose originality is certain even by Hertel’s standards (for the prose passage III § 155 occurs also in ishe « subrecension of Tantr;, “ A 226 a ”). It seems to me equally clear that the true reason for Pahlavi’s failure to refer to two owl-ministers is a very Mmple one, and the same in both cases. It is, that the original puts no into the months of two owl-ministers (the second and the fifth,, ICrurSksa and PrlkSrahariia). This made it easy for the Pahlavi to overlook the brief references to the consultation with these two. The Pahlavi alludes only to as many owl-ministers as have stories to tell. It runs together Dlpl^ga’s speech with that of Kraraksa, and leaves out PrSkarakarna altogether. — At any rate the alteration in Pahlavi, which drops one owl- minister at this point, conoems only the undoubtedly original § 155 (Tantr. “ A 225 a ■”), and does not at all concern the following story.
(c) If Hertel were right in his hypothesis of the “ secondary archetype K,” to which he believes all Parle, versions except Tantr., and in part even Tantr. go back, then of course the agreement of all these versions would not prove the originality^ of the story, I shall show (in Chapter V) that this “ archety^pe K” seems to be a fiction of Hertel’s imagination. But it happens Ihat Hertel denies even to “ K ” this particular story, since- he thinks it was inserted hy the immediate archetype of Pahlavi., This apparently means that he would deny it also to his imaginary “ JST-W,” which he supposes to be the common original of Pahlavi, the Ur-SP, and Bimplicior. In short, it appears that Hertel, unless I misunderstand him,
Critique of Hertel’a method
65
me, as to Winternitz {DLZ, 31 [1910], 2760), to lay much too great weight on the rule which he lays down {ZDM6. 64.631 f. and elsewhere), that fuller versions must be assumed a priori to be later, and briefer ones earlier. There is, perhaps, some justification for this rule, tho it has many exceptions. But Hertel seems to come dangerously near to operating with it as a hard-and-fast axiom. Yet he ignores it when it suits his purpose. For instance, the Southern Pancatantra is briefer than the Tantrakhyayika ; but Hertel does not hesitate to declare
believes that this story was inserted, purely independently, by at least four different redactors of Paficatantra versions, viz. those of (1) Tantr. p, (2) tTr-SP, (S) Palilavi, (4) Somadeva,— or their respective immediate arche- types. (He would presumably suppose that the Jain versions and K§emendra might have got it from Tantr. p.) That this actually is his theory of the story seems indicated hy his remark (Einleitung to TantrSkbySyika Ober- setzung p. 141) that it is “ ein Schulbeispiel fUr Interpolation derselben Er- zkhlung in den verschiedensten Rezensionen.”
Just what does this theory ask us to believe? That at least four redactors should have happened to pick out the same story [from where? is not dear] — should tell it in the same way [the narrative is closely similar]-~and should insert it, by mere luck, at the same identical spot in Paficatantra Book III; a spot, hy the way, in which it is by no means called for by the context. There are dozens, perhaps hundreds, of other places in the Paiicatantra where it would fit (juite as well. Is this rational? Is it not far more reasonable to suppose that all these versions, including the Ur-Tan- trakliyayika, inherited the story from the original, and that only the sub- recension Tantr. a — or the one single manuscri<pt (note this I) which we have of it at this point — omitted it, for some reason or other? Does a single Hindu manuscript, full of lacunae and corruptions (as Hertel admits), really have so much authority as to outweigh the agreement of all other existing versions of the Paiicatantra, including the other manuscripts of its sister subrecension, Tantr. p? Why may not the archetype of this manuscript have been corrupt, or had a lacuna, at this point? Or why may not Its copyist, or om© of his predecessors, have been offended by the awkwardness of the introduction to the story in Tantr. p (referred to by Hertel himself), and so left the story out deliberately, for esthetic reasons? (Personally I think it probable that this is the true explanation; cfi p. 122 below.) Or why may not some other reason — any of a dozen conceivable reasons — have led to its omission, deliberate or accidental, in this one ms. of Tantr. «?
Hertel’s treatment of this story is worth considering at length, it seems to me, as a literal redifctio ad adsurdum of his theory that omission of a story in any one of certain recensions (Tantr. a, Tantr. p, Pahlavi, Bomadeva, Southern Paiicatantra, Nepalese Pafic.) constitutes good reason to suspect an interpolation,
Edgerton, Pancatantra. 11.
66
Chapter III: Methods employed in the reconstruction
tljat it is an abbreviated text, and that Tantrakbyayika’s text is on the whole much more original. Even more abbreviated is the text of Somadeva, as Hertel has also clearly indicated; it is not for that reason more original. But more important is the fact that even versions which are on the whole expanded can be shown to have omitted some things from their originals. Simplicior is an expanded version; yet it omits many details which are found in all the older versions, so that they surely would not be denied to the original by Hertel. Numerous in- stances can easily be found from my table of correspondences, Chapter VIII. Nay more : Simplicior omits at least one entire story which Hertel accepts as indubitably original (Brahman, Thief and Ogre, Reconstruction III. 7, Tantr. ed. ITL 6). This shows that no such absolute rule can be laid down. There is no version that does not contain both omissions and insertions, be they deliberate or accidental. Some versions tend more or less strongly in one direction, some in the other ; but none are consistent— no, not even Somadeva, which contains a few un- questionable insertions, nor Pur^iabhadra (the most expanded of the versions handled by me), which contains some unquestionable omissions. Nor is it fair to demand, as Hertel does, that we prove just why a version omits something^ in every given case. It would be just as rational to demand that we prove why it inserts something. If we were omniscient, we could no doubt answer both questions. Sometimes we can guess the reason — tho seldom, I think, can we be as confident as Hertel often sounds. Frequently there is no discernible reason. Once more, all that we can do in individual instances (after once deciding that wo cannot assume secondary relationship between the versions concerned) is to ask ourselves the question, which is more likely: (1) that an identical variation or insertion was made independently in two or more versions at the same spot in the text, or (2) that this identity was inheiuted from the ori- ginal? The answer will vary with the importance and de- finiteness of the identity, with the habits of the versions in question, and with the extent to which other (discordant) ver- sions may tend to support one or the other alternative. But it is a fandamental error of principle to make the assumption a prioiij even tentatively, that when two or more versions
Critique of HertePs method
67
have a ])assage of which the rest have no trace, the former have inserted it secondarily.^
Hertel’s remarks ZBMQ, 69. llSf. are entirely beside the point as far as my position is concerned; their only weight is derived from the fact that 'Winternitz (see note 6 above, page 63) said ‘‘ alien Eezensionen instead of “ gegenseitig unabhangigen Eezensionen ”, wbicb be presumably meant. For instance: K§einendra is dependent on TantrUkhyayika, and therefor© agreements between these two versions prove nothing. The Jain versions are interdependent with Tantrakhyayika, and Ptirpabhadra is directly de- pendent on both TantrSkhySyika and Simplicior, or their immediate arche- types. The AnwSri Suhaili is known to have used other sources of Indian origin besides the Kalllah-wa-Dimnah. Meghavijaya and other late versions which have the story III. 1 (Ass in Panther’s Skin) in its original place of course got it from some version on which they depended (probably the TantrakhyiLyika, cf. page 33). In short, when Hertel says ‘‘‘ der Wintemitzsche Grundsatz fiihrt uns wieder zu Kosegarten zurttck ”, he is perhaps making a good point in dialectics, but all he really does is to prove that Winternitz was unfortunate in his phraseology. If we correct this as I have suggested, the “ Grundsatz ” is entirely sound. Of. the preceding footnote 6.
CHAPTER IV
SECONDARY INTERRELATIONSHIPS OF VARIOUS VERSIONS
Old Syeiao awd Ababio
Common archetype of the Old Syriac and the Arabic. — That the various Arabic versions and their descendants go back to the same archetype (the Pahlavi) as the Old Syriac — not, for instance, to a separate translation from Sanskrit— is shown by three considerations,
1. They contain one interpolated story (Treacherous Bawd, L 3c) at the same point; and both transpose the story of the Three Fish (original L 11), making it I. 7. In addition they show a number of common omissions of original stories— which might, however, conceivably have been omitted independently.
2. They are in general very close to each other in verbal details thruout the work* This has never been, and could not be, doubted by any one who takes the trouble to acquaint himself with the texts. It is hardly necessary to quote examples. Where unoriginal details are inserted in either Old Syriac or Arabic, they are usually found in the other also.
3. The Pan catantra sections of both are found imbedded in a larger whole, most of which is found alike in both (the parts which precede the Pahcatantra in the Arabic are not found in our ms. of the Syriac, which is fragmentary at the beginning; they include some material inserted by the Arabic translator) A
We may designate as ^^Ur-Pa” the hypothetical Sanskrit version from which the Pahlavi translation was made.
* Benfey believed tkat tke OT%imal Sanskrit work included not only the five books of our Paficatautra, but also the other sections peculiar to the Pahlavi. He supposed that these had been lost in the Sanskrit Paii catantra versions. This opinion would surely never have been exprest if Benfey had been in possession of all the evidence which we possess.
69
The “ Ur-SP,” archetype of SP, N, and 11
SOMADEVA Am> K.^EMENDRA
Common archetype of Somadeva and Ksemendra. — On the reasons for supposing that these two authors got their Panca- tautra sections from their general common archetype, the North- western Brhatkatha, see above, pages 51 f. As there stated, it seems to me that this common archetype of the Pancatantra sections rests on a presumption — a quite strong presumption, to he sure — but not on any absolute proof.
Southern PAf^OATANTRA, Nepalese PAfJcATANTRA^ and HitopadeSa
The “TJr-SP,” archetype of SP, N, and H.—Tlie fact that these three versions go back to a common archetype is proved by the following facts,
1. They all contain a secondary story, the Shepherdess and her Lovers (SP L 12, N II. 12, H II, 6). In SP and N it occurs at the same point; in H in the same book, but not at the same point (H, as we have seen, rearranges its stories to a very considerable extent).
2. In verbal details they correspond most strikingly and constantly, and often in cases where the other versions suggest that they are unoriginal. {Cf. llortel, Pane., p. 432 ff.) N has only the verses preserved, and H has omitted many of the stories altogether ; but in so far as the same text-units occur in these three versions, they agree so strikingly that no one, I think, can doubt their connexion. So far as I know no one has doubted it Since the fact seems to be unquestioned and un- questionable, I shall not take the space to prove it by examples here. My Critical Apparatus contains numerous examples.
The the secondary archetype of Iff and H. — That the
Nepalese Pancatantra, containing only verses, goes hack to a common archetype (called by me “Ur-N”) with Hitopadesa, an archetype closely related to the Ur-SP but not quite the same, is indicated by the following facts.
1. Boolcs I and II are transposed in these two texts, and in them alone,
2. In many verbal details of the verses found in both texts they agree against all other versions, even SP. See Hertel,
70 Chapter IV; Secondary interrelationships of various versions
Paflc. p. 433 f., for examples. Much more numerous examples can easily be got from my Critical Apparatus.
TaxtuIki-iyIyiica and Ksemdndka
Xsemendra used a Tantrakhyayika manuscript. — That oii.e of the sources of Ksemendra was a Tantrakhyayika text seems to me (following Hertel) to be clearly enuf indicated by the fact that Ks has five unoriginal stones, all of which occur in the Tantrakhyayika in the same places. One of the five (T and Ks IV. 1, Punisht Onion-Thief) occurs in no other version; another (T and Ks III. 11) occurs nowhere else in the same place (in Pn in Book I). These circumstances seem to indicate that the text used by Ks for these stories was either precisely our Tj or a manuscript very close to it. The other three stories are found also in the Jain versions at the same points. Thej^ are: Blue Jackal (T. I. 8, Ks I. 7), Jackal Outwits Camel and Lion (T 1. 13, Ks L 12), and Potter as Warrior (T and Ks IV. 3 ; in T ed. put in Appendix because not found in a, c/. p. 78).
Agreements in verbal details between T and Ks are neces- sarily few, because Ks abbreviates and omits so many details that it leaves only a very bare skeleton of the stories. But there are some cases in which Ks seems to have followed T in details that are secondary. See e. g. my Critical Apparatus on I § 547.
TaNTEIKHYAYIKA^ SiMPLIOIOB, and PteiiTABHADEA
The ‘^ITr-Simplicior,” source of our SimpHcior, and one of the main sources of Ptrnabhadra. — ^I have already (page 31) referred to this older form of SimpHcior, the realily of which seems to me to be indicated with gi‘eat probability by Purnabhadra's treatment of Book III, in which he has apparently followed a SimpHcior text, but one which had not yet introduced the extensive alterations, in the latter part of that book which are found in all manuscripts of SimpHcior now known to us. I have also referred (page 31) to the fact that Purnabhadra’s text is for the most part a mosaic of this Ur-SimpHcior (or at least of a text which must have been practically identical in language with our SimpHcior) and the Tantrfikhyllyika. This fact has been proved by Hertel, especially in the Parallel
The “ Ur-Simplicior,”— Duplications in Ptlrriabhadra 7 1
Specimens o£ text in HOS, Vol 13. These cases are quite typical, and are confirmed by my Critical Apparatus. It is hardly necessary to quote furtlier examples here. But it does seem to me worth while to quote a few very curious passages in which Purnabhadra has done this mosaic work so poorly that he has double versions of the same passage side by side, taking the one from Tantrakhyayika, the other from Simplicior. This seems to hav'e escaped Hertel’s notice.
Duplications in Plirnabliadra, due to Ms use of two sources.— I have noted four clear cases of this sort; there are probably others.
1. Reconstruction KM §§ 11, 12. In reply to the king’s request that he instruct his sons, the brahman Visnusarmau replies:
Spl p. 2, 1. 19 deva srhyataih me tathyavacaiiam, naliaih vidyavikrayaih ^asanasatenapi karomi. punar etliis tava putrEn masa§atkena yadi niti- &trajfian na karomi, tatab svanSmaty^aih karomi.*— athisau raja &c, T A 2 (after vs found only in T, the brahman says) tat kiih bahuna; srilya- tim ayaih mama vacanasihhanEdab. nSham arthalipsur ity evaih braYimi, na ca mamSiitiYari^asya YySYpttasarvendriyasya kascid arthopabhoga- kSlafu kith in tvaddhitirthaih buddhipUrvako arambhab* tal likh- yatgm adyatano diyasab. yady abaih na ^anmasabhyantarat tava putran nitiiastrath praty ananyasaman karomi, tato mamarhasi margasathdar- sanena bastasatam apakramayitiim. iti.— etarn asaihbbavyaih brah- manasya pratijhaih srutva sasacivo raja &c,
Pn p. 2, 1. 4 deva, sruyataih me tathyavacanam. naliaih vidyavikrayaih karomi sasanasateiia. etan punar masa^atkena yadi intisastrajhan na karomi, tatab svanamaparityagaih karomi.
kiih bahmm, ^ruyataih mamai^ siillianadab* naham arthalipsur bra- vimi, na ca me ’sitivar^asya vyavifttasarvendriyartlmsya kiihcid arthena prayojanam. kiiii tu tvatprarthanasiddhyarthaih sarasvativinodaiii kari^- yami. tal likbyatSm adyatano divasab. yady ahaih §aijmE&Sbhyaiitare tava putran niti^astraih praty ananyasad;-to na karomi, tato ’rhati me devo devamErgaih saihdar^ayitum. iti.—etlih brElimapaByasaitibhavySih pratijfiaih ^rutva sasacivo raji &c.
It seems as clear as possible that Ptirpabhadra has simply taken over bodily first Simplicior’s, and then TantrEkbyEyika’s, version of this passage, so that it has two variant versions of the same matter.
The next case is perhaps even more striking, since it introduce an internal inconsistency in PUrpabhadra’s text.
2. In the story of the Gat, Partridge, and Hare (Reconstruction III. 4), as told in Tantrakhyayika (whose general senee is supported by most versions and is clearly close to tbe original), the partridge and hare set ofi to have their dispute decided (our text, HI § 95). In § 97 the partndge
72 Chapter IV: Secondary interrelationships of various versions
asks the hare (so T, SP ; Pn with Pa makes the hare ask), who shall be the judge? In § 98 the other replies suggesting the pious cat who, he says, lives by the river engaging in austerities etc. In § 99 the former opposes this suggestion, because the cat is lisudra\ here T, followed by Pn, quotes a verse (our III vs d-S) to back up this opinion. In § 100 the cat, overhearing this conversation, engages in prayer (Jain versions, preaches a sermon), striking a religious attitude to deceive them.— Now Simplicior introduces its equivalent of § 100 before the question of the judge has been raised at all. The cat hears the partridge and the hare quarreling and decides to deceive them, by acting as described. After this (§ 100) Spl makes the hare suggest (without any preliminary question by the partridge, contrast our §97), in what corresponds to our §98:
Spl p. 67, hl5: sasaka aha, bhob kapihjala, e§a naditire tapasvi dliarma-
vadi tisthati, tad enaih prcchavali.
To which the partridge replies, in what corresponds to our § 99, not im deed rejecting the proposal outright, but:
Spl p. 67,1. 16: kapinjala aba, nanu svabhavato ’yam asmakaiii satriibhu-
tab; tad dare sthitva pircchavali-
Now Pai-pabhadra, as I indicated above, follows Tantrakbyayika closely (the exact language may be found in my Critical Apparatus ad loc.) in §§ 95—99 and vs 48,— reversing, however, the roles of the partridge and the hare in the conversation. (Pahlavi does the same, hut the agreement is doubtless purely accidental; the like occurs not infrequently in all versions; SP supports T, the Brhatkatha versions are indecisive, and Spl rather supports T, as j\ist stated.) Pm':^abhadra’s § 100 seems to combine T and SpL But after § 100, Pfiri^abhadra follows with Bim^flicior's version of ^,98, &9, as quoted above, iu the position where Spl has them, and 0Ub^uiaally identical (Pp P* 1^0, 1.23). In other words,
PUruabhadra, anxious to omit nothiug found in either of his primary BOUrci^s, forgets that he has already represented the partridge as suggesting Ihe cat as and the have as opposing the suggestion; and here he
m.akes the Iwere offer the same suggestion, as if nothing had been said on the subject before fbhos tittke, e^a nadiiaxe tapasvi dharmavadi ti§thati, tad mm prt^chlvab), while the pai’tridge counsels caution (as in Spl), ■diko according to ihe preceding part of Purnabhadra (taken from Tantr.) it was the parta-idge Mmself who first made the suggestion!
8 and 4. Oth^ eases in which Purnabhadra has clearly reproduced Ihe same pai^afu twice, once in .ite Tantr§:khygyika form and once in its Simplicior form, will be found in my Critical Apparatus on I §§216 and 217 (which must be couMdered togethei*) and I § 442. To save space I refrain ftom quo&g or discussing these passages here.
The ^'ITr-rautrSkhyltyiha;’ archetype of Tantrakbyayika and the « Br-Simplioior,”— I have indicated above (pages $6 f.), very briefly, flie nature of my reasons for assuming a common
The “ Ur-Tantrlikhyayika
73
secondary archetype for TantrakhyUyika and the Ur-Simplicior (and, of course, Puniabhadra). This secondary archetype I call the ‘‘ Ur-TantrakhyUyika,” for lack of a better name. That the two versions in question are secondarily, related can be shown by the two first methods outlined on pag'cs 49 ff.^ especially the first of them. That is, they both contain a number of se- condary stories inserted at the same points; and they agree to a considerable extent in verbal details, many of which may reasonably be suspected of being secondary. These correspon- dences can hardly be explained by supposing that either Tau- trakliy^yika or Ur-Simplicior is based directly on the other. For each contains original features which the other lacks. And I believe there is no reason for supposing that either is a con- taminated version. Of course, it is hard to disprove contami- nation. The Simplicior, in particular, has (as we have seen, page 30) many striking features that did not belong to the original. And if anyone chooses to suppose that these secondary features were not the work of the author of Ur-Simplicior, but were taken by him from some older Pancatantra version, now lost— there is no way to prove him wrong. This much, however, is clear to me: there is not a shadow of reason for believing that Simplicior has been contaminated \vith any otlier Pancatantra version of which we now have knowledge, or whose former existence wo have any conclusive reason to assume. In other words, I believe that when Sim])lieior agrees with any version other than Tantrakhyayika, or Purijiabhadra, or other (later) offshoots of these versions, such agreements are always either inheritances from the original Pailcatantra, or chance coincidences in petty details. Nowhere do I find signs of secondary connexions between Simplicior and, for in- stance, the Southern Pancatantra, Somadeva, or the Palilavi, (See Chapter V for a critique of HertePs contrary opinion,) Secondary stories inserted in XTr-Tantrakbyayika ” and found only in its descendants.— I believe that the Ur-Tantnikhyli,yika contained certainly three, — probably five, and very possibly a sixth, if not even more, — secondary stories. On page 86j note 29^ I give a list of the six stories which may, in most cases with virtual certainty, be attributed to this secondary archetype. The reason for this is that they are all (except the
74 Chapter lY: Secondary interrelationships of various versions
sixth) found in the same place in T, Spl, and Pn, and in most cases also in Ksemendra (which used Tantrakhyayika), hut in no other Pancatantra versions. If I am right in the principle laid down on page 61, this in itself would he enuf to make us strongly suspect that they do not belong to the original Pancatantra. But on the principle establisht on page 68, that stories found at the same place in several offshoots of an archetype pretty surely belong to that archetype, we should have to attribute the first five of them, at least, to the Ur-Tantrakhyayika (as the arche- type of T, Spl, and P^i, in all of which these stories occur at the same place). To be sure, two of these five are not found in Tantra- khyayika a. Their presence in Tantr. (3 might be explained by assuming with Hertel that Tantr. 13 is contaminated from some other Pancatantra version. But I shall show later (pages 121 ff.) that this opinion seems untenable. Furthermore, I have failed to find the slightest reason for regarding any of the differences between Tantr. a and p as due to influence from any outside version. I am satisfied that the features which p contains and which a omits are mostly original features which a has lost, pre- sumably in most cases as a result of lacunae or corruptions in the manuscripts or their archetypes. (We have only two mss. of T a in all, and for a large part of the work we have .only one. Both contain many lacunae, sometimes recognized by the copyists, sometimes not.) If we reject the theory that Tantr. g is contaminated, as I think we must, there remains no other plausible explanation of the discrepancies between the two sub- recensions. I have shown above (page 63, note 6) that Tantr. a omitted one story which belonged to the original Pancatantra.
All tliese stories are regarded by Hertel, also, as not parts of the original Pancatantra. But since Hertel seems to me to reject stories much too lightly, I think it desirable to show just how much definite reason there is, from my own point of view, for rejecting tliem. In addition to the general considera- tion referred to above, that they occur at the same place in only one of the four independent streams of Pancatantra tra- dition, I find the following specific grounds in each case.
1. The Blue Jackal (T 1. 8, Spl I. 10, Pn I. 11, I. 7; also H m. 6 Pet., m. 7 Mil.). — To begin with, the occurrence of this story in Hitopade^a cannot he considered an indication of
Secondary stories in Ur*TantrakliySyika; Blue Jackal
75
its belonging to the original Pancatantra. Not only does it occur in a different place (which means little, since the Hit. transposes its stories very generally) ; hut it is told there in a wholly different way, and with a wholly different catch-verse. Moreover, it is not found in any manuscript of the Southern Pancatantra, nor in the Nepalese Pancatantra. This indicates that it almost surely did not occur in the “Ur-SP,” which was the archetype from which the Hitopadesa got its Pancatantra materials. Plence, the story in Hitopadesa is an interpolation.
The insertion of the Blue Jackal story where it is found in T, Spl, Pn, and Ks disturbs the context. The situation, in the original Pancatantra, is as follows. By telling the story of the Louse and Flea (I. 7), Damanaka tries to prove to the lion that ‘‘ one should not grant asylum to one whose character is unknown ” (na t» avijfidtaHlclya kaicid dadydt pratisrayam^ I vs 86). Upon hearing the story, the lion in § 309 quite na- turally inquires what, then, is the nature of the bull: ‘^how can I recognize his hostility to me, and what is his manner of fighting? ” Damanaka’s suggestion that he does not know the bull’s real character bears fruit at once; the lion makes inquiries on the subject Compare the parallel situation where Damanaka, later, makes the same suggestion to the hull re- garding the lion (with Story I. 9, Strandhirds and Sea, tlie moral of which is that one ought not to take irrevocable steps without knowing what one’s enemy can do), and immediately the bull is prompted to inquire (I § 453) what the lion’s style of combat is.
But the versions which insert the story of the Blue Jackal at this point (just after the story of the Louse and Flea, and just before the lion’s question to Damanaka, our I § 309) disturb the continuity of the tale. The moral of the Blue Jackal story is that it is dangerous to slight old friends in favor of strangers. This is a wholly different point, which Damanaka liad previously mentioned (I § 271, and vs 76), If the Blue Jackal story had been told in the original Paiicatantra, it should rather have been told at that place. Where it stands in Tantr. etc., it spoils the logic of the lion’s question in I § 809 j for that question is evidently the appropriate reply not to the Blue Jackal story, but to that of the Louse and Flea.
76 Chapter IV: Secondary interrelationships of various versions
2. Jackal outwits Camel and Lion (T I, 13^ Spl I. 16, Pn I. 21, K? I 12). — This is a part of a longer insertion, an ex- pansion of the brief conversation between Karataka and Daina- naka in the original I §§ 456—458 and vs 128, After vs 128, Tantr, and the related versions insert several sentences and verses spoken by the two jackals to each other; and finally this story told by Damanaka to Karataka to illustrate the wisdom of looking out for number one.” None of the other versions contain any trace either of the story or of the sur- rounding material. The story itself is furthermore an obvious piece of secondary patchwork. It is made up of elements stolen from two other stories, which belonged to the original Pahca- tantra^ namely, the story of the Lion’s Retainers and Camel (reconstruction I. 8), and, that of the Ass without Heart and Ears (IV. 1). This will be evident, I think, to anyone who examines the storyj the imitation of the former story is noted by Hertel, Tantr., Einleitung, p. 134, top line. These con- siderations seem to make it practically certain that the story is secondary.
3. Weaver Somilaka (T n, 4, Spl II. 5, Pn 11. 6). — As in the
preceding case, this story iS found in the midst of some un- original material, which disturbs the context; one particulary foolish feature in it is noted by Hertel, Tantr., Einleitung, p. 136, second paragraph. The consensus of other versions shows that the order of the Tantrakhyayika is otherwise badly con- fused in the vicinity of this passage ; see my Critical Apparatus and the conspectus of text-units, Chapter VIII. That is, Tantr. not only has inserted much secondary material here, but has confused the arrangement of the materials inherited from the original. As to this story, it appears to have been built up around the theme of a verse which apparently was found in the original, vm, the vs yad abhavi na tad IMvi &c., recon- struction II vs 68. This vs is found in SP and N, at the same place, as well as in T in the middle of the Somilaka story. In SP it stands between two bits of prose that are found in Sim- plicior and PHro^abhadra just after 1he, Somilaka story, as it were driving home the moral of the story, which is identical with the moral of the verse and of these hits of prose (viz. that fate, or karma, decides everything). As so often, the
Secondary stories in Ur-TantrHlcliyEyika: Weaver Somilaka 77
Southern Pancatantra is here the most faithful representative of the original. What evidently happened was that this familiar moral, stated in the original in a few prose words and one stanza, was developt by the Ur-Tantrakhyayika in tlie long Somilaka story (which incidentally is a wretched piece of work, stupidly composed and awkwardly presented). The ori- ginal verse was then included in the now story. The original prose disappears from our Tantrakhyayika text altogether, but is preserved in the Jain versions, being placed just after the story. It is reasonable to assume that the Jain versions have followed the Ur-Tantrakhyllyika in this, and that our Tantra- khy^yika has lost this prose owing to the utter confusion into which its text has fallen in the vicinity of this passage,®
4. Talking Cave (ip III. 11, Appendix to ed.; Spl in. 4, Pn III. 15). — This story (not found in Ta ; must have been in the version of T used by Ks, which refers to the catqh-vs, see my Critical Apparatus) occurs in a passage (our III § 249) which as a whole is found only in T, Spl, Pn, and K§, and is there- fore very possibly secondary in its entirety. In it the wise owl-minister Raktak§a, foreseeing that the crow is going to destroy the owls, and having warned them in vain, summons his family and departs with them, thereby escaping destruction. Nothing is said in the sequel by which we could tell whether this much belongs to the original or not. On the principle (cj\ p. 61) that a short passage such as this may conceivably have been omitted from the other three streams of tradition, and that it fits the context well enuf, I do not feel like absolutely rejecting our III § 249, tho of course I enclose it in paren-
® The fatalistic or karma-rnorhl of tho story is regarded by Hertel as sufficient proof of its unoriginality, since he believes the original contained only stories teaching lessons of trickiness (nUi)\ cf. p, 6 above. While this argument may have some force, by way of confirmation of results otherwise proved, I do not believe that it has very much. I should never admit that such a moral in itself alone would justify us in doubting the originality of a story. There is no question that the original contained at least Hanms teaching this moral (cf. for instance II vss 70 and 71, just after this passage in ray reconstruction; these two vss are found in T and Pahlavi, and I presume, therefore, that Hertel would not deny that they are original). And if stanzas, why not stories? Hertel expects a deal too much single-
mindedness, and too much care, from a Hindu composer.
78 Chapter IV: Secondary interrelationships of various versions
theses as doubtful; the chances are, in fact, that it is unoriginal. These chances are much greater with the story. Nevertheless I think the story probably belongs to the Ur-Tantrakhyayika, tho surely not 'to the Ur-Pancatantra. Its omission in Ta is probably due to the fact that the T archetype (preserved in Tj3) was corrupt at the point where the story was introduced.^ — Incidentally the story is very poorly told in T; the Jain ver^ sions handle it much better, and certainly come closer to the way it was originally told. The inferior style of the story in T may have been one reason why the redactor of Ta omitted it^ if he omitted it deliberately.
5. Potter as Warrior (Tp, Spl, Pn, and IV. 3; not in Ta). — The omission of this story in Ta proves nothing at all, since Ta demonstrably has lost part of the original matter both before and after the point at which the story is inserted (namely, Tp IV vs 18, reconstruction IV vs 20, before the story, and T|3 A 301, with vs 23, our IV §84 and vs 21, after the story). Ta ends the fourth book very abruptly with its vs 17 (our vs 19), and there is no doubt in my mind that the original was longer. Nevertheless it seems to me unlikely that the ori- ginal Pane, contained the story here under consideration — for the general reasons mentioned page 61. In this case, as in the preceding (Talking Cave), I am unable to reinforce them by any internal evidence pointing to the insertion of the story. It is appropriate enuf (if we assume the originality of T§ A 297 and what follows ; this passage and the story go hand in hand, and if one is unoriginal, the other evidently is). And it is, at least in the Jain versions, very well told; in the Tantrakhyayika, not quite so well. — The general probabilities are, therefore, that the story belonged to the Ur-Tantrakhyayika, but not to the original Pancatantra.
3 Tp reads, after ’oatsydmaht. (p vart^) in the text of § 249 (Tantr. p. 136, 1. 3, and. Appendix, p,165, 1.1); ca gukam asannavma^opaspr^tdm andgatdm (v.l. ^tam) tyagyatMi (v. L saMgajga) ireya (v. 1. 4reyali sydt). uHai^i ca:~-At which point follows the catch- verse of the Talking Gave story, and the story itself. No words resembling this sentence occur in Spl or Ta makes reasonable sense out of them (a lecido faciUor)^ as follows: imdih . . . tyajdma
iti\ and then omits the story. Hertel regards Ta as the original, and thinks Tp has inserted the story. The opposite theory seems at least as likely. On the general question of passages found in Tp and omitted in Ta see page 121.
Other secondary stories in the Ur-TantrakhySyika
79
6. Tke Clever Hansa (T HI. 11, Ks HI. 11, P?i L 19).~-Here we have a story whose antiquity is even more qnestionahle. It occurs in the same place only in Ksemendra and TantrU- khyayika ^ (but it may well have occurred also in Ta; we cannot be sure, since Ta has a long lacuna at the point where the story is found). Even the Jain versions do not have it at the same place; Purriahhadra has it in the first hook, and Simplicior does not have it at all. Hence it is doubtful whetlier it was found even in the XJr-Tantrakhyiyika; while there is no reason whatever to suppose that it belonged to the original Paucatantra.
7. Other stories which may possibly have been found in the Trr-Tantrakhya3rika. — Our Tantrakhyayika contains two other stories (not to mention the story of the Treacherous Bawd, interpolated in Ta as HI. 5; see page 40, note 30) which are not found even in the Jain versions (Spl and Pij). One of them, King ^ivi (T od. IIL 7), is found in no other version used by me (it is not even found in Ta, but since the ms, of Ta has a lacuna at the place where it occurs, we cannot tell whether it occurred in it originally or not). The other, T IV. 1, the Punisht Onion-Thief, occurs in the same place in Ksemendra, but nowhere else (the sole ms. of Ta has a lacuna where it occurs, also). The failure of these two stories to occur in the Jain versions may conceivably be due to omission by them (Simplicior, at least, omits some original stories). Likewise, it is conceivable that some of the numerous stories found in the Jain versions, but not in Tantrakhjayika nor any other Paucatantra version, may have occurred in the Ur- Tantrakhyayika. But here we cannot do more than conjecture; and speculation on this subject is not likely to be fruitful. There is, in any case, not the slightest reason for supposing that any of these stories belonged to the original Pancatantra.
Verbal correspondences between Tantrakhyayika and Simplicior and Pur]^abhadra.— The secondary relationships between Tautrl.- khyayika and the Jain versions are, I think, sufficiently esta- hlisht by these unoriginal stories inserted in them. We should expect, however, to find them confirmed by minor* agreement in sense and language more striking and extensive than is the case with versions whose only connexion is thru the ori-
80 Chapter IV; Secondary interrelationships of various versions
ginal Pancatantra. In fact we do find that SimpHcior (not to speak of Purnabhadra, which as we have seen used Tantra- khyayika directly) agrees at many places with Tantrakhyayika much more closely than either of them with other versions. To be sure, it is often hard to tell whether these agreements are secondary, or whether they go hack to the original Panca- tantra. Since both the Southern Pancataiitra and^its relatives, and the Bphaikatha versions, tend to abbreviate the text in details, we have in Tantrakhyayika and the Jain versions the only Sanskrit versions that are not essentially abbreviated. Therefore, when they are fuller than the other Sanskrit versions, we must always consider the possibility that they preseiwe the original; and frequently the Pahlavi offshoots prove that this is the case. Failing such confirmation from the Pahlavi, it is often impossible to tell whether we are confronting an abbreviation of the original by SP etc. and the Brhatkatha versions (and an accidental omission in the Pahlavi), or an expansion by the Ur- Tan trakhyS-yika. The greater part of the phrases and sentences which I print enclosed in parentheses in my reconstructed text, indicating that their originality is uncertain, are of just this sort: they are found in Tantrakhyayika and its relatives (Spl or Pn or both), but nowhere else (unless in Ksemendra). They maybe ori- ginal, but there is no definite proof of it. It is probable that many of these passages are really unoriginal. For there is no doubt that the Ur-Tantraihyiryika contained some expansions in minor details, in addition to the above-mentioned insertions of stoines.
Clearly secondary correspondences in detail between Tantra- khyayika and SimpHcior (and Pdrpabhadra), — A few examples will now he given of minor agreements between TantrUkhyEyika and the Jain versions (especially Simplicior), all of which must,
I think, be regarded as secondary, and most of which must have originated in the Ur-Tantrakhyayika, the common secondary archetype of these versions. Otherwise they would have to be purely accidental, which at least in some of the cases seems to me impossible.
1. Eeeonstrnetion 1 §§ 18—22, Inclndingr yss 4, 5..— Here we have a passage in wbieh the order of the original, as proved by the general agreement of SP, H, So, and Pa, supported by the requirements of the sense, is departed from in T and the Jain versiom. The latter also, and
►Secondary correspondences between Tantrnkhyayika and Jnin versions 81
especially T and Pn, have a greatly expanded version. The expansion probably goes back to the Ur-Tantrakhyayika, but, in part at least, certainly not to the original Paficataiitra.
The passage includes T A7 and 8 and vs d; SP lines 56 5*. with vs 5: N vs 3; Hp p. 48, 11. 19 ff. with vs 16, Hm p. 5, 11. 5 if. with vs 19; So 18, 20-23; K? 261-263 (Maiik. 6-8); Spl p. 7, 11. 12ff.; Pii p. 4, 11. 18 ff. with vss 5, G; Sy A 2; also in Arabic versions.
The situation is near the beginning of Book I, We have just heard how the bull Samjivaka, abandoned by the caravan, had recovered from his accident and was enjoying himself on the banks of tlie Jnnina, eating his till and bellowing mightily. Now the text proceeds to introduce the lion Piilgalaka, as follows. I quote first the readings of the other texts, tlien those of T, Spl, Pn, and K§.
§18:
SP tasmin vane mrgadhipatib pingalako naina svaviryarjitarajyasukham anubliavann aste. tathS ca (a hi).
PI tasrnin vane piugalakanama sinhah svabhujoparjitarajyasukham anu- bhavann aste. tatha coktam.
Bo tatkalaih c^bhavat tatra nEtidilre vanEntare, sihhab pifigalako nama vikramSkrgntakan anah.
Sy In einiger Entfeimiing von ihm war ein L5we, der jene Ebene im Be- sitz hatte, und bei ihm befanden sich in Menge Sehakale, Fiiclise und wilde Tier© aller Gattungen.~Ar as Sy. vs 4:
(In Sanskrit only in T, Pp; see below.)
Sy Dieser Ldwe war nnklug [so Schulthess by emend.; the ms. reading means ‘^klug”] nnd nnpraktiseh \ef, anitisastrajfle in T, Pn] nnd durcli sein Regiment iibermtitig gemacht [cf. sattvocchrite].— Ar, JCap 39. 19 Erat antem leo inagnanimis [Hebrew probably “proud,” says Beren- bourg] in suis negociis, singularis in siio eonsilio. KF 8. 14 Now this lion was exceedingly haughty in spirit, and whatever he wisht to do, he did independently, without employing the advice of anyone. Not- withstanding, his knowledge was not very perfect, vs 5:
SP, N, H: nabhi^eko na saihskErab siffhasya kriyat© mi-gSib vikramaijitavittasya svayam ova m^rgendrati.
Variants: a, N satkarab- c, SP "jitasattvasya.— For Pp’s reading see below. Cy. So vikramakrantakananab, under § 18; this perhaps represents pada c of this vs. Possibly Sy and Ar also confuse this vs with the precjed- ing.
§19:
SP sa caikadE (« sa kadEcit. so read!) pjpisEkulita udakErth! yamunltim^ a gat (a yamunEkaccham avEtarat, so read!).
H sa eEikadE pipasEkulitah paniyatli patuih yamunEkaccham avEtatat So (20 a b) sa sidho jitu toySrtham Egacclmn yamunatatam.
Not in Pa.
Paficatwitra, II. 6
Chapter IV: Secondary interrelationships of various versions
§20:
SP teiia cinanublmtapurvain akalapralayaglianagarjitam iva saiiijivaka- narditam asravi.
H teiia ea tatra siiihenananubliutam (Hm “ta-purvakani) akiilapralayaghana- garjitam (Hm om pralaya, Hp om ghana, but v. 1. has it) iva saiiijivaka- narditam asravi.
So tasyaran nadam asrausit saitijivakakakudmatah, srutva casrutaptirvaiii taili tannadam diksu mdrcliitam.
Sy Als nun der Lowe imd sein Gefolge die Stimme des Stieres Snzbng hdrten [fiirchteten sie sick, cf. next], wcil sie nocli nie einen Stier ge- sehen, noch seine Stimme gehd-rt hatten.'
Ar as Sy, except that the versions refer only to the lion, not to his attendants.
§21:
SP Srutva ca kimcic chankitamanah (a cakita°) svagatam iilocya (a °cayanj tii§nini athitavan: kirn idam, ko ’treti.
H tac chrutva pamyam apitva sacakitah parivrtya svasthanani itgatyakiin (Hp svagatam for kirn) idam ityalocyayam (Hm alocayans) tusniih sthitah. So {ef. preceding, Srutva &c.) sa sidho ’cintayat kasya bata nado ’yam idrsab, nunam atra mahat sattvaiii kiilicit tisthaty avaimi (Brockhaus apaimi) tat, tad dhi drs^vaiva maih hanyad vanad vapi pravasayet. iti so ’pitapaniya eva gatva vanaih drutam, bhitah siuho nigribyasid Skaram anuyayisu. [This is interesting as one of the rare cases in which So has expanded the text.]
Sy [cf. preceding, fiirchteten sie sich]~~aher in der Erwagung: Mein Gc- folge darf nicht merken, daU ich in Furcht geraten bin, stellte sicli der Lowe furchtlos imd blieb rnhig auf seinem Platze stehen.
Here follows, in all these versions, § 23, introducing the two jackals, Kara^aka and Damanaka.
The version of T and Pn is markedly different from the above; and Spl and K§, while much briefer, apparently indicate that their archetypes agreed with T and Pp. The differences concern in part additions to the text (as I believe), but especially markt alterations in order, which result in a much poorer arrangement of the materials than that indicated by the other, independent versions.
Let us first consider T. 1 italicize the words which literally reflect the common original. T reads:
(§18, beginning) atha [hadaeitf cf. SPof, §19] w^esarvamrgaparivytab
[(?/. last clause of Sy] pingalaJco ndma svhha—
(§ 19) tfrda/cagrahanSrthaiii yamunakacehaw avatitb'sub —
(§20) sarhjiMhxBj^ mahantaih garjitam [so mss., ed. em. garjitasabdam] as^uot. (§ 21) ca im^t'dtlYak^ubhitah^daya akdram [cf. So] achadya mancjala- V afapradese catiu'mandalSivasthanenavas^/iitoii.
Here follows, in our § 22, a section found only in T and its relatives Spl and Pi?, an explanation of the curious teims inti’oduced hy them in § 21
Secondary correspondences between TantrSEkhylyika and Jain versions 83
(which explanation, by the way, leaves ns more in the dark than ever; ohscurum per ohscuriusl). This § 22 mmj be original; that is, its originality cannot be disproved.
After § 22 T proceeds: atha pingalakah — and here follows a