STUDIES IN ORIENTAL CULTURE NUMBER SIX
PHANTASIES OF A LOVE-THIEF: THE CAURAPARCASIKA ATTRIBUTED TO BILHA~A
THE
CAUR...
131 downloads
1102 Views
10MB Size
Report
This content was uploaded by our users and we assume good faith they have the permission to share this book. If you own the copyright to this book and it is wrongfully on our website, we offer a simple DMCA procedure to remove your content from our site. Start by pressing the button below!
Report copyright / DMCA form
STUDIES IN ORIENTAL CULTURE NUMBER SIX
PHANTASIES OF A LOVE-THIEF: THE CAURAPARCASIKA ATTRIBUTED TO BILHA~A
THE
CAURAPA~CASIKA ATTRIBUTED TO
PHANTASIES
BILHA~A
a Critical Edition and Translation
of Two. Recensions with sixteenth-century illustrations of the text
COLUMBIA UNIVERSITY PRESS NEW YORK AND LONDON
~
OF A
~
LOVE-THIEF
Barbara Stoler Miller
Barbara Stoler Miller, Assistant Professor of Oriental Studies at Barnard College, is the translator of Bhartrihari: Poems.
to UNESCO COLLECTION OF REPRESENTATIVE WORKS INDIAN SERIES
This book has been accepted in the Indian Series of the Translations Collection of the United Nations Educational, Scientific and Cultural Organization (UNESCO)
Copyright
©
1971
ia University Press 0-23 1--< It in., 9-10 lines to a side. Caura, following 88 verses of caritam, is verses I-52; 53-63 conclude the legend. This mixed-codex contains verses from Nand WS, as well as readings from different versions (e.g. verses 42 and 49 are variants of verses 2 and 14, respectively); the manuscript is MC4 in the Classification of Manuscripts, below. Begins: SriJ:!. bilhaI).acaritam. No colophon. MAR. Stiftung Preussischer Kulturbesitz Staatsbibliothek, Marburg. One manuscript was examined from a microfilm prepared in Marburg through the kind cooperation of Dr. Voigt. It is a conflated text bearing resemblance to the manuscripts AHM.7 and BORI.5 (R* in the Classification of Manuscripts, below). It is described in Schubring andJanert, Verzeichnis, pp. 171-72 (no. 340). MAR-No. 649 (Orientalische Abteilung). Devanagari script, paper, 4 folios, 131 in. x 8 in., 10 lines to a page. Caura is verses I-s I; verse 52 begins bhavatkrte; verses 53-54 refer to the legend. Begins: srigaI).eSaya nama~. Ends: iti bilha~lapaficasika samapta. bhaXXjayarameI).a likhito 'yam grantha~. srir astu. srisarasvatyai nama~. MEHTA. N. C. Mehta Collection, now with the Gujarat Museum Society, Ahmedabad. Each of 18 extant leaves of this manuscript has one Caura verse (N, type I, version A), accompanied by a painting in early Rajasthani style of the beginning of the sixteenth centl:rry. The leaves have been studied from photographs prepared in [116]
Manuscript Sources
Caurapafidisika
Ahmedabad; these were procured through the kind offices of Dr. Moti Chandra, director of the Prince of Wales Museum; they are reproduced in the Appendix, where the paintings are discussed. MEHTA---'---Devanagari script, paper, 18 folios, 81 in. x 6i in., one verse and painting to a leaf. Caura verses: NIS, 18-20, 22, 24-27,29,32-33,36-38,40,42-43. MY. Oriental Research Institute, M ysore. Nine manuscripts have been examined from transcriptions and microfilms prepared at the library through the kind cooperation of the director, G. Marulasiddaiah. All nine belong to WS, type IV. MY. I-No. MY.2-No. MY.3-No. MY.4-No.
1638. Telugu script, 91 verses of caritam only. 2338. Grantha script, belongs to version X. 104. Grantha script, l:>elongs to version Y. 2106. Telugu script, a conflation of versions V and Y.
MY.5-No. 2 0Grantha 2 I .script }} . . 4300. ' related to the manuscnpt I2BO'}T I . BOM.I (MC6). e ugu scnpt, 4 0 14. MY.6-No. 4235. Devanagari script, has characteristics of version Yand MC6. Caura, following 91 verses of caritam, is verses I-50. Readings, order of verses 10 and I I, omission of verse 33, and eight concluding verses mark its relation to MC6; but N verses which are found in MC6 do not appear here, and verse 50 is Y2.42, not found elsewhere. Note has been made in the critical apparatus of readings which differ from those ofMC6. NAG. Nagpur. University Library, Nagpur. Three manuscripts were examined after Dr. Patkar arranged with the manuscripts officer of the library, Dr. S. M. Ayachit, to have them sent on loan to Poona. They have been collated from microfilms prepared in Poona. NAG. I-No. 779. Devanagari script, paper, 5 folios, 12 in x s!in., I I lines to a side. Caura is verses I-s I; verses 52-63 are verses which are usually associated with the piirvapaiiciisat as concluding verses of the legend. Begins: srigaI).eSaya namal).. Ends: iti sribilhaI).akavina bilhaI).akavyarh sarhpiirI).am astu. chao chao chao srigajanarpaI,lam astu. chao chao chao
[I 17]
Textual Criticism qf the Caurapaficasika NAG.2-No. 780. Devanagari script, paper, 17 folios, 121 in. x 5-1 in., I I lines to a side (in tri-pafha form). Caura, following 74 verses of purvapaiiciiJat (called pilrvacatu~saptatiM, is verses I-50; each verse is accompanied by the commentary which is elsewhere attributed to BhaveSvara. Verses 51-62 conclude the legend as in NAG.1. Begins: srigaJ,leSaya namal:t. Text begins: atha corapaficasika. Ends: iti sribilhal).akavina krtam bilhaJ,lakavyam sampiirJ,lam astu. srigajanarpaJ,lam astu. NAG.3-No. 83g. Devanagari script, paper, 5 folios, 6-1 in. x 3 in., 10 lines to a side. Caum is verses I-50. Begins: srigaJ,leSaya namal:t. Ends: iti srimahdvaraviracita corapaficasika samapta. ox. Bodleian Library, Oxford. Folios 137-55 of the manuscript Chandra Shaun Shere e.g7 have been collated from a microfilm prepared at the library. The manuscript is a collection of works. ox-Devanagari script, paper, 18 folios, I I in. x 6 in., 7 lines to a side. Caura, following 7I verses of pilrvapaiiciifat [which appear also in the Oxford manuscript Walker 20g], is verses 72-122 (number 112 is skipped in the order of numbers) ; verses 123-33 conclude the legend. Begins: srigaJ,leSaya namal:t. Ends: iti bilhaJ,lakrtam kavyam. PAR. Departement des Manuscrits, Bibliotheque Nationale, Paris. Two manuscripts have been collated from microfilms prepared in Paris. PAR. I-No. 704, III two parts. This is a manuscript copied in Poona in 1844 by Charles D'Ochoa, entitled by him Les Cent Stances du Pocte BilhaTJa du Kachemir; the Devanagari script shows influences of ]aina Devanagari. Part A is a conflated codex with text readings bearing relation to type III, version P. Caura, following 72 verses of piirvapancafat, is verses 1-46; each verse is accompanied by commentary. Verse 47 is found also in cop and AHM.6. Seven additional verses, called fafikalaviikyiini, precede part B. Part B is a virtual duplicate of the manuscript BORI.g. PAR.2-No. 705 (part I). Grantha script, palm leaf, 15 folios, about 171 in. x It in., 5 lines to a side. Caura, following 68 [118]
Manuscript Sources verses of caritam, is verses 6g- I 15; verse I 16 begins bhavatkrte; verse I I 7 begins pancatvam; verse I 18 concludes the legend. Entitled: corapaficasad atha va bilhaJ,lacaritram. Ends: with verse 118: bilhal).akavina racitam bahudha srutvaha rajacandro 'pi I tam eva rajakanyam tasmai datvadya sukham anubhaveti The text of this manuscript, with only slight variations, is printed in the edition of Ariel in Journal Asiatique, XI, 469-8g. PU. Manuscripts Collection, University of Pennsylvania Library, Philadelphia. Two manuscripts were examined in Philadelphia and have been collated from microfilms prepared at the library. They are nos. 2I 38 and 2I 39 in Poleman's Census. pu.I-No. 528. Devanagari script, paper, 2 folios, g1 in. x 41 in., 9-23 lines to a side (unevenly spaced). Caura is verses I-50; although the manuscript contains no commentary, the. text is that associated with the GalJapatifikii. Begins: sri mahala~mi jaya. Ends as noted for ASB.6. Colophon (barely legible): brahmavarte sravaJ,laknJ,le 7 sa~ [for sam ?] 1860 agivrekara [?] ityupanamakabalakr~J,labhat!aramajavinayakena likhitam idam pustakam. [Ca. A.D. I803?] pu.2--No. 1524. Devanagari script, paper, I I folios, 13 in. x 5 in., 10-12 lines to a side (in tri-pafha form). Caura is verses I-50, with the commentary of GaJ,lapati on each verse. Begins and ends as noted for ASB.6. Postscript: subham bhavatu lekhakapathakayol).. TAN]. Tanjore Maharaja Serfoji's Sarasvati Mahal Library, Tanjore. Ten manuscripts have been studied from hand-copies and transscriptions in Devanagari prepared at the library through the kind cooperation of the librarian, V. Gopale Iyengar. These fall into three groups, which represent two versions of WS and a mixedcodex. TAN]. I, 2, 3,5,8 have the same text; they are version X2. TANJ.4 and 9 are version W. TANJ.6, 7, 10 are MC6. The manuscripts are described in P. P. S. Sastri's Descriptive Catalogue, V, 2745-53. The manuscripts are here identified according to the serial numbers of this catalogue; Burnell's numbers are recorded [IIg]
Textual Criticism of the Caurapaficasika there. One manuscript listed in the catalogue (No. 3788) is missing from the library. It is noteworthy that eight of the ten manuscripts from Tanjore are in Devanagari script, one with Pr~!hamiitras; this suggests some relation to the Maratha hegemony in Tanjore in the seventeenth century. It is not unlikely that these versions were composed in South India at this time by Maharashtrian brahmins who traveled between the courts of Sivaji and his brother. See P. K. Gode, Indian Culture XII (1945), pp. 47-56; Davies, An Historical Atlas ofthe Indian Peninsula, pp. 48-49. TAN].I-No·37 lk Devanagari script, paper, 16 folios, 9i in. x 4* in., 7- 1 I lines to a side. Caura, following 84 verses of caritam, is verses 1-62. Verses 54-57 are variants of verses 41,31,37, and 23 respectively (i.e., WS 40,31,37, and 22). Verses 58-60 are N22, 21, and 18; verses 49 and 51 are N34 and 19, but these are found generally in manuscripts of type IV; verse 61 begins adyiipi naiva; verse 62 is N 50. Verse 63 begins bhavatkrte; verse 64 begins pancatvam. Verses 65--66 conclude the legend. Begins: srigal)eSaya nama!).. Ends: iti bilhal)akrtau corapaficasatinamagrantha!). samapta!).. TAN].2-No. 3782. Dcvanagari script (with Pr~!hamiitras, according to Sastri), paper, 12 folios, 8i in. x 3t in., 10 lines to a side. Caura, following 84 verses of caritam, is a duplicate of TAN]. I. Begins: srigal)eSaya nama!).. srilak~mivefikateSaya nama!).. Ends: iti kavibilhal)akrte corapaficasatinamagrantha!). samapta!).. srilak~mivei1kateSaya nama!).. TAN]·3-No . 3783. Devanagari script, paper, 15 folios, 8t in. x 3t in., 9 lines to a side. Caura, following 84 verses of cantam, is essentially the same as TAN]. I (minor differences in readings have been taken into account in establishing the text of X2, but these are not incorporated in the Classification of Manuscripts, below). Begins: srigal)esaya nama!).. sr'igurucaral)aravindabhyarh nama!).. Ends: iti corasatakarh samaptam. srinrjanneSarpaJ;lam astu. sriI' astu. TAN] .4-No. 3784. Telugu script, palm leaf, 7 folios, 14t in. x It in., 9-1 I lines to a side. Caura, following 87 verses of caritam, is verses 1-49; verse 50 begins bhavatkrte and verse 5 I begins [120]
Manuscript Sources paiicatvam. This is followed by verses 1-10, which conclude the legend. Begins: sri!).. subham astu. Ends: sri!).. srityagarajaya nama!).. sriramaya nama!).. sri!).. sribhayadarsam anukrtam. TAN].5-No. 3785 [incomplete]. Devanagari script, paper, 9 folios, lOi in. x 4t in., 6-9 lines to a side. Caura, following 84 verses of cantam, is verses 1-12, after which the text breaks off; the text is that of TAN]. I. Begins: srilak~minrsirhhaya nama!).. TAN].6-No. 3786 [incomplete]. Devanagari script, paper, 16 folios, 9t in. x 4i- in., 7-8 lines to a side. Caura, following 88 verses of cantam, is verses 1-26, after which the text breaks off. This is a highly conflated text containing verses from both recensions and readings from different versions; it belongs to MC5. Begins: srigaJ;leSaya nama!).. TAK].7-No. 3787 [incomplete]. Devanagari script, paper, 8 folios (nos. 11-18), 10 in. x 4i in., 9 lines to a side. Caura is verses 90- 138; verse 139 begins paiicatvam. Verses 140-49 conclude the legend (as in MAD.3, TAN]-4, and so on). This is a highly conflated text of the same version as TAN].6 and closely related to MAD.3; it cbntains verses from both recensions and readings from different versions. Ends: iti sribilhaJ;lanatakarh samaptam. TAK].8-No. 3789 [incomplete]. Devanagari script, paper, 9 folios, 91 in. x 4 in., 6-8 lines to a side. Caura, without cantam, is verses 1-61, where the text breaks off; it is essentially the same as TAN].!. Begins: Srigal}.adhipataye nama1:J.. sri subham astu. TAK].9-No. 3790. Grantha script, palm leaf, 4 folios, 18 in. x 11- in., 7 lines to a side. Caura, without caritam, is verses 1-49; verse 50 begins blwvatkrte and verse 51 begins pancatvam. Another verse follows, but the transcription is unclear. The text is quite similar to TAN]+ Begins: veSya sumukhim anime~am alokayan mukharil tasya!). kathayan kathas ca vividha!). kalak~eparh kada kari~yama1).. Ends: iti corapaficasas samaptal). TAN]. I a-No. 3791 [incomplete]. Devanagari script, paper, 6 folios (I6-21 only), 91- in. x d- in., 8 lines to a side. Caura is verses 20-50; verse 5 I begins paiicatvam; verses 52-6 I conclude the legend. The text duplicates that of TAK]. 7. Ends: iti sribilhal)ana.takarh samaptam. srisarhbasadasivarpal)am astu. sri. srir astu. sri.
Textual Criticism of the Caurapaficasika TlR. Sri Veiikatdvara Gniversity Oriental Institute, Tirupati. Two manuscripts have been collated from transcripts in Devanagari prepared at the institute through the kind cooperation of Dr.E. R. Sreekrishna Sarma. TIR. I-No. 6385. Grantha script, palm leaf, 6 folios, 15! in. x I! in., 6-7 lines to a side. Caura is verses I-51 (17 is skipped in the order of numbers); verse 52 begins bhavatkrte and verse 53 begins paiicatvam. Begins with verse I. Ends: iti caurapaficasat samaptam. TlR.2-No. 7405. Telugu script, palm leaf, 3 folios, 17 in. x I in., 8-9 lines to a side. Caura is verses I-58; the text is essentially the same as TANJ.I. Verses 51-53 are variants of verses 39, 30, 21 (i.e. WS 40,31,22). Verses 54-56 are verses 22, 21, 18 of N (cf. TAl'iJ.I). Verses 57 and 58 = TANJ.I. 61 and 62. Verse 59 begins bhavatkrte and verse 60 begins paiicatvam; verses 63-69 conclude the legend. Begins with verse I. Ends: iti corapaiicasatakaril sarilpiirl}.am. sril).. WAI. Prajiiapathasala, Wai (Satara District, Maharashtra). Two manuscripts, examined in \Vai and collated from microfilms photographed there. The pandits of the pii!hafiila are thanked for their cooperation. WAI.I-No. 8g02. Devanagari script, paper, I I folios, 8 in. x 4 in., 12 lines to a side. Caura, following 72 verses of pilrvapanciifat, is verses i3-121; verses 122-31 conclude the legend. Begins: srigal}.eSaya nama!).. Ends: iti caurapaiidisika samapta. chao sake 1720 kalayuktasarhvat marge. suo IO cittalopabhidhananasarmal}arh idarh kirhcit likhitaril likhapitaril ca. [Ca. A.D. I7g8.] WAI.2-No. 8g03' Devanagari script, paper, 32 leaves, I It in. x 5 in., 7 lines to a side. Caura, following 74 verses of purvapaiiciiSat, is verses I-54; verses 55-59 include bhavatkrte; verses 60-6g conclude the legend. Following the colophon, iti bilhal}.akavyarh samaptam, is the type IV caritam in 83 verses. The text is highly conflated and careless, but is clearly based on a manuscript close to AHM.5. Begins: srigal}.eSaya nama!).. Ends: iti bilhal}.akavyaril samaptam. srihariJ:1. chao [ 122]
---~
3. PRINTED EDITIONS
~,---
ARIEL, M. "Les Cinquantes (Couplets) de Tchora ou Histoire de Bilhal}.a," Journal Asiatique, 4th ser., vol. XI (Paris, 1848), pp. 46 9-534. The text here is based on the manuscript PAR.2.
BOHLEN, PETRUS VON. Bhartrharis Sententiae et Carmen quod Chauri Nomine Circumfertur Eroticum. Berlin: Ferdinand Duemmler, 1833, pp. 1-20. The text here is accompanied by the Gavapatilika; it is based on a manuscript of type I, version B.
HAEBERLIN,]. Kiivyasamgraha. Calcutta, 1847, pp. 227-36. The text here, entitled "Caurapaiicasika," is based on a manuscript, or a printed edition ofa manuscript, of type II, version D.
SIVADATTA, P., and K. P. PARAB. Kavyamala, no. 13. Bombay: Nirnaya Sagara Press, Ig03, pp. 145---69. This is a conflated text, probably based on manuscripts of type III, versions S and T, or on an already conflated manuscript or edition.
SOLF, WILHELM. Die Kafmir-Recension der PaiiciiSikii. Halle: Friedrichs-Universitat Halle Wittenberg, 1886. The text is based on the manuscript BORI. I.
TADPATRIKAR, S. N. Caurapaiiciifikii: An Indian Love Lament Bilha1Jakavi. Poona: Oriental Book Agency, Ig66.
of
The Caura text is based on the manuscripts BORI.4and BORI.g, both ofwhich belong to type I, version A. Tadpatrikar also includes appendices with different versions of the Bilhava legend.
_ _ _ _ ~ 4. CLASSIFICATION ~1--- OF MANUSCRIPTS The manuscripts which have been collated to establish the critical edition of the Caura are here grouped according to their relationships. The broadest grouping is into recensions; the manuscripts are divided into a Northern Recension (N) and a Western-Southern Recension (WS). The manuscripts then divide into types (designated [ 12 3]
Textual Criticism
of the
Caurapafidisika
I, II, III, IV), which represent two basic subdivisions of each recension; these are generally found to fall into definable geographic regions. A type is considered to be "located" in the region where its codices occur in the greatest concentration; this has been corroborated by considerations of script and dialectal variants. The four types are further subdivided into versions (designated A, B, D, and so on); this grouping is based on concordances and similarities of readings, omissions, legend verses, commentaries, and the like. I It is worth noting that all versions are not equally homogeneous; this is apparent in the descriptions which follow. The usefulness of the version classification is that readings generally fall into these classes and that a well-established version can often be located geographically by its colophons, vernacularisms, and scribal errors. When a manuscript of a given version is copied in a territory belonging to some other version, contamination usually occurs from the local text. The versions provide a framework, at least, in what would otherwise be haphazard collections. The final division is into codices (designated AI, A2, A3, and so on). Most codices are represented by only one manuscript, but there are some which are represented by several copies. Besides this fourfold division into recension, type, version, and codex, there is also a category of mixed-codices. There are six different mixed-codices in the Caura apparatus; these are designated MCI-6. 2 They are characterized by their redactors' use of verses from both recensions and readings from different versions to constitute the texts. There are other conflated texts which are also clearly mixed-codices, but have not been considered independently because of the relative limitation and transparency of the conflations; these are generally closely associated with a particular version and are designated by an asterisk plus the version designation, e.g. P*. I The versions are so designated that the first version listed under each type represents the most extreme statement of the type, i.e. versions LA. and n.D are the most widely different from one another, both in readings and geographical location; likewise HLP and IV.V. This was done to avoid the notion of necessary continuity between less extreme versions of each type. 2l'vICI, the manuscript BORI. I, in Sarada script, is the codex which is referred to as the" third" recension in the literature. That there is no basis for this should be clear from the critical apparatus.
Classijication
of Manuscripts
The sequence of verses within each of the recensions is fairly consistent. The critical text of the Northern Recension follows the order of type I, version A; that of the Western-Southern Recension follows the order of type III; divergences will be noted only when they are relevant to the relations of manuscripts. Omissions and addenda will be noted as they occur in various versions.
N- THE NORTHERN RECENSION Manuscripts belonging to this recension are known from Maharashtra to Punjab in Western India and as far east as Bengal; the majority are written in Devanagari script, but examples also abound in Bengali, and there is one codex in Nepali script. Stray manuscripts are found in South India, but they are clearly importations (e.g. KER.3 belongs to versionA; ADY.5 belongs to version D). TYPE I: VERSION A This version seems to have originated in the region of what is now Maharashtra and Gujarat. It is known only from manuscripts in Devanagari script, the oldest of which is the illustrated manuscript MEHTA, which is datable to the early part of the sixteenth century. This- is the only version of N whose manuscripts either attribute the Caura to BilhaJ).a or make no mention of the author. The readings of these manuscripts generally represent the most conservative tradition in N. A I = BORl.3; also ANA, BISM. I. Verse 5 I is the verse bhavatkrte; it does not otherwise appear in N, but is extremely common at the end of WS manuscripts. A2 = BORI.4; also DC.5. A3 = BORI.g; also PAR.I (part B). The text is accompanied by the commentary ofBhaveSvara. A4 = (a) NAG.2, (b) BAR.2. The Caura texts are identical and each is preceded by the pilrvapaiiciisat version of the BilhaJ).a legend, which is otherwise found only in manuscripts of WS, type III. The two manuscripts differ only in that NAG.2 has the BhaveSvara commentary and BAR.2 does not. AS = (a) ASB-4, (b) ASB.3, NAG.3. The Caura texts are identical; the manuscripts differ in that ASB.4 has the BhaveSvara commentary while ASB.3 and NAG.3 do not. [ 12 5]
Textual Criticism
of the
Caurapafidisika
A6 = BAN+ A 7 = MEHTA. The text is incomplete, but extant verses show a marked relation to codex AI. The poet-lover depicted in the paintings is labeled Bilha7Ja.
Classification TYPE
F
This is the Gal).apati version of the Caura text, accompanied by his {ikii. The text, which is repeatedly emended by the commentator at points of difficulty or obscurity, is based on that of version A. The codices of this version show remarkably few variants; this may be explained by the clarity of the text.
TYPE
II:
VERSION D
This is the easternmost version of N. Manuscripts are written III Bengali script and the poem is attributed to Sundara. Verses 6, 29, 40, and 43 of N are omitted and alternative verses substituted. D I =ADY.5. Known from a transcript into Devanagari of a Bengali manuscript in the Adyar Library, dated ca. A.D. 1820. D2 = BAN.6; also 10+ D 3 = ASB. I ; also ASB.2, 5, 7. The Caura text is 48 verses only. TYPE
II:
VERSION E
The two codices of this group are quite independent of each other, but their readings show enough similarity with one another and independence of other versions to infer their relation to a single hypothetical source. EI = BAN.5. This manuscript, in Devanagari script, agrees in many places with the readings ofD. Verse 47 of Nis omitted; verse 48 is repeated. E2 = 10.3. This manuscript, in Bengali script, shows sporadic agreement with version F. Variations in the sequence of verses 20-30 are the same as those of version B. The poem is attributed to Sundara. [126]
VERSION
F
Represented by a single manuscript, in Nepali script.
TYPE I: VERSION B
B = ASB.6, BOM.2, BORL8, 10. I, PU. I, PU.2; also the printed edition of Bohlen, which has many errors.
II:
of Manuscripts
KAT. The manuscript contains 57 verses; the Caura is verses 2-56. The manuscript has characteristics in common with version A and version D, but there are many readings and several verses which are not known from either of these; verses 52 and 54 do not occur in any other manuscript examined and verse 57 is found elsewhere only in AHM.7 (WS, type III, R*).
=
TYPE
II:
VERSION
G
Represented by a single codex of two manuscripts, in Devanagari script. The majority of readings classify this codex in type II, but there is sporadic agreement with readings of type I, version A. The poem is attributed to Sundara. G
=
BAN.2 and HOS.2. TYPE II: PSEUDO-VERSION H
Under this have been gathered three Devanagari manuscripts that have many common features without sufficient agreement of readings or verse sequence variations to form a version. The basic unity of the group derives from the shifting pattern of agreements and similarities of readings among the manuscripts. All three also suggest origin from version A, but each is strongly divergent enough to suppose a series of intermediary codices. It is my hypothesis that as the Caura traveled northward and eastward from the region of origin in Maharashtra and Gujarat, certain features of these and related manuscripts were incorporated into other manuscripts, which proceeded to become better established through protected transmission, i.e. versions G and D. This notion will be developed below. HI = HU.2. H2 = BORI.6. This manuscript omits verse N 15 and substitutes one of the alternative verses otherwise found only in version D manuscripts. The poem is attributed to Sundara. H3 = BAN.I.
Textual Criticism
of the
Caurapaiidisika
WS-THE WESTERN-SOUTHERN RECENSION Manuscripts belonging to this recension are concentrated in two regions. Type III manuscripts, written in Devanagari and its variant, Jaina Devanagari (see AHM, above), are known from Maharashtra and Gujarat. These manuscripts, with the exception of R4 (DC.2), give the title BilhaTJapaficiifikii to the poem. Mixed-· codices closely related to this type are known from Punjab and fromJammu and Kashmir. Type IV manuscripts, writteninGrantha, Telugu, and Devanagari (as well as one manuscript in Kanarese), are known from Madras, Andhra (Tirupati only), Kerala, and Mysore. TYPE III: VERSION P This is generally the most conservative version of WS. The two known manuscripts, both in J aina Devanagari script, show many difficult readings which could be the basis for variants found in related versions; there are relatively few attempts in either manuscript to emend the text for the sake of simplicity or clarity. Allowances must bemade in reading these manuscripts for confusion between long and short vowels and between dental and palatal sibilants, as well as for carelessness with regard to the anusvara and visarga signs. Such allowances must also be made in reading manuscripts of versions Q, R, S, and, to a lesser extent, T.
Classification of Manuscripts translation is attributed to Jiianiidirya and is said by Dr. Sandesara (see Sahitya, XX, 385-400) to date to the sixteenth century. The manuscripts are written inJaina Devanagari script and all omit verse WS 21. In each manuscript, the Caura and its translation are preceded by a translation of the piirvapanciifat legend into copais and dohas of Old Gujarati; they are followed by several Old Gujarati verses which conclude the legend. A fourth codex, iri Devanagari, is somewhat independent of these three in that it does not contain the Old Gujarati translation and does not omit verse WS 2 I, but readings made the common ancestry of the Caura text seem certain. All four codices conclude the Caura with the verse N 50. QI = AHM.I. Dated ca. A.D. 1720. The Caura is in 50 verses; 4 additional Sanskrit verses refer to the legend. Q2 = AHM.2. Dated ca. A.D. 1658. The Caura is in 50 verses; Sanskrit verse 5 I refers to the legend and 52 is the verse bhavatkrte. Q3 .= AHM.6. The Caura is in 48 verses; verses 46 and 48 of WS are omitted. Sanskrit verse 50 refers to the legend. Q4 = BAR. I; also BAR.3. The Caura, following 7I verses of the pilrvapanciifat, is in 50 verses. TYPE III: VERSION R
TYPE III: VERSION Q. In three of the known codices of this version, each Caura verse is followed by a translation into two copais of Old Gujarati; the
The unity of this version is based on patterns of concordant and related readings. All codices include the verse N 50. One manuscript is written in Jaina Devanagari script; the others are 10 Devanagari. RI = BORI.2. Written in Jaina Devanagari. The Caura is in 51 verses, each followed by an anonymous commentary in Prakrit. Verse 52 begins bhavatkrte. R2 = HU.I. The Caura is in 51 verses (nos. I-50, 52); verse 51 begins bhavatkrte and verse 53 refers to the legend. R3 = DC. I [incomplete: first folio missing]. Dated ca. A.D. 1636. The Caura is in 52 verses; verse 5 I is N 7 (the readings agree with those of M C4, 5, 6 for this verse) . R4 = DC.2. The Caura is in 5 I verses. R * = BORL5 and AHM. 7. This is a conflated text, apparently based on more than one manuscript of version R; it shows agreement in a number of cases with }'1CI, 2, 3. Variants are indicated only when they are ofspecial interest.
[128]
[ 12 9]
PI = AHM.4 [incomplete; the text breaks off after the seventh ak~q,ra of pada c of verse 46]. P2 =AHM.8. The Caura is in 46 verses only; verses 17,34,49, and 50 of WS are omitted. The 47th verse (no. 49 in the manuscript) begins bhavatkrte; this verse is common to many manuscripts of WS and is almost always found immediately following the final Caura verse. P* = PAR.I (part A); also cOP. This is a conflated text which is clearly based on a manuscript of version P, mixed with readings and additional verses from versions Sand T; unless otherwise noted readings are those of PI.
Textual Criticism
cif the
Caurapai'icasika
TYPE III: VERSION
s
The version is comprised of three codices in J aina Devanagari script and a mixed-codex in Devanagari. They are characterized by inversion or omission of plidas c and d of verse WS I 7 and by the presence of the verse adyiipi tlim kamalapattravifiilao at the beginning of the Caura. S I shows a lacuna where this verse would have been, suggesting that the copyist knew it, but chose to omit it for some reason. SI = AHM.3 [incomplete]. The Caura is 46 verses only; the text breaks off after the twelfth ak~ara of piida c, verse 46. It is preceded by the piirvapaiiciifat in 72 verses. Verse 17 reads a, b, d, c. S2 = AHM.5. The Caura is in 51 verses, each accompanied by commentary in Sanskrit. Verse 17 reads 'a, b, c, omitting plida d, but the commentary has the text. S 3 = AHM.9. The Caura is in 49 verses; verses 18 and 26 of WS are omitted. It is preceded by the piirvapanclifat in 72 verses; 10 additional verses conclude the legend. Verse 17 reads a, b, d, omittingpiida c. S* = DC.4. This is a mixed:codex whose readings show a conflation of texts from versions R, S, and T. The piida sequence of verse 17 (a, b, d, c), the presence of the verse adylipi tlim kamalao, and the majority of readings associate it with version S. Variants are indicated only when they are ofspecial interest. TYPE III: VERSION T
This version is characterized by the presence of the verse adylipi tlim kamalao at the beginning of the Caura (cf. version S). All known manuscripts are written in Devanagari script and all contain the same verses concluding the BilhaQ.a legend. T I and T3 also have the piirvapaiiclifat, which is not in T2. Readings show sporadic relation to those of type IV manuscripts. TI = DC.3; also WAI.I. Dated ca. A.D. 1794 and 1798, respectively. The Caura is in 49 verses; verses 39 and 41 of WS are omitted. T2 = NAG.I. The Caura is in 50 verses; verse 27 of WS is omitted. T 3 = ox. The Caura is in 50 verses; verse 28 of WS is omitted. [13 0 ]
Classification
cif Manuscripts
TYPE IV
The Caura text of codices from South India is distinct from that of type III in several ways. It omits verses 48-50, found in most complete codices of type III. It adds three verses which are known (rom all versions of N: 7a + 6b-d, 19, and 34; these are found within the final ten verses of each type IV codex. The inclusion of the verses from N probably occurred early in the develop~ent of the type, for the variant readings of these verses in type IV manuscripts are diverse and unrelated to variants within N. TYPE IV: VERSION V
This version, known from manuscripts In Grantha and Telugu scripts, is characterized by a Caura in 47 verses which omits verses 9, 13, 20, 23, 25-30, 33, as well as 48-50, of fVS. Eleven alternative verses occur; these include the 3 verses from N. The Caura text is usually preceded by the caritam, or South Indian account of the BilhaQ.a legend, in approximately 70 verses; it is followed by the verses bhavatkrte and paiicatvarh. The redactors of this version were notably liberal in emending the text for the sake of clarity and simplicity. The version represents the "maturity" of the South Indian type, since it developed in relative isolation from continued comparison with versions of type III. V I = PAR.2. Written in Grantha script. The manuscript MAD. D.No.I 1976, written in Te1ugu script, is a duplicate of this text. V 2 = ADY.2 [incomplete; the Caura text breaks off in verse 34]. Written in Telugu script. V3 = ADY-4 (also consulted: manuscripts MAD.R.Nos.539b and 1878). Written in Telugu script. The Caura is not preceded by the caritam and each verse is followed by an anonymous commentary. V 4 = MAD.2 [incomplete; the Caura text breaks off'in verse 40]. Written in Grantha script. TYPE IV: VERSION W
This version is characterized by a Caura in 49 verses which omits verses 13 and 33, as well as 48-50, of fVS. Four of the alternative [13 1 ]
Textual Criticism
if the
Caurapancasika
verses found in version V are also found here, as are the verses beginning bhavatkrte and paficatvarh.
Classification
if Manuscripts
of this codex (found as no. 50 in the manuscript MY.6) is a variant of N 7 :
WI = TANJ+ Written in Telugu script. The Caura is preceded by the cantam in 87 verses; 10 additional verses conclude the legend. \-\12 = TANJ.9 (also consulted: manuscript KER.2). Written in Grantha script. The Caura text is not preceded by the caritam.
adyapi kundavadanaIh kucabharanamraIh piirQ.endusundaramukhiIh mama vallabhaIh ca antarvisalanayanaIh hrdi me prasannam alokakuntalakalapavatiIh smarami II TYPE IV: VERSION
TYPE IV: VERSION x This version is characterized by two confl,!-ted codices, the common parent ofwhich is independent of any other known version. For this reason and because the "conflation" seems to have resulted in the addition of verses rather than the blending of readings, the codices are considered to be worthy variant bearers. Besides the three verses of N which are common to versions of type IV, three others are found here: N 18,21, and 22. Several verses within WS are repeated in variant versions, which agree with readings of other type IV versions. The verses bhavatkrte and paiicatvarh also occur. XI = TIR.2. Written in Telugu script. The Caura text in 58 verses is not preceded by the cantam. Verses 13, 36, and 4 I, as well as 48-50, of WS are omitted. X2 = TANJ.I; also TANJ.2, 3, 5, 8. Written in Devanagari script. The Caura text in 62 verses is preceded by the caritam in 84 verses.
Z
This version is not as well determined as others of type IV; it is represented by only one complete manuscript, one manuscript of eighteen verses only, and a highly conflated manuscript which blends features of this version with those of version X. All are written in Grantha script. The version is notable for the sporadic concordance of readings with those of type III in places where type IV manuscripts are otherwise in agreement, e.g. 6a, 12d. Neither ZI nor Z2 has the caritam. TIR. I. The Caura is in 50 verses. Verses 4, 2I, and 3 I of WS are omitted; WS 13 is omitted in sequence, but is no. 48 in the manuscript. The same four alternative verses which are found in version Ware found here; in addition, there is one verse found elsewhere only in version X, and also the verse N 50. The verses bhavatkrte and pancatvarh follow the Caura. Z2 = MAD.I [incomplete; the Caura text breaks off in verse 18]. Verses 4, 13, 19, and 21 of WS are omitted in sequence before the text breaks off. Z* = KER.I.
ZI
=
TYPE IV: VERSION Y The relation between the two manuscripts which make up this version is based on the pattern of concordant and unique readings. YI = ADY.I. Written in Tel.ugu script. The Caura text in 53 verses is preceded by the caritam in 88 verses. Verses 19 and 22 of WS are omitted; alternative verses found in version V are also found here, as are the verses bhavatkrte and paficatvarh. Y 2 = ADY.3 [incomplete: the Caura, following the caritam in 79 verses, is in 42 verses only, after which the text breaks off]. Verses 19, 27, 28, 29, and 33 of WS are omitted. Verse no. 42 [13 2 ]
MIXED-CODICES This' group of six mixed-codices is characterized by texts which blend verses from both recensions as well as readings from different versions. ~1CI, M.C2, and MC3 are composed from manuscripts of type III of WS and both types I and II of N. MC4, MC5, and MC6 are composed from manuscripts of types IV and II. All show liberal emendation of the text at points of difficulty. MCI = BORI.I. Verses 3, 5, 35, 37, and 38a-b + 7c- d belong to both recensions. Verses 6, 3-34 are WS 2, 7-33; verses 40-49, [133]
Textual Criticism
if the
Secondary Evidence
Caurapaficasika
51-56 include N 9, 10, 16- 19,22-25,30,3 6,37,45,47,4 8,5 0 . Verses 4a-c [d = WS/N Id], 36, 39, and 50 are not found in known manuscripts ofeither recension, though verse 36 is found in several anthologies. MC2 = (a) JAM, (b) HOS.I. Verses 1-6,8-41 are WS 1-41, with verse 28 of WS omitted; verse 42 is WS 50. Verses 7, 43-50 are N 16, 43-50. MC3 = BORl.7; also Dc.6. The 88 verses ofthis mixed-codex are a blend of verses from both recensions. The codex omits verses q, 33,44,49, and 50 of WS; verses 37,39, and 44 of N. MC4 = MAD.3. This mixed-codex of 52 verses also blends verses from both recensions. I t includes verses 1-7, 9- II, 15- I 7, 2 I, 23, 25, 28-3 1, 45-47 of WS; verses 1-14 (five of which are in both Nand WS), 16-19, 22-25, 28, 34, 45-48, and 50 of N; verse 2 (= WS 2) and verse 14 (= N 19) are repeated in variant versions. MC5 = TANJ.7; also TANJ.6, TANJ.IO. This mixed-codex of 50 verses blends verses from both recensions. It is closely related to MC4. It includes verses 1-5, 7-12,14-17,20,21,23-31 of WS; verses I, 2,4-1 I, 13, 14, 16-19, 21, 24, 25, 28, 34, 42, 45, 4 6, 48 , and 50 of N. MC6 = BOM.I (compared with manuscripts MAD.R.No.902b and MY.6). Verses 1-44, 46 are WS 1-47, with WS 33 omitted; verse 42a-b = WS 43a-b and 42C-d = WS 44C-d. Verse 45a-b = N 22a-b and 45c-d = N 19c-d. Verses 47-52, 54-57 include N 5-7, 13, 17, 25, 28, 34, 35, 50. Verse 53 is not found elsewhere.
vardhana (p. 60) and of the entire verse, without attribution to any author, in Kuntaka's Vakroktijivita (p. 60), where it reads as follows: nidranimilitadrso madamantharaya~ napy arthavanti na ca yani nirarthakani adyapi me varatanor madhuralJ.i tasya~ tany ak~af(llJ.i hrdaye kim api dhvananti
I II
The order of the piidas in BORl.1 is c, d, b, a and the verse reads: adyapi me varatanor madhuralJ.i tasya~ yany arthavanti na, ca yani nirarthakani nidranimilitadrso madamantharayas tany ak~aral).i hrdaye kim api dhvanti II
I
S. K. De claimed with reference to the Caura that the text had existed "in some form already in the IOth century" (Dasgupta and De, A History oj Sanskrit Literature, I, 369nI). This was based on the existence of piida d of verse 36 of the manuscript BORI.l (MCI) in Abhinavagupta's commentary on the Dhvanyiiloka of Ananda-
The essential point is that the verse appears only in this one manuscript, which is highly conflated. The verse also appears in the Kashmiri anthology Siirfigadharapaddhati (Petersen, ed.) attributed to BilhalJ.a and it is likely that the redactor of MCl took the verse from there or some similar source. The "existence of the text in some form already in the loth century" is doubtful, though the verse-type was known. A single verse of the Caura (N 12, WS 3) is found, without attribution, in the Calcutta edi.tion of Bhoja's SarasvatikaT}!hiibharalJa (p. 52, verse 1.152), but not in any of the other editions checked. Most notable is its absence in the Madras edition. Since the readings in the Calcutta edition are the same as those in the Siirfigadharapaddhati, it is likely that the verse was at some point drawn from this or a related anthology. The oldest anthology in which verses of the Caura are found is the Siiriigadharapaddhati. The date of this work is generally given as ca. A.D. 1363. Of the verses attributed to BilhalJ.a, two are found in both recensions of the Caura: No. 3469 = N I and WS 1; No. 3470 = N I I and WS 5. No. 3468, attributed to BilhalJ.a, appears only in the manuscript BORl. I, no. 36; it reads here as in the Vakroktijivita version quoted above. No. 3467, which is attributed to Amam rather than BilhalJ.a, is N 12, WS 3. Other verses attributed to BilhalJ.a are found in the Vikramiifikadevacarita. Verse no. 2 of WS appears in the Long Recension ofW. Norman Brown's critical edition of the Vasantaviliisa. It is one of two Sanskrit
[134]
[135]
--,~
5. SECONDARY EVIDENCE
~--
Textual Criticism qf the Caurapaiidisika
Relationships of Versions and Determination of Readings
verses accompanying the Old Gujarati verse no. 59; it is no. 60 in Professor Brown's manuscript S, the illustrated scroll in the Freer Gallery, which is dated A.D. 145I. The text of WS, type III, accompanied by the purvapaiicafat, was translated into Old Gujarati verse in the sixteenth century by]iianadirya. The translation is found in the manuscripts AHM.I, AHM.2, and AHM.6, which are QI, 2,3 in the Classification of' Manuscripts, above. Three verses of N (I, I I, and 50) are found in chapter 28 of Bharatchandra Ray's Viqyasundara (Bandyopadhyay and Das, eds., pp. 300-I), the eighteenth-century Bengali poem based on the legend of the poet and the princess. Verse 50 of N, which also appears in manuscripts of WS, is verse no. 202 of D. D. Kosambi's critical edition of the Satalcatrayam of Bhartrhari (p. 8 I). It is also found in Otto Bohtlingk's Indische Spriiche, verse no. 203 (1,39).
method of textual criticism to this text, for unless we could disentangle the versions sufficiently to establish certain of them as absolutely independent, it would be impossible to base the text's "originality" on what is common to two or more "independent" versions. The Caura case is further removed from the province of the strictly genealogical method by a factor which operates between the two recensions and within each of them, a factor which we may call "independent creation." It has been pointed out that the two recensions share in common only five verses. 2 Such diversity makes it impossible to speak of an "archetype" of the two recensions. The recensions are therefore dealt with separately. A variety of speculations could be put forth in an attempt to account for the two sets of verses. It is possible to imagine that a poet, whom we have tentatively decided to call BilhaI).a (see above, Introduction), himself recited different verses based on the aqyapi ... smarami formula. The theme and form are focal points of the texts; these were maintained, as were a few of the verses, to set the tone, and then the poet ranged freely within the pattern. Assuming that this poet did exist and recite his verses, further alternatives are possible: (I) BilhaI).a never composed paiiciiSikas, but merely recited his lyrical poems and a few were recorded, later giving· rise to each of the recensions; (2) BilhaI).a himself, or his followers, did collect the verses into a single pafit;ajilca, and one extant recension represents old and "original" verses, while the others are "derivative," or (3) the verses were collected into paiicajiktis by BilhaI).a, and the two extant recensions are survivals of these. It is impossible to decide on the basis of the available evidence.
~
6. THE RELATIONSHIPS OF THE VERSIONS AND THE DETERMINATION OF THE READ INGSI
~~
Although a grouping of manuscripts of the Caura text into recensions, types; and versions has been effected on the basis of internal evidence, the interrelationships of these groups are not easy to determine. The study of the manuscripts themselves indicates that interrelationships of versions are of a particularly complex nature. In the case of a secular work like the Caura the complexity is attributable, at least in part, to the absence of any reason to preserve the text in a particular form and the resultant indiscriminate conflation and inflation of versions through the centuries. Such widespread interpolation would itself militate against the strict application of the genealogical I This consideration of the problems has been guided by the following discussions: Sukthankar, Iv[ahiibhiirata, "Prolegomena," pp. lxxvi-xcii; Katre, Indian Textual Criticism, pp. 35-62; Kosambi, Bhartrhari, pp. 7 1 -75.
2 This includes a pada-by-pada analysis of each recension to determine possible points of contact on the basis of a repeated pada. The occurrence of three additional verses in both versions of Nand WS, type IV has also been considered, as has the occurrence of N 50 in many WS manuscripts (see above, Introduction). The three verses are 7a + 6b-d, 19, and 34 of N; they always occur at the end of It'S, type IV manuscripts, often along with other verses which are clearly borrowings from other manuscripts. The suggestion is that the verses were" appended" to an already developed and fairly complete text and that this was probably some text of type III, upon which type IV built. Just why these particular verses were used is impossible to say. It is also impossible to "prove" that they do not belong to some lost uT-text and are" preserved" in Nand WS, type IV.
Textual Criticism of the Caurapafidisika What is the case between the two recensions is also true of other relationships. We are not, even within each recension, dealing with a situation in which a concatenation of copies and exemplars reaches back ultimately to a single "authentic" archetype, which can be reconstructed by eliminating certain" errors." We may be able to reconstruct the hypothetical parent of two or more codices, but at the level of different versions transmission is usually too uncontrolled for this process to be effective. Because the Caura model is so easily imitated and varied, verses could be omitted and others substituted, either by conflation or by "creation," and verses could be readily altered through word and phrase changes. Given this situation the determination of the texts of the two recensions on the basis of a priori criteria is unsatisfactory. The criteria which were followed in editing the Caura have, therefore, been developed and tested during the examination of the manuscripts themselves. After having collated the manuscripts and grouped them into versions on the basis of recurring patterns of concordant readings, omissions, and other internal peculiarities, several noteworthy features appeared. The first was that the readings of manuscripts of versions A of Nand P, Q, R of WS are generally those which could best account for the readings of other versions. In repeated cases the readings of the other versions are attempts to deal with difficulties or obscurities in these versions. Versions Band D of N are particularly liberal in their conflations and emendations; version D even omits several somewhat troublesome verses (N 6, 29, 40, and 43) and substitutes others in their places. The same liberal tendencies are found in versions V, W, and X of WS. When this observation is considered in relation to the geographical distribution of manuscript versions, it is striking that versions A and P, Q, R are all found to be concentrated in the region ofGujarat and Maharashtra, while versions D and V, W, X are respectively found concentrated in Bengal and Madras. 3 When presented with such 3 Since manuscripts are not always found in the region in which they were written, script differences have been taken to be a stronger indication of origin than place of deposit. While manuscripts written in such scripts as Sarada, Nepali, Bengali, Grantha, Telugu, and J aina Devanagari are generally found in certain areas and manuscripts written in them are likely to have been composed in these areas, this is not true of manuscripts in Devanagari. With regard to manuscripts in Devanagari, the antiquity of a codex's association with a certain
[13 8 ]
Relationships
if Versions and Determination if Readings
·JAM'''''' o Hoskilrp.....
The Distribution oj Caura .Manuscripts
a picture, supplemented by the fact that the earliest datable manuscripts of each recension are known from Gujarat and areas bordering on it (MEHTA is probably located in Mewar or Malwa), the notion of a "region of origin" for the Caura verses suggests itself. This is illustrated in the accompanying map. On the basis of an analysis of readings in the manuscripts and the area is determined, if possible, through information in colophons and, in the absence of this information, by the quantitative occurrence of manuscripts of it and related codices.
[139]
Textual Criticism cif the Caurapafidisika
Relationships cif Versions and Determination cif Readings
loci of these manuscripts, it is suggested that each of the recensions developed in migration. N spread northward and eastward from the region of origin; WS spread southward. This is represented here in a greatly simplified diagram.
of variations in the manuscripts arise from slips of the pen or of the tongue.4 Some of the commonest orthographic and phonetic confusions are noted on the following pages. Most of these present no particular problem and have not been recorded in the critical apparatus unless the reading is in doubt.
REGION OF ORIGIN
SOME RELEVANT ORTHOGRAPHIC
of both recensions of the Caurapaficiifikii
I I
PECULIARITIES
I I
N
The orthographic confusions listed below are responsible for the commonest variants. s
WS
jfl--------~M-------_F~~;~:~:,~:::~ DEFGH
B
A
PQRST
ZYXWV
indicates lines of conRation \Vhat began as a series of observations based on the nature of specific variants and patterns of variant readings has been generalized into a working hypothesis of the origin of the manuscript traditions of both recensions in the Gujarat and ]\;Iaharashtra region. The texts in general determine themselves by the concordance of the overwhelming majority of manuscripts. In places of difficulty, readings of manuscripts at the extremes of each recension rarely agree; there is usually a range of variants. In many cases one reading can be discovered to be the basis of the others and in most of these cases, as has been indicated, this reading is found in manuscripts of versions A and P, Q, R. When the situation is less clear, and no other means is available for choosing among readings, the "best" reading derived from A or P, Q, R is provisionally taken. At the level of choosing the "best" of related variant readings within a manuscript version or type we are again faced with the task of discovering the reading which could best account for the others. The mechanisms of change are manifold, but the majority
[140 ]
WITHIN DEVANAGARI
rava-kha (but metrical considerations exclude this her~) va-ba va-na-ta va-nu (less common) va~a
pa-ya-ma-sa (which in turn can be pronounced fa, which can be confused with tra) su-sva pu-pha tta-tra The instability of subscript and superscript vowel signs (u, ii, e, ai, 0, au) can also give rise to variants. JAINA DEVANAGARI
There is a frequent substitution of sa for fa (a confusion which is phonetic in basis); this in turn may be reflected in a confusion of sa with mao There is carelessness regarding anusvara and visarga. There is inconsistency in the representation of vowel signs; for example a single manuscript may alternate between Tifi and ifi for ke. This may lead to confusion when the manuscript is being transcribed into Devanagari or some other script. 4 cr. Sukthankar, Mahiibharata, pp. xxxvii-xxxviii; Katre, Indian Textual Criticism, pp. 55--63; Kosambi, Bhartrhari, pp. 72-73· 5 See Renou, L'Inde classique, pp. 691-7°1; Buhler, Indische Paleographie.
Textual Criticism oj the Caurapaficiisikii
Variant Readings and Related Notes
There is the common interchange of ~a and kha, and ofya andja, which is also phonetic in basis. There is the peculiarity of certain Jaina Devanagari characters, which bear resemblance to other characters in Devanagari and therefore often cause confusion:
anusvara is used to represent final m before mutes and nasals in external sandhi, though some manuscripts use the class nasal. 6 Although the piidas of each hemistich have been separated, they are considered a unit insofar as the application of consonant sandhi is concerned. Because most manuscripts do not separate the piidas of a hemistich and because the insertion of anusvaras is often careless, there are several ambiguous cases with regard to the existence of a compound; these have been noted as they occur. Less common and less explicable variants occur in the form of syllable and word substitutions which cannot be directly attributed to phonetic or paleographic confusions. As has been suggested, these are frequently the result of attempts to deal with some' difficulty in the text being copied. Or they may be attempts to "improve" the text by clarifying a word or words which the copyist feels is ambiguous. Sometimes metrically equivalent synonyms, or words and phrases of similar meaning, are substituted for one another without apparent reason. Such variants are not all equally interesting or important to the critique of the text, but they are included where found because they cannot be otherwise explained. In observing the patterns of variation it is striking that certain syllables and words and even complete piidas are relatively stable while others are highly vulnerable to change. The reasons for this' are not clear in every case.
f!a in]aina resembles glga in Devanagari cha ha tha ~a dra rja
The characters rja and ttha also seem to cause confusions. Within J aina Devanagari there are frequent confusions among: ca-va-ha ttha-ccha tha-gha hbha-jjha dda-ddha-!!a-!!ha-rjrjha BENGALI AND DEVANAGARI
Confusions are more often phonetic than paleographic. PHONETIC CONFUSIONS Phonetic confusions are also common and are often at the basis of apparent orthographic confusions. They are likely to occur when the copyist transcribes his manuscript at someone else's dictation. Gujarati and Maharashtrian scribes generally confuse the length of i, t, u, ii; sibilants are also often confused. Bengali scribes reduce all sibilants to one, and they confuse ba and va (both of which usually become ha under Bengali influence). OTHER VARIANTS
7. VARIANT READINGS "'f ~ AND RELATED NOTES ~!l9---Nl This verse is also found as WS I, with variants. (b) AI navaromao. (d) F, H2 °gu.t:litiirh.
N2 The above sorts of variants are omitted from the critical apparatus. Also omitted are common permissible variants of sandhi, which have been normalized in. the text, for example (1) after r, consonants have ~ot been doubled (read tiryag, though some manuscripts read ttryyag; read vartD, though some manuscripts read vartt D); (2)
[142]
This verse is also found as WS 4, with variants. (b) B Dkiintim. 6 Many scribes use anusvara to represent the final m of a hemistich, as well as all nasals in combination with a following consonant, for example damta for danta. These representations are widespread and random and have not been recorded.
Textual Criticism
£if the Caurapaiidisika
N3 Cf. WS6 (b-d are essentially the same). MCI.38a-b + 7c-d is this verse; the only variant is in d: pronmattavan, which is the reading of WS, type III manuscripts. N4 (a) B nidhuvanao. A3, H2, MC3 sramao [= A4 (manuscript NAG.2 only)]. MC4 vidhrtanetrapariplutiiiigim; MC5 °plavaiigim. (b) D, E, F, MC4, 5 °patitakulao. Ar, 2, H3 °kuntaHini. (c) A3, 4, 5, MC3 °mantharam; D, E, G antaram. MC4,5 °bii~patatimantagatirh[?]. X5 (b) DI,3, Er tiryakskhalat ..... tiirakamavahantim. D2, E2, F, G, H, MC6 °tiirakamiiyatak~im. MC4 °tiiramivek~amaI).am; MC5 °tarakamik~amaI).am. (d) A4 °vadanarh sumukhirh ; M C4, 5, 6 °vadanam manasa. N6 D omits the verse. PMas b-d are found in all versions of WS, type IV; pMa a in these manuscripts is N 7a. (a) AI, 2, 3, 4, 5, MC3 kamalayatak~im. [This may in fact be the more original reading, the other being an attempt to avoid repetition of N 3a; the omission of the entire pMa in WS, type IV and of the verse in D supports this. The reading fravattaO was finally chosen because of its presence in the usually conservative manuscript A7 (= MEHTA); this suggests its antiquity, if not " originality."] (b) EI °virahavyathitaO; H2, MC4, 5, 6 °virahaglapita'. F, G ghoravirahanalapi