Candra Vyakarna Some Questions

Author(s): Johannes Bronkhorst
Publisher: Johannes Bronkhorst
Catalog link: https://jainqq.org/explore/269536/1

Page #1
183 The Candra-vyakarana: Some questions JOHANNES BRONKHORST 1. CANDRA-SUTRA AND CANDRA-VRTTE: ONE OR TWO AUTHORS? Some arguments seem to indicate, at first sight, that the authors of the Candra-Vrtti and of the Sutra were different people. The most important among these bases itself on the use in the Vrtti of vaksyati "he will state and karoti "he makes" on the one hand, and of vaks yamah "we will state" on the other. The third person verbal forms refer nine out of a total of ten times to one of the surviving Candra-sutras; the one remaining case pertains to a particular accent, not dealt with in the surviving text. Of the ten occurrences of vaks yamah, one refers demonstrably to another passage of the Vrtti, eight to the treatment of accents which is missing in the surviving text, while one would seem to concern a sutra. If we leave out of consideration, for the time being, the cases concerning accents, and suppose that the one puzzling use of vaksyamah refers to the explanation of a sutra in the Vrtti rather than to the sutra itself, we may be tempted to conclude that the Vrtti uses the third person to refer to the Sutra, and the first person to refer to other parts of the The Candra-vyakarana: some questions Vrtti. What further conclusions can be drawn from this? It goes without saying that the temptation is great to see in this use of the first and third persons proof that the author of thi Vrtti did not compose the Candra-Sutra. Yet it would be overhasty to draw this conclusion without considering the habits of the age concerned. These habits appear to have been rather varied, for we find that a text like the Yoga Bhasya uses vaksyamah to refer to the Yoga Sotra, the different authorship of which is not in doubt.2 The author of the Tattvarthadhigama Bhasya uses both first and third person verbal forms to refer to the sutras on which he comments, and whose author appears to have been different. It is not so easy to find out how authors of both the basic text and the commentary referred, in their commentary, to the basic text, for few certain cases of such combinations are known from the first millennium. But one undoubted example is Mandana Misra's Brahmasiddhi, which consists of verses and commentary. Mandana uses the third person on several occasions in the commentary to refer to his own verses.' Another example is the first chapter of Dharmakirti's Pramanavarttika. Unlike Mandana, Dharmakirti uses the first person (vaksyamah) a few times in his commentary to refer to the text commented upon. In other words, the use of vaksyari in the CandraVrtti does not allow us to conclude anything whatsoever. It has been suggested, on the basis of the frequent references to accentual questions by means of the word vaksyamah, that the author of the Vitti composed the (now missing) eighth Adhyaya of the Candra-vyakarana, on accents. But this position is not without difficulties, for the Candra-Sutra already uses a number of anubandhas that indicate accents (without explaining their significance). The surmise, meant to explain these anubandhas, according to which the author of the Sutra only intended to compose a section on accentuation, but did not succeed in doing so, has obviously no other justification than the wish to uphold the hypothesis concerning the author of the Vrtti: it need not detain us here. 2. On YS 2.29, 40, 46. 3. Bronkhorst, 1985: 169-170. 4. E.g. p. 75 1. 4 dariayati, p. 231. 17: aha. 5. Sea Gnoli, 1960: xvi n. 6. Dash, 1986: p. 2 fr. 5. * An earlier version of this paper was read at the Vitith World Sanskrit Conference held in Vienna, 1990. This earlier version is frequently criticised in a paper by Thomas Oberlies (1996), which however offers further material in support of some of the theses presented in it. While discussing my review (WZKS 36, 1992, 239-240) of his book on the Candra-vyakarana (Oberlies, 1989), Oberlies (1996: 266 n. 2) accuses me of the 'Ungenauigkeit of having claimed that he had prepared a critical edition. The 'Ungenauigkeit is however his, for my review does not mention the expression critical edition nor does it suggest that Oberlies' book contains one. I thank Jan E.M. Houben, who made the Jainendra-vyakarana available to me. 1. These arguments are presented in detail by P.C. Dash (1986: 8-21). Cp. Rau, 1996: 336.

Page #2
The Candra-vydkarand: some questions preserves a locative case-ending. Candra represents this whole discussion in one single stra: saptamya bahalam (CS 5.2.11). Do we have to conclude that Candra was not interested in the details provided by Panini (and by the Mahabhasya)? No such conclusion is necessary if we accept that the Vrtti is a complement to the Candra-Sutra: the Vrtti presents the information which the Sutra omits. The Vrtti on CS 1.3.106, similarly, presents the contents of no fewer than eight Papinian rules, P. 3.3.131-138. These and many other examples almost force us to conclude that Candra-Sutra and Candra-Vrtti were conceived of together. This does not necessarily exclude the possibility that two authors composed these two works-say a teacher and his student. But it makes it extremely unlikely that the Candra-Sotra was ever conceived of as a selfcontained work." JOHANNES BRONKHORST Another argument is more interesting the Vrtti uses in certain cases the words sam jria and naman where the Stra has naman and akhya respectively. This deviation between Satra and Vrtti constitutes the strongest argument I know of in support of a double authorship, even though it is hard to assess how strong an argument it really is. There is another circumstance that seems to me relevant in the present discussion. The Candra-Sutra is not complete without the Vitti! A glance at Licbich's Konkordanz Panini. Candra (1928) shows that the Candra-Stra was not meant to be shorter than Panini's grammar. It omits, to be sure, in its present form rules on accent, Vedic rules and samjha-stras; but we know that the first two either existed or were planned (see the above remarks on accentual anubandhas), while samjhi-stras were left out on purpose. In general the Candra-Sutra follows Panini's grammar in all its details. Indeed, no attempt is made to leave out rules that produce non-current forms. And yet, sometimes the Candra-Sutra skips a number of Paninian sutras. Why? Does the author of the Candra-Satra lose interest in these cases in the forms prescribed by the Astadhyayi? In practically all these cases the skipped stras, or the forms they are meant to produce, recur in the Candra-Vrtti. In other words, Candra-Sotra and Candra-Vrtti together represent practically the whole of the Astadhyayi, with the exception of the portions specified above. The Candra-Sutra by itself does not do so: it presents the irregular image of a text which sometimes follows the Astadhyayi step by step, and at other occasions walks through the field covered by the Astadhyayi with seven-league boots The twelve stras P. 6.3.9-20 constitute an example. These stras discuss in detail the cases where the first member of a compound 2. Is THE KASIKA INDEBTED TO THE CANDRA-VYAKARANA? Candra-vyakarana and Kasiki contain a large number of similar or even identical passages. A priori this suggests one of the following three explanations: a) the former borrowed from the latter; b) the latter borrowed from the former; c) both borrowed, directly or indirectly, from a common source. I will not here consider the possibility that the Candra-vyakarana borrowed from the Kasika: chronological considerations make this unlikely. How do we choose between the two remaining options? Personally I consider it a priori improbable that the Kasika commentary in the Paninian tradition should have as a major source a text like the Candra-vyakarana, which belongs after all to a different grammatical tradition. Wouldn't one rather expect the Kasiki to draw its inspiration primarily from the Paninian tradition, 7. Not everywhere! On eight casions both the Candra-Seurs and the Vise samjha (Dash, 1986:59, read 1.1.123 for 1.1.23). Note that the Vitti does not always reintroduce the Paninian term valytarankhyd (P. 6.3.7) becomes naman in CS 5.2.10, mjid in the Vitti, samjid in P. 3.1.62 becomes dikya in CS 4.1.65, man in the Vytti; samjid in P. 73.67, on the other hand, becomes dyd in CS 6.1.95, and remains akhyd in the Vrtti. 8. Proof is constituted by Oberlies' (1989: passim comment nicht belegen". which occurs on virtually every single page of his translation of parts of the Candra-vyakarana in connection with the forms to be produced by Candra's 9. Compare Anna Radicchi's (1985: 67) remark: "Nel Candravykarana e generalmente rispettata la ripartizione nella presentazione della materia: i sitra danno l'inquadramento teorico generale mentre la vrti raccoglie il lessico,..." 10. Oberlies (1996: 272) makes the useful observation that one should unlichst einmal die uns tatsachlich erhaltenen grammatischen Werke auf Abhingigkeiten (etc.) untersuchen, the man daran geht, die Existenz nun verloren gegangener Grammatike u postulieren". He seems to think that this observation might help to choose between options b) and c). I fail to see how it could possibly do so.

Page #3
185 187 JOHANNES BRONKHORST say from earlier commentaries on the Astadhyayi? Is there evidence that any such commentaries existed? I have presented some such evidence in an earlier publication (1983): the Kasika explicitly mentions an earlier Vrtti, in which connection the Nyasa mentions the names of Culli, Bhatti, Nallura etc.: Bharthari refers to earlier Vrttikaras in his commentary on the Mahabhasya, and mentions onc by name ("Kuni')." But there is more evidence, some of which I will now present. (1) The Vrtti on CS 4.2.8 refers to an alternative interpretation of that sutra, which it ascribes to others' (anye), and which agrees with the interpretation presented by the Kasika under the corresponding rule P. 5.2.5. Chronological considerations do not allow us to think that the Vrtti here rejects the Kasika. This leaves only one possibility: both the Candra-Vrtti and the Kasika found this opinion in another, carlier work, most probably belonging to the Paninian tradition. Oberlies (1996: 285-86) agrees with this conclusion, but prefers to think here and in some other casesthat the source of both Candra-Vrtti and Kasika is a lost commentary (by Devanandin) on the Jainendra-vyakarana. I shall now discuss some passages from Bharthari's commentary on the Mahabhasya (the 'Dipika") which throw further light on our present question: (ii) Consider first the line The Candra-vyakarana: some questions (arkupninumvyavaye 'pi): numgrahanam arus dropalakpandrtham drastavyam. It does not occur in the Mahabhasya," nor, to my knowledge, in the Candra-vyakarana. How to explain this? It seems implausible that the Kasika should borrow an example from Bharthari that occurs in an otherwise completely unrelated context. No, we must rather assume that Bharthari borrowed his example from a work belonging to the Papinian tradition-most probably a commentary on the Astadhyayi-which was also used by the Kasika. He cannot have borrowed it from Devanandin's lost commentary on the Jainendra-vyakarana, for the corresponding sutra in this grammar (5.4.86) reads arkupvarivavaye 'pi, without num. (iii) Patanjali's Mahabhasya (Maha-bh vol. I p. 61. 6-7) mentions in a discussion the two Paninian stras karmany an (P. 3.2.1) and ato nupasarge kah (P. 3.2.3). Bharthari's commentary gives the following illustrations: kumbhakara and kandalava for P. 3.2.1; parinitra and goda for P. 3.2.3 (Ms 6b3; CE 1.15.1-2; AL 17.1314: Sw 22.7-8). All these examples are also given by the Kasika under the sutras concerned. Moreover, kumbhakara and kandalava are mentioned in the Mahabhasya on P. 3.2.1, and in the CandraVrtti under the corresponding rule CS 1.2.1. Goda does not occur in the Bhasya on P. 3.2.3, but the word is used a number of times elsewhere in the Mahabhasya; it is also mentioned in the CandraVrtti under the rule corresponding to P. 3.2.3, i.c., under CS 1.2.2. But parsnitra is mentioned neither in the Mahabhasya nor in the Candra-vyakarana. Where did Bharthari find this form? We can again exclude the possibility that the Kasika borrowed this example from Bharthari. Rather, Bharthari uses here some known illustrations which he, apparently, found in an earlier commentary on the Astadhyayi, which was used by the Kasika, too. The example parsnitra also occurs in the Mahavytti on sutra 2.2.3 of the Jainendra-vyakarana, which corresponds to P. 3.2.3. but this does not interfere with this conclusion. (iv) VL. 6 on P. 1.1.38 (taddhitas casarvavibhakti) proposes to enumerate the faddhita formations that are avyaya 'indeclinable'. and the Bhasya thereon actually does so, in the following passage (Mahl-bh vol. I p. 95 1. 9-11): yatha numgrahanam anus dropalaksanartham "Like the use of num which serves to characterize an anusvira" given in Bharthari's commentary (Ms 9a8; CE 1.22.11-12: AL 26.21; Sw 32.1). This line constitutes here an example, meant to illustrate Bharthari's statement to the effect that artha in arthasambandhe is not used for its own sake, but in order to characterize the connection (sambandhopalaksanalvena). Where did Bharthari find this example? The same line is found in the Kasika on P. 8.4.2 11. Cp. Bronkhorst, 1983: 382; Oberlies, 1995: 311 n. 150. Note that the Nyasa (vrti Panini-pranitnim stranam vivaranam Call-BharriNallarddiviracitam) and the Padamanjari (Panini-praplanam satrapam Kw prablyihir wiracitam waragam), both on the first introductory stanza of the Kaski, state explicitly that these names refer to commentators on Plini's grammar. 12 The Mahabhasya has twice a dravidesnom ugrahanam (Mahl-bh vol. I p. 291. 11; vol m p. 454 L 4. on P. 8.4.2).

Page #4
JOHANNES BRONKHORST siddham tu pachat (vt. 6) pathad wa siddham etarl katham pashah kartavyah/ tasiladayah prak pafapah/ sasprabhrlayah prak samdsantebhyah/ mantahl krivo'rthah/ tasivaril nandaa iti But the desired result) is obtained by enumeration. (vt. 6) Or this [desired result is obtained by enumeration. How must the enumeration be made? From lasit until pdsap (i.c. the faddhita suffixes taught in P. 5.3.7-46), from fas until the compound endings (taught in P. 5.4.42-671. (a suffix) which ends in m (ic. am and am. P. 5.4.11. 12). (a suffix) which has the meaning of krvas (P. 5.4.17-20). tas! and vail (P. 4.3.113 and 5.1.115), nd and nd (P. 5.2.27). The passage in Bharthari's commentary which discusses the above enumeration is unfortunately very corrupt, but the following partial reconstruction seems in the main correct (Ms 76c1-7; CE VI(2).6.18-25; AL 226.1-10): Salah The Candra-vyakarana: some questions 189 The suffix] that (prescribed by P. 5.3.111 pratnapurva-Jvisvemar that Ichandasi] is not included in the Mahabhasya. But an effort is made to include it) as follows: "The same [suffix thal (prescribed by P. 5.3.23] when expressing manner (prakaravacane) is the one which comes, in Sacred Language, after the words] pratna etc. in accordance with P. 5.3.111), in their own meaning.' (What is described in the Bhasya as) 'ending in m are am and dim. "Having the meaning of krvas' are krvasUC, SUC, and dha. 'asl and vail, na and nd in the Bhasya]: (tas/ is prescribed by P. 4.3.112 and 113) renaikadik and tasis ca. vat is prescribed by P. 5.1.115) tena tulyam kriya ced watib. There is enumeration of this many in the gana, thus it is read. We see that Bharthari, while primarily commenting on the enumeration of indeclinables in the Mahabhasya, makes use of a gana he knows, and which contains more than just what is enumerated in the Bhasya. Bharthari himself states in so many words that asl and thaL are not read in the 'Varttika', i.c., in the Bhasya. The gana known to Bharthari seems, moreover, to have contained the phrases tasiladis taddhita edhacparyantah; it also had tasl, as and thal, in that order. The gana known to Bharthari appears to be very close to the one contained in the Kasika on P. 1.1.37 (svaradinipatam avyayam). which has: tasiladih taddhita edhacparyantah, sastasi, kyrvasuc, suc, asthalau, cvyarthas ca, am, am. The Candra-vyakarana, on the other hand, does not, to my knowledge, contain anything like it. The Mahavrtti on Jainendra-vyakarana 1.1.74 contains an enumeration corresponding to the one accompanying P. 1.1.37, but nothing remotely resembling Bharthari's words can be found in it. Again we are led to believe that Bharthari and the Kasika made use of the same carlier text; since our earliest source of Paninian ganas is the Kasika, a commentary on the Astadhyay, we may assume that this earlier text used by both Bharthari and the Kasika, too, was a commentary on the Astadhyayi. The preceding examples indicate that Bharthari used a commentary on the Astadhyayi which was also used by the Kasika. But we know that Bharthari knew more than just one such commentary. At times he may have followed another commentary than the one that influenced the Kasika, or simply deviated from tasiladayah prak pafapah...gano 'py evam pathyatel taslladis taddhita edhacparyanta ini fasprabhstayah prak samasantebhya iris bahwalparthdcchas ity atah prabhrti yavan madrat parivapane dac in ... lasipratyayas ca pratiyoge yas ca tenaikadik tasis cetil as ina asis ceri aya itil varitike tu faddhitah prakta iti asir na pashitah/ thal visvend that iti ayam vartrike nopasamgrhilah yarnas i kriyatel ya cva prakdravacane thal chandasi sa eva prainddibhya svarthe bhavartti mantah amam/ krtvorthah/ krvasnc suc dhaltasivatli nandriau tatha lenaikadik tasis cal vatih tena tulyam kriyd ced vatir itil game parha erdvaram ili parhitam! 'From tasll until pasap.' ... The gana, too, reads like this: "The addhita [suffixes) from fasil until and including edhac' i.e. P. 5.3.7-46). (The line in the Bhasya) 'from Sas until the compound endings' (corresponds to the section] from [P. 5.4.42] bahvalparthac chas [karakad anyatarasydn) until Dac [in P. 5.4.67) madrat parivapane. The suffix tas/ (in the gana) is the one connected with prati (prescribed in P. 5.4.44 pratiyoge pancamyas tasih) as well as the one (prescribed] by P. 4.3.112 and 113:] tenaikadik and tasis ca. [The suffix) as is prescribed in the Unadi stra) ina asit ca (which gives rise to the form] ayas. But the Unadi (and therefore krt) suffix] asl is not read in the Mahabhasya" because taddhita (suffixes) are under discussion (there). 13. This reconstruction differs in several respects from the ones in the two published editions. 14. The straina sit or inaf cash is present in the surviving versions of the Unadi Stra. Not all commentaries mention that aws is an indeclinable: an exception is Mahadeva's Unadikosa 4.221. yassi prakrteriala 15. On the use of "Varttika' in order to refer to the Mahabhasya, see Bronkhorst, 1990

Page #5
190 JOHANNES BRONKHORST the latter. The following example belongs to this category. It also shows that the Kafika, though almost identical with the CandraVrtti, cannot in this case have borrowed from the latter. (v) In the course of a discussion in the first Ahnika of his commentary (Ms 8a8-9; CE 1.19.17-18; AL 23.4-5; Sw 28.1-2) Bharthari gives the following example: radyathal ganga hi ripenasrita tannamika nami wpadayati ganga til devatadirupenairiyamana dhakam gangeya itil For example, the river] Ganga when referred to in its own form, having that (i.c. 'Ganga') as its name, produces the suffix) aN. [which gives rise to ganga 'son of Ganga'. When it is being referred to as the goddess etc. (of that name, it produces the suffix) dhak, (which gives rise to gangeya. The occurrence of tarnamika in the first half of this passage shows that Bharthari derives ganga with the help of P. 4.L.113 avrddhabhyo nadimanusibhyas tannamikabhyah. This is interesting because the Kasika lists ganga under P. 4.1.112 sivadibhyo 'n, and does not use P. 4.1.113 in the derivation of ganga. Apparently Bharthari did not find gariga in the gana sivadi, in the commentary on the Astadhyayi which he decided to follow in this respect. Consider now the following explanation in the Kasika under P. 4.1.112: The Candra-vydkarana: some questions (with phil), gargeya (with dhak). Since we know (Bronkhorst, 1983: 373 f.) that the Kasika did not tamper with ganas, we must conclude that it found the word garigd in the gana fivadi already in the earlier commentary. Yet the above passage reads more or less like a justification of the presence of ganga in sivadi. Does this justification make sense? Doesn't Bharthari succeed equally well in deriving ganga, without a mention of garga in fivadi? Why didn't the Kafika, like Bharthari, derive ganga with the help of P. 4.1.113 avrddhabhyo nadimanusibhyas tannamikabhyah? The reason is found under P. 4.1.121 dvyacah, which is, according to the Kasika, an exception to P. 4.1.113. P. 4.1.121 prescribes dhak (= eya) after words of two syllables ending in a feminine suffix. This would account for gangeya, but-4.1.121 being an exception to 4.1.113--would at the same time exclude the form ganga. The Kasika-or rather, the commentary which it follows-solves the problem by avoiding both the sutras 4.1.113 and 121 in this connection. Ganga is now derived by P. 4.1.112 Sivadibhyo . gangeya by P. 4.1.123 subhradibhyas ca. This is accomplished by adding the term ganga to both the appropriate ganas. (Note in passing that for Bharthari and his example P. 4.1.121 cannot have been an exception to P. 4.1.113.) Interestingly, the Candra-Vrtti on CS 2.4.41 (which corresponds to P. 4.1.112) agrees with the Kasika. We read here: gangaiahdadiha pathad an gangah/ Subhradipathad dhak: gdrigeyahl rikadipahat phir: gdigdyanih But here we find no explanation whatsoever of this derivation of ganga. Nor do we find any indication that CS 2.4.51 (dvyacah; = P. 4.1.121) is an exception to CS 2.4.42 (nadimanusinamno 'nadai jadyacah; P.4.1.113). So why did the Candra-Vrtti include ganga in the gana sivadi? Apparently for no other reason than that it found the word there in one of the commentaries on the Astadhyayi it followed. It is therefore not possible to maintain that the Kasika here simply borrowed from the Candra-vyakarana. Quite on the contrary, in order to understand what underlies the procedure of the Candra-Vyakarana in this case, we have to consult the Kasika, which better preserves the information contained in the earlier commentary which influenced the Candra-vyakarana. gangasabdah pathyale nikddiphina subhradidhaka ca samdve drtham! tena frairupyam bhavati gdigahl gangd yanih/ gangeyah/ The word ganga is listed in the gana siva etc.) in order to include [the suffix analong with (the suffix] phi on account of [ganga being included in the gana) tika etc. (P.4.1.154) and with (the suffix] dhak on account of Iganga being included in the gana) subhra etc. (P. 4.1.123). There are therefore three forms: ganga (with aN), gangayari 16. This mode of expression is once connected with Apidali in the Mahabhiya (vol. II p. 281 L. 3-5: on P. 4.2.45 Slokart. 2): tatha capisaler widhih/ dhenar anal kam atpadayatil dhenanam samiho dhain kam/. It also occurs elsewhere, eg. in the Nyksa (vol. III p. 332 1. 27-28. on P. 4.1.41; vol. IV p. 117 L. 29, on P. 5.1.129) and in the Padama jarl (vol. III p. 405 1. 18, on P. 4.1.86). Compare in this connection Kumarila Bhatta's Tantravirttika on MIS 3.4.13, p. 368: karanavam evedam wbtena nyayena karttavat svdiyi vibhakrim notpadayati; and p. 369: ... tad ena siddhaphaladt phalabhutdim vibhakrim aruddayad api sisani punar utpadayisyanti.

Page #6
JOHANNES BRONKHORST The Mahavitti on Jainendra-vyakarana 3.1.101 contains some lines which correspond in their content with the lines from the Kasika cited above. The most probable conclusion to be drawn from this is that the Mahavrtti was influenced, directly or indirectly, by the same early commentary that also influenced the Candra-Vitti and the Kasika. 192 We cannot overestimate the importance of Bharthari's commentary on the Mahabhasya, a closer study of which might bring to light much more evidence pertaining to the questions we are investigating. The above examples must, for the time being. suffice. They show clearly, as it seems to me, that any misgivings about the existence of pre-Candra commentaries on the Astadhyayi, and their influence on Kasika and Candra-vyakarana, are without foundation. (vi) The contents of two sutras of Candra correspond to a verse in the Kasika. The two sutras read: CS 4.4.72: pandudakkrsnad bhumeh After [compounds ending in) bhumi preceded by pandu etc., [comes the samasanta-suffix ac] CS 4.4.73: samkhyaya nadigodavaryos ca After [compounds ending in] nadi, godavari and [bhami] preceded by a numeral, [comes the samasanta-suffix ac]. The Candra-Vrtti gives as examples: pandubhumah, udagbhumah, krsnabhamah (for 4.4.72), and pancanadam, saptagodavaram, dvibhumah prasadah, dasabhumakam sutram (for 4.4.73). The Kasika on P. 5.4.75 contains the following verse: krsnodakpandupurvaya bhumer acpratyayah smrtah godavaryas ca nadyas ca sankhyaya uttare yadill This justifies the following examples: kysnabhumah, pandubhumah, udagbhumah, pancanadam, pancagodavaram. However, the examples dvibhumah prasadah and dasabhumakam sutram are not covered by this verse, yet they are desired. The Kasika, therefore, adds the line: bhumer api sankhyapurvayah acpratyaya isyate, followed by these two examples. The question is: did the Kasika in this case borrow from the Candra-vyakarana? One argument pleads in favour of this position: The Candra-vyakarana: some questions 193 dasabhumakam sutram looks like a Buddhist expression. However, if we accept this position, we must not only assume that the author of the Kasika at times wrote in verse, but that he was not capable of formulating the verse-which has already twice the number of syllables as the two sutras of Candra combined-in such a manner as to express the same meaning as those two sutras; he has to add an isti in simple prose. Borrowing in the other direction seems far more likely. That is to say, the verse appears to have preceded both Candra and the Kasika. Either Candra or someone before him realized that the examples dvibhuma and dasabhuma should also be included. Candra managed to express the new situation very elegantly in two short sutras. The Kasika preserved the verse but, following either Candra or the earlier unknown grammarian, added the above line. In any case it seems more than likely that both Candra and the Kasika were influenced by the same earlier grammarian. The Mahavrtti (under Jainendra-vyakarana 4.2.71) cites the same verse as the Kasika, but instead of adding an isti so as to justify the forms dvibhuma and saptabhuma, it derives this justification from the word ca in the verse (cakarad bhumir api bhavati). It does not give the example dasabhumakam sutram. One might be tempted to conclude from this that the Kasika borrowed this verse from the Mahavrtti, but the arguments presented above suggest that the verse and its examples are older than the Candra-vyakarana and therefore much older than the Mahavrtti. It is not therefore necessary to assume that the Kasika borrowed in this case from the latter. (vii) Wilhelm Rau (1996: 336) makes the following observation: "[CS] 4.4.119 sampraj januno jnah [ist] gegen [P.] 5.4.129 [prasambhyam janunor jnuh]. M.W. ist dies die alteste Stelle, wo das Ungetum auftaucht. Das Mahabhasya hat nichts zur Sache. Im Vakyapadiya 2.220 wiederholen samtliche (!) karika-Mss den Fehler, dagegen bleiben der Amarakosa 2.6.47 und die Kasika 5.4.129 beim richtigen, ohne verhindern zu konnen, dass spatere [Candra's] Irrtum am Leben erhalten." The Candra-vyakarana prescribes prajna instead of prajnu, and samjna instead of samjnu. Vkp 2.220ab reads: prajnasamjnadyavayavair na casty arthavadharanam. The editor has corrected the first compound into prajnusamjnvadyavayavair, in the light of P. 5.4.129, but points out in a note that this emendation goes against

Page #7
JOHANNES BRONKHORST all (1) mss. The temptation is great to conclude that Bharthari knew P. 5.4.129 in a slightly different form-perhaps prasambhyam janunor jnah-which was also the reading known to Candra. Since the Kasika has this sutra in its correct form, we may have to conclude that in this case the Kasika follows another sutra readingand therefore another commentary on the Astadhyayi-than the Candra-vyakarana. It certainly does not borrow here from the Jainendra-vyakarana, which has the incorrect sutra sampraj januno jnah (4.2.130). The Mahavrtti first gives the two incorrect forms samjna and prajna, then observes that according to some -jnu should be used instead of -jna. However, both opinions, according to the Mahavrtti, are correct. (jna ity ukarantah kesamcid adesahl matadvayam api pramanam!) (viii) To conclude this section a few words must be said about the form of P. 3.3.122. We start from Kielhorn's brief observation regarding the form of this sutra in the Mahabhasya (1885: 192-193 [195-196]; 1887: 181 [229]; I quote from the 1887 article): 194 P. III.3,122 adhyayanyayodyavasam karadharavayas ca originally did not contain the words adhara and avaya, which have been inserted from Katyayana's Varttika on the preceding rule (Vol. II. p. 155). The word avahara, which is mentioned in the same Varttika, is in the Kasika given in the commentary on P. III.3.122. In the Mahabhasya, Vol. II. p. 146, 1. 20, where the rule has been quoted, the MSS. give it as read in the Kasika, excepting that the MS. K omits from it adhara. Kaiyata on P. III.3,121 has the remark-adhyayasutra adharavayasabdau varttike dariandd abhiyuktaih praksiplau. This laconic passage presents a real and serious problem. Do we have to assume that Patanjali himself changed the sutra? He never does anything like it. And even if we assume that here, exceptionally, he interfered physically with the wording of a sutra, why didn't he include avahara? Or must we, alternatively, believe that the Kasika presents us the sutra in its original form? In that case P. 3.3.121 vt. 1 becomes unintelligible. Neither of these two alternatives is therefore satisfactory. There is however a third alternative. Thanks to the researches of V.P. Limaye, W. Rau and M. Witzel we now know what was not yet known to Kielhorn, viz., that the surviving mss. of the Mahabhasya (or at any rate the ones used for Kielhorn's edition) all go back to an archetype that may date from around 1000 C.E. The Candra-vyakarana: some questions 195 (see Bronkhorst, 1987: 14 f.). This archetype may have been contaminated, "improved upon", by the then standard reading of Paninian sutras. A rule like P. 3.3.122, which is only once cited in the Mahabhasya, would be particularly vulnerable to such "improvements". This is a hypothetical solution, yet it is the only one which satisfactorily explains the situation. If it is correct, we must believe that someone after Patanjali and before the Kasika added the words adhara and avaya to P. 3.3.122, but not avahara. The Candra-Vrtti on CS 1.3.101 has the list with adhara and avaya, and without avahara! Candra did not borrow his list from the Mahabhasya, because we now think that the Mahabhasya did not contain it. Nor did he borrow it from the Jainendra-vyakarana (2.3.103), which has a different list, containing avaya and avahara, but not adhara. Inevitable conclusion: Candra borrowed his list from an earlier, but post-Patanjalian, work in the Paninian tradition, the same work, probably, from which the Kasika borrowed P. 3.3.122 in its present form. 3. THE GEOGRAPHICAL LOCATION AND DATE OF THE CANDRA-VYAKARANA In order to discover the geographical location of the author(s) of the Candra-vyakarana we must make use of the "second index fossil" drawn attention to by H. Scharfe (1976). The relevant discussion in the Candra-Vrtti concerns the use of the two future tenses lyf (first future) and luf (second future). It is prescribed for the future in general (CS 1.3.2), lug for the future other than today's (CS 1.3.3). The Vrtti on CS 1.3.106 gives further clarifications; the for us important part reads: maryadavacanabhave 'pi viprakarsaparatvad vivaksaya anadyatanavidhir bhavaty eval yo 'yam adhva niravadhiko gantavyas tasya yad avaram kausambyas tatraudanam bhoktasmahel If no limit is expressed, the rule regarding "not that same day" (CS 1.3.3) is certainly [applied), because distance is intended to be expressed. [An example is:] "The limitless road that must be traversed on the part of it which is this side of Kausambi we shall eat (bhokiasmahe; second future) rice."

Page #8
196 JOHANNES BRONKHORST Distance is intended to be expressed by the verbal form bhoklasmahe. Yet this verb is used only in connection with the early part of the journey, between 'here' and Kausambi. The example also makes mention of a "limitless road", and this cannot but concern the part from 'here' to Kausambi as much as the part beyond Kausambi. For this expression is added in order to bring out the sense of the second future; it is absent in the parallel example concerning the first future. It follows that a limitless road separates Kausambi from the position of the author of the Vrtti. And even though no precise conclusions can be drawn from this information, it is none-the-less justified to think that the distance from the position of the author to Kausambi was considerable. It excludes an area too close to Kausambi, as proposed by Scharfe." Scharfe's discussion of this "index fossil" is marred by the fact that his comparative treatment of it in a number of grammatical texts made him lose sight of the introductory phrase reproduced above, which does not appear to introduce the "index fossil" in any of the other texts. Yet this phrase, as we have seen, is of vital importance for understanding the precise significance of the example in the Candra-Vrtti. Scharfe makes a further mistake: the fact that some of the other grammarians specify "that they would eat twice on the first leg of their respective journeys" (my emphasis) leads him to the conclusion that Candra, who does not add this specification, lived a one day journey away from Kausambi. This conclusion does not only sin against "the distance intended to be expressed" and "the limitless road", but also against "the rule regarding 'not that same day""" We turn to the date of the Candra-vyakarana. Since Sutra and Vrtti were apparently composed more or less simultaneously, evidence derived from the Vrtti is valid for the Candra-Sutra too. The Candra-Vrtti cites Kalidasa's Raghuvamsa and 17. For further inferences regarding Candra's location, see Bronkhorst, 1983: 397. Aklujkar (1991: 26-27 n. 6d) thinks that the view there expressed "is partly based on what Scharfe thought to be justified"; he further sees some (remote) similarities with the ideas of Satyakama Varma. Aklujkar disagrees with both these authors (as I do), and concludes: "Bronkhorst's composite view, therefore, stands doubly refuted and need not be discussed separately." I hope that the present exposition will allow Aklujkar to arrive at a better understanding of my point of view. 18. For further criticism of Scharfe's discussion, see Aklujkar, 1991: 29-30 n. 11. The Candra-vyakarana: some questions 197 Kumarasambhava (Oberlies, 1989: 13; Rau, 1996: 337). The concluding verses of the Vakyapadiya-Vrtti mention "Acarya Candra and others". They further suggest that Bharthari is later than Candra etc. The Vakyapadiya-Vrtti, in its turn, is older than Dignaga. This provides the following chronological sequence: Kalidasa Candra-vyakarana Bharthari Vakyapadiya-Vrtti Dignaga These different authors and works must probably all be placed in a period of at most one hundred years, most of it in the fifth century. If it is true that at least the first four of these were located in more or less the same area in the west of India, this chronological proximity is in no way problematic. 4. CONCLUSIONS The conclusions to be drawn from the above material are not very different from those presented in my 1983 article. And indeed, the aim of this article was not to present new findings, but to better support earlier conclusions. It can now with more certainty than before be maintained that Candra-Sutra and Candra-Vrtti-even though different authorship of these two works cannot altogether be ruled out-must be looked upon as belonging together, as essentially one work conceived as such right from the beginning. It has also been more satisfactorily established that the Candravyakarana and the Kasika shared at least one earlier source (other than the Mahabhasya and the Jainendra-vyakarana). This does not, of course, exclude the possibility that the Kasika knew the Candravyakarana, but decisive evidence to that effect is not known to me." One might in this connection cite, with Oberlies (1989: 10), the 19. See also Hahn, 1992: 93. 20. See Bronkhorst, 1988: 111, which states the reasons for believing that these verses belong to the Vakyapadlya-Vitti. It can of course not be proved with absolute certainty that the Candra here mentioned is the author of the CandraSutra, and the identification is not self-evident. Yet the strong influence of the Mahabhasya on the Candra-vyakarana agrees with this identification.

Page #9
JOHANNES BRONKHORST example yenagnis tena gatah, found both in the Candra-vyakarana (on CS 2.2.8) and the Kasika (on P. 2.1.14). This seems no doubt "typical Buddhist idiom", even though it may not be completely unknown to Brahmanical literature." It is however to be noted that this idiom occurs, in Buddhist Sanskrit literature, almost exclusively with upa-sam-kram, upa-gam, and upa-i, and probably never with only gam." Of slightly (but how much?) more weight may be the expression dasabhamakam sutram considered above, and the expression ajaryam aryasam gatam (which resembles Aryasura's Jatakamala 22.88) both in the Candra-vrtti (on CS 1.1.116) and in the Kasika (on P. 3.1.105). It is, finally, hoped that a misinterpretation introduced into the "second index fossil" by Scharfe has now been cleared away. 198 REFERENCES Aklujkar, Ashok (1991): "Interpreting Vakyapadiya 2.486 historically (part 3)." In: Paninian Studies. Professor S.D. Joshi Felicitation Volume. Ed. Madhav M. Deshpande and Saroja Bhate. Ann Arbor: Center for South and Southeast Asian Studies, University of Michigan. Pp. 1-47. Bronkhorst, Johannes (1983): "On the history of Paninian grammar in the early centuries following Patanjali." Journal of Indian Philosophy 11, 357-412. Bronkhorst, Johannes (1985): "On the chronology of the Tattvartha Sutra and some early commentaries." Wiener Zeitschrift fur die Kunde 21. Oberlies (1996) finds fault with this remark (p. 271 n. 17), and turns the situation on its head on p. 314: "[Es] wird deutlich geworden sein, dass es keinen zwingenden Grund gibt zu zweifeln, dass Candrasutra/-vrtti... der Kasika bekannt [war]. Es bleibt daher zu hoffen, dass zukunftig diese Abhangigkeit nur auf Grund wirklicher schlussiger Indizien in Frage gestellt werden wird..." 22. Consider, e.g., Mahabharata (crit. ed.) 14.16.41: acirat tu gamisyami yenaham rvam acucudam. This might be translated: "But I will soon go [there], where I just urged you to go." 23. See Simson, 1977. 24. Cp. Rau, 1996: 335. Note that both CS 1.1.116 and P. 3.1.105 read: ajaryam samgatam. The example ajaryam aryasamgatam may have been invented by any commentator, and the influence of ajaryam hy dryasamgatam in the Jatakamala is far from certain. The Candra-vyakarana: some questions Sudasiens 29, 155-184. Bronkhorst, Johannes (1987): Three Problems Pertaining to the Mahabhasya. Poona: Bhandarkar Oriental Research Institute. (Postgraduate and Research Department Series No. 30. "Pandit Shripad Shastri Deodhar Memorial Lectures" [Third series].) Bronkhorst, Johannes (1988): "Etudes sur Bharthari, 1: L'auteur et la date de la Vrtti." Bulletin d'Etudes Indiennes 6, 105-143. Bronkhorst, Johannes (1990): "Varttika." Wiener Zeitschrift fur die Kunde Sudasiens 34, 123-146. 199 Dash, Prafulla Chandra (1986): A Comparative Study of the Paninian and Candra Systems of Grammar (Krdanta Portion). New Delhi: Ramanand Vidya Bhawan. Edgerton, Franklin (1953): Buddhist Hybrid Sanskrit Grammar and Dictionary. Volume 1: Grammar. Delhi: Motilal Banarsidass. 1977. Gnoli, Raniero (ed.) (1960): The Pramanavarttikam of Dharmakirti. The first chapter with the autocommentary. Roma: Istituto Italiano per il Medio ed Estremo Oriente. (Serie Orientale Roma, XXIII.) Hahn, Michael (1992): "Uber den indirekten Beweis bei literaturhistorischen Fragestellungen." Wiener Zeitschrift fur die Kunde Sudasiens 36, 91-103. Jainendra-vyakarana. The following edition has been used: Jainendra Vyakaranam by Pusyapada Devanandi with Jainendra Mahavritti of Shri Abhayanandi. Ed. Shambhu Nath Tripathi. Kashi: Bharatiya Jnanapitha. (Jnanapitha Murtidevi Jaina Granthamala, Samskrit Grantha no. 17.) Kasika. 1) Ed. Aryendra Sharma, Khanderao Deshpande, and D.G. Padhye. Hyderabad: Sanskrit Academy, Osmania University. 2 vol. 1969-70. 2) Ed., with the Nyasa of Jinendrabuddhipada and the Padamanjari of Haradatta Misra, by Swami Dwarika Das Shastri and Pt. Kalika Prasad Shukla. Varanasi: Prachya Bharati Prakashan. 6 vol. 1965f. Kielhorn, F. (1885): "Der Grammatiker Panini." Nachrichten von der K. G. d. W. zu Gottingen 5, 185-199. Repr.: Kleine Schriften 1 (Wiesbaden, 1969) pp. 188-202. Kielhorn, F. (1887): "Notes on the Mahabhashya, 6. The text of Panini's Sutras, as given in the Kasika-Vritti, compared with the text known to Katyayana and Patanjali." Indian Antiquary 16, 178-184. Repr.: Kleine Schriften I (Wiesbaden, 1969) pp. 226-232. Kumarila Bhatta: Tantravarttika. In: Srimaj-Jaimini-pranite Mimamsadarfane sampurnas trtiyo 'dhyayah. Bhatta-Kumarilapranita-Tantravarttika-sahita-Sabarabhasyopetah, Kasinatha Vasudevasastri Abhyamkara... tatha Ganesasastri Jost... ity etaih pathabhedatippanyadibhih sahitam punah samsodhitam. Poona: Anandasrama. 1984.

Page #10
200 JOHANNES BRONKHORST The Candra-vyakarana: some questions 201 Ms Manuscript of Bharthari's commentary on the Mahabhasya ("Mahabhasyadipika") reproduced, Poona: Bhandarkar Oriental Research Institute, 1980 Paninian sutra Edition of Bharthari's commentary on the Mahabhasya ("Mahabhasyadipika") by V. Swaminathan, Benaras Hindu University, 1965 Bharthari, Vakyapadiya, ed. W. Rau, Wiesbaden 1977 varttika Yoga Sutra Vkp vt. Liebich, Bruno (1928): Konkordanz Panini - Candra. Breslau: M. & H. Marcus. (Indische Forschungen, 6.) Mahadeva Vedantin: Unadikosa. Edited, with the Unadi sutras and full glossarial index, by K. Kunjunni Raja. University of Madras. 1956. (Madras University Sanskrit Series, 21.) Mandana Misra: Brahmasiddhi. Edited, with Sankhapani's commentary, by S. Kuppuswami Sastri. Second edition. Delhi: Sri Satguru Publications. 1984. (Sri Garib Das Oriental Series, 16.) Nyasa. See Kasika. Oberlies, Thomas (1989): Studie zum Candravyakarana. Eine kritische Bearbeitung von Candra IV.4.52-148 und V.2. Stuttgart: Franz Steiner. (Alt- und Neu-Indische Studien, 38.) Oberlies, Thomas (1996): "Das zeitliche und ideengeschichtliche Verhaltnis der Candra-Vitti zu anderen V(aiyakaranas (Studien zum Candravyakarana III)." Sill 20 (Festschrift Paul Thieme), 265-317. Padamanjari. See Kasika. Radicchi, Anna (1985): La teoria paniniana dei Samasa secondo l'interpretazione delle scuole grammaticali indiane dal quinto all'ottavo secolo d.c. Parte prima. Firenze: Elite. (Materiali dell'Istituto di Glottologia, Universita di Cagliari, Anno accademico 1982-1983.) Rau, Wilhelm (1996): "Die vedischen Zitate in der Candra-Vstti." Stil 20 (Festschrift Paul Thieme), 327-338. Scharfe, Hartmut (1976): "A second 'index fossil' of Sanskrit grammarians." Journal of the American Oriental Society 96, 274. 277. Simson, Georg von (1977): "Zur Phrase yena ... tenopajagama/upetya und ihren Varianten im buddhistischen Sanskrit." Beitrage zur Indienforschung. Ernst Waldschmidt zum 80. Geburtstag. Berlin: Museum fur Indische Kunst. Pp. 479-488. ABBREVIATIONS Edition of Bhartshari's commentary on the Mahabhasya ("Mahabhasyadipika") by K.V. Abhyankar and V.P. Limaye, Poona: Bhandarkar Oriental Research Institute, 1970 "Critical edition" of Bhartshari's commentary on the Mahabhasya ("Mahabhasyadipika") by various scholars, Poona: Bhandarkar Oriental Research Institute, 1983-1991 Candra-Sutra; ed. B. Liebich, Candra-Vrtti, Leipzig 1918. Patanjali, (Vyakarana-) Mahabhasya, ed. F. Kielhorn, Bombay 1880-1885 CS Maha-bh Patanjali, need. B. Lieb; 1983-1991