Paninian Studies

Author(s): Ashok Aklujkar
Publisher: Ashok Aklujkar
Catalog link: https://jainqq.org/explore/269664/1

Page #1
Paninian Studies: Professor S.D. Joshi felicitation volume. Ed. Deshpande, Madhav M; and Bhate, Saroja. Ann Arbor: University of Michigan, Centre for South and Southeast Asian Studies, no. 37. 1991 Ashok Aklujkar 1.1 The first two parts of this study were published in the Adyar Library Bulletin (1981:581-601, Dr. K. Kunjunni Raja Felicitation Volume) and in Indological and Buddhist Studies: Volume in Honour of Professor J. W. de Jong on his Sixtieth Birthday (Canberra: Australian National University, Faculty of Asian Studies, 1982, pp. 1-10). I am happy that this third part is also appearing in a volume dedicated to a scholar who has contributed substantially to our understanding of Sanskrit sastras. 1.2 Vakya padiya (VP in abbreviation) 2.486, the first word of which I intend to discuss here, runs thus: parvatad agamam labdhva bhasya-bijanusaribhih / sa nito bahusakhatvam candracaryadibhih punah // The question of the precise import of this verse has given rise to a substantial body of literature extending over 125 years (Aklujkar 1978:9). As I have already examined this literature directly and indirectly in the publications mentioned above, I shall merely state here that I prefer to translate the verse along the following lines: 'Having acquired the traditional knowledge from parvata, Candracarya and others, who followed the indications in the Bhasya, again made it (i.e., the

Page #2
traditional knowledge) many-branched'. I should also clarify that in my view, as argued in the 1978 article, the verse was probably authored by a student of Bhartr-hari (B in abbreviation) and not by B as has been commonly supposed. 1.3 Although parvata is a common Sanskrit word with 'mountain, mountain range' as its definite meaning and it would not be incompatible in that meaning with the other words of VP 2.486, it has caused much reflection on the part of scholars. Goldstucker (1861:258), Weber (1862:161), Kielhorn (1874b:285-86; 1876c:244-45), and Bhandarkar (188385a/1933:184) refrain from translating it as 'mountain', although they must have known its most frequently attested meaning. Instead, they leave it untranslated as "Parvata" (note the capitalized initial) in their translations or paraphrases of the verse, giving the impression that they take it as a proper name, most probably that of a person. Nearer to our time, Joshi and Roodbergen (1976:xxxiii) and Joshi (1976:138) do the same, while Varma (1971a:206) and Laddu (1981:193-94) explicitly take parvata as a reference . to a particular person. Then there are scholars who, while not giving up the common meaning of the word, pay special attention to it in order to identify the mountain intended by the author of the verse. These are: Tara-natha Tarkavacaspati (1864: [introduction), p. 2; 1902: (introduction), p. 2), Peterson (1883-85:183), Ramakrishna Kavi (1930:239), Thieme (1956:20), Scharfe (1976:276), Cardona (1978:97, n. 36), and Bronkhorst (1983:393-97). 2.1 On the basis of Raja-tarangini (RT in abbreviation) verse 1.176, which is related in content to the VP verse we are discussing, Varma identifies parvata with King

Page #3
of 486, vyakaranagamah Abhimanyu of Kashmir (and also with the land of Kashmir; see note 2). This is patently absurd. According to the context daksinatyesu vyavasthitah, parvata must be in the South (3.3). A king of Kashmir (or the Kashmir country) is hardly likely to have been viewed as 3 Southern. Secondly, regardless of the reading one accepts of the RT verse, Abhimanyu cannot rightly be viewed as the giver or source of the agama in the case of Candracarya and others (Aklujkar 1987:228). 2.2 Laddu, too, takes parvata as a reference to an individual, but this individual is the Vedic, and to some extent Epic and Puranic, seer whose only distinction seems to be that he is always found in the company of Narada; otherwise, he is singularly characterized by a lack of individuality. The Classical authors, as far as I know, do not refer to him. He is not even remotely connected with grammatical studies. 3.1 Among those who prefer to take parvata as standing for a mountain, we find (a) the author of the Tika, Punyaraja or Hela-raja (Aklujkar 1974); (b) modern scholars like Raghunatha Sharma (1968:574), Raghavan Pillai (1971:146), and Subramania Iyer (1969:3; 1977:204) who reproduce the Tika explanation; and (c) other modern scholars such as Ramakrishna Kavi, Peterson, Thieme, Tara-natha Tarkavacaspati, Scharfe, Cardona, and Bronkhorst (precise references in 1.3) who either add to the Tika explanation or differ from it in the identification of the mountain. 5 3.2 Peterson, apparently following Muller, accepts the view that the parvata involved is "the hill of Chittore" in modern Rajasthan. The reason he gives in support of this 3

Page #4
identification is that the hill of Chittore "was a centre of learning for the southern country." However, the reason is hardly adequate. Peterson did not prove, and in fact could not have proved, that there were no other centers of learning for the southern country, irrespective of what he means by "centre of learning" and "southern country" (see note 3). Nor did he prove that there was something so special about the center of learning at Chittore that only it could have preserved the agama in question. He did not even attempt to answer preliminary questions such as: Is it known that Chittore specialized in the study of grammar or of Patanjali's Maha-bhasya (MB in abbreviation) in the early centuries of the Christian era? Was it considered or likely to be considered 'southern' by B or his students (1.2)? Was it, or was it at least believed to be, a repository of manuscripts of rare works? Is Candracarya said to have visited it? To propose an identification without raising even a few of these questions is to show disregard for the context of parvatat (see also the point I make in note 13). 3.3 The more elaborate statement made by Scharfe is open to the same charge. In suggesting that parvata should 6 be identified with Citra-kuta, he asks none of the above questions. In addition, he makes a series of unproved assumptions. Why identify parvata with Citra-kuta? Because, he tells us, the oral tradition of the MB was alive there. How do we know that the oral tradition was alive there when verse 485 says vyakaranagamah daksinatyesu grantha-matre vyavasthitah "The traditional knowledge (which would include the oral tradition) of grammar remained among the Southerners only in book form'? Because, according to 4

Page #5
If he Scharfe, "the verse speaks only about the South Indian Mahabhasya tradition which had withered to the point that oral instruction had ceased." Now, how can one assert this when there is no indication in the preceding verses 481-84 that their author intends to speak of Southern Vaiyakaranas and Northern Vaiyakaranas separately? Are we going to say that samksepa-ruci, alpa-vidya-parigraha, and akrta-buddhi grammarians existed only in the South or that the confusion Baiji and others caused in MB studies was confined to the South? In the latter case, how do we account for the fact that the names Baiji, Saubhava, and Haryaksa, although unusual, have nothing south Indian about them? In any case, why would the author of verses 481-86 be concerned with making a statement that applied only to south India? spoke of disintegration of MB studies only in the South, would it not follow that the state of these studies was satisfactory in the North (cf. Bronkhorst 1983:395)? Why, in that case, would he mention a single location in the North as the place where the agama was recovered?" How would his assertion accord with the commonly noticed state of affairs that South India in general managed to preserve transmission lines for a longer period (Aklujkar 1981:599600)? If the Northern tradition had not been disrupted, would Candracarya's acquisition of the agama be an achievement so remarkable as to deserve special mention? Why would Candracarya try to improve the state of MB studies in the South? Was he a Southerner? If he was, how would one account for the traditions that connect him with either Kashmir or 7 Bengal?8 5

Page #6
3.4 Even if Scharfe's contextually improbable interpretation of vyakaranagamah daksinatyesu grantha-matre vyavasthitah is accepted, how would it imply that Citra-kuta was the place where the agama was preserved? At most, the implication would be that the study of the MB did not suffer as much in the North as in the South. Furthermore, how does Scharfe establish Candracarya's association with Citra-kuta? He does so on the basis of examples in the Candra Vrtti pertaining to sutras 1.3.106-7. Therefore, the further assumptions are that: (a) Candracarya of VP 2.486 is identical with Candra-gomin; (b) the Candra Vrtti was in fact written by Candra-gomin; 10 (c) the sentence "we shall eat rice," given as an example in the above-mentioned part of the Candra Vrtti, means 'we shall eat rice only once'; (d) if rice was eaten only once before reaching Kausambi, the journey to Kausambi could not have taken more than a day; (e) if Candra-gomin thought the journey to Kausambi required only a day's travel in the direction of the eastern city of Pataliputra, his place of residence could have been anywhere up to fifty miles west of Kausambi;12 and (f) since Citrakuta is about fifty miles west of Kausambi, it must be the place where Candra-gomin resided. 3.5 Of the six assumptions outlined above, (a) and (b) are open to doubt, and those remaining, the ones that are crucial to establishing Candra-gomin's/Candracarya's presence at Citra-kuta, do not form a logically tight series of propositions. They constitute a novel, and for that reason commendable, attempt to determine a location, but one that is far from definitive. Assumption (c) forms their basis, and once its validity is called into question, the

Page #7
* others no longer remain tenable. Now, from my notes 11 and 12 here and from Bronkhorst 1983:397, it is evident that there are serious problems in relying on assumption (c). 3.6 To sum up, the impression I get from Scharfe's discussion of VP 2.486 and Candra Vrtti 1.3.106-7 is that the possibility of connecting the two occurred to him and the force of that new idea made him overlook many indications to the contrary. He has proved nothing more than the likelihood that the author of the Candra Vrtti was, at the time of writing his Vrtti, in an area which lay to the west of Kausambi and from which a journey to Kausambi and Pataliputra could be foreseen.13 4.1 The sika provides a specification of parvatat with the expression tri-kutaika-desa-varti-tilingaika-desat. This expression is printed as "vartti-tri-lingaio in Manavalli's (1887:285) and Raghunatha Sharma's (1968:574) editions and as vartino lingaio in Kielhorn's (1874:286) article. 14 An examination of the manuscripts of the sika reveals that the second variant reading is found only in a few inferior manuscripts and the first variant reading is not found in any manuscript accessible at present. Since this is the situation, an attempt like Thieme's (1956:20), which follows Kielhorn's reading and gives a siva-linga in the Tri-kuta region as the location of Candracarya's acquisition of the agama, must be set aside." 4.2 Besides, the word eka-desa does not really compound well with linga (=siva-linga) in the present context. A statement to the effect that 'the agama was acquired from a part of the siva-linga which stands on a part of the (mountain) Tri-kuta' is not likely to be intended, for there

Page #8
. ran- is no propriety in relating the acquisition to a part of the siva-linga. Thieme realizes this and remarks: "lingaikadesat is perhaps a mistake for simply: lingat, or, else, for lingavisesat 'from a particular linga,' the ekadesa-part having been nonsensically repeated, or put in place of -vicesat, from the preceding compound by a copyist." However, since the reading lingaio is objectively weaker than tilingaio, there is no need to indulge in the kind of speculation that Thieme's statement contains. Besides, a corruption of linga-visesad into lingaika-desad is transcriptionally improbable. The available manuscripts do not even indirectly indicate that deggaika-desad could have resulted from anything like oga-visesad (see note 15). 4.3 Explaining the sika identification is partly easy and partly difficult. The easy part comes later, viz., tilingaika-desat. Undoubtedly, tilinga is an earlier form. of the modern name Telangana and refers to a part of Andhra Pradesh having approximately the same boundaries as modern Telangana. An element of convenience in this fact is that . the identification of parvata as a mountain in Telangana will stand even if no agreement is reached on the identification of Tri-kuta (4.7) and on the connection of the expression tri-kutaika-desa-varti with the expression tilingaika-desat (see note 24). I shall, therefore, postpone the consideration of tri-kutaika-desa-varti and turn first to determining the specific area of Telangana that the sika author and, if the sika author has accurately preserved the tradition, the author of 486 are likely to have had in mind.

Page #9
4.4 It is indeed strange that the author of 486 should give the location of a major achievement by using a general term like parvata. There are two possibilities under which such use seems sensible. Either the author does not know the precise location or parvata does not carry for him in this instance the general meaning it usually conveys. The first possibility can be entertained in two ways: (a) The author is uncertain about the location beyond the fact that it was a mountain; or (b) the author cannot make up his mind as to which of the two or more mountains identified as the site of Candracarya's achievement was the true location. If (a) were the case, a form like kasmad api, indicating indefiniteness, would probably have accompanied the expression parvatat. Besides, a tradition identifying the parvata rather precisely would not, in all probability, have arisen. It seems proper, therefore, to entertain the first possibility only in the form of alternative (b). This I shall do in a later section (5.1). It is more convenient at the present stage of our discussion to explore the second possibility. 4.5 Let us assume for a moment that the ordinarily general term parvata could have been used in 486 as a singular term referring to a mountain. In other words, the question to be asked is: Is there a specific mountain in the South that could be referred to simply as parvata? I think Sri-parvata is the only mountain that meets the conditions contained in this question because (a) the stem parvata is more closely associated with it than with any other mountain frequently mentioned in the ancient and

Page #10
medieval literature of India. The other prominent mountains are quite freely referred to with compound names that contain one of the synonyms of parvata such as adri, giri, saila, or acala. Sri-parvata, on the other hand, does not commonly attract designations like sri-giri (see note 20c). In addition, (b) if a mountain is to be referred to by dropping a part of its name for the sake of brevity or for the sake of preserving the meter, the part to be dropped should be dispensable from some point of view. The stem sri, figuring in the name Sri-parvata, is so commonly employed as an honorific before names (including the names of other prominent mountains) that we should not be surprised if it was occasionally looked upon as an honorific even in the name Sri-parvata and thus thought to be dispensable. The thoughts expressed in the preceding paragraph are not mere speculation. There is in fact a tradition of referring to Sri-parvata simply as parvata (Dhere 1977:106, 121-23, 135). It may not have arisen for the reasons alluded to above and may be due to the importance of Sri-parvata as perceived by the people of India, particularly of south India ("Sri-parvata is the mountain; when someone uses the word parvata, it is Sri-parvata that first comes to mind'). However, what matters for us is that it is clearly there and that even the interpretation 'mountain of Sri' of the name Sri-parvata did not preclude occasional dropping of the component sri. 4.6 It will thus be seen that the second possibility (general term used particularistically) need not be viewed as a possibility in theory only. Besides explaining what looks like a lack of precision (it is not that an ancient 10

Page #11
22 author's usage is imprecise; it is our understanding of his usage that is deficient), it leads us to an identification that ties in well with tilingaika-desat of the Tika, for Sri-parvata is a part of Tilinga or Telangana (see notes 18 and 20). Furthermore, the identification agrees with indications in the Tibetan tradition and can be said to receive 21 confirmation in the accounts of Chinese travellers.1 According to the Tibetan tradition, Candra-gomin encountered the MB exposition when he returned from Simhala (Ceylon) to southern Jambu-dvipa (India). For the travellers from Ceylon to northern India, Sri-parvata was not only a natural, well-trodden, and time-honored region to pass through (Moticandra 1953, map preceding index), but it was also a holy and fascinating place on account of its association with siddhas and siddhis (Aklujkar 1982:6-7). In fact, there was a regular provision at Sri-parvata in the early centuries of the Christian era for receiving travellers (particularly Buddhist travellers from Ceylon) as we learn from an inscription of Vira-purusa-datta dated in the third 23 century A.D. (Sircar 1965:235). 4.7 Now, to turn to the first half, tri-kutaika-desavarti, of the identification in the Tika, Tri-kuta is said to be the name of at least four mountains in Sanskrit liter24 ature and Indian inscriptions:" (a) A mythical mountain in the North that forms the southern ridge of the Meru, bears the Tri-pathaga Ganga, is surrounded by the Milky Ocean, and has peaks made of iron, silver, and gold; cf. Maha-bharata 2.39.11, 2.82.11; Markandeya-purana 55.6; Bhagavata-purana 8.2.1ff; Sabda 11

Page #12
kalpa-druma pt. III, p. 74; Wilson 1894:141, n. 2; Kirfel 1954:10. (b) A mythical or semimythical mountain beyond the ocean in which Ravana's Lanka is situated; cf. Maha-bharata 3.261.53, 3.266.54-55; Ramayana 5.2.1, 6.30.18 (and passage 18 on p. 950 of Appendix 1), 7.5.21-22, 7.11.20; Vayu-purana 1.48.26-29; Panca-tantra, book 5, story 11 (Buhler's fourth ed., p. 63); Rudrata's Kavyalamkara 7.20; Bhoja's Srngaraprakasa, p. 419. (c) A range of hills near the west coast of India which extends from northern Konkan to the west of the present district of Nasik; cf. Mirashi 1955:xl-xli, 1963: 106-7; Gupta 1973:45, 246. Probably the same as the Tri-kuta mentioned by Saida Mahammada in his Kalpa-samuha (Dhere 1977:201). (d) A mountain in the eastern part of the Deccan (Mirashi 1975:186), probably the same as the one which B. V. Krishna Rao (Journal of the Andhra Historical Research Society 10:191; reference according to Gupta 1973:246) specifies as Kotappakonda near Kavur in the Narasaraopeta taluq of the Guntur district. 4.8 Of these four mountains bearing the name Tri-kuta, the one described in (a) is clearly alien to the spirit of this inquiry and can safely be ignored for that reason as well as for its uniform association with the North (3.3). Mountain (b), too, ceases to be historical if Ravana's Lanka is identified, as is generally the case,26 with the island of Ceylon (Sri Lanka) or some part thereof. As far as I can ascertain, no mountain in Ceylon was or is called Tri-kuta. Secondly, not only is there no corroboration of 26 12

Page #13
discovery in Ceylon in the other accounts of Candra's achievement, but the Tibetan accounts specifically state that Candra found the equivalent of the VP agama after he left Simhala or Ceylon (4.6). However, while we can dismiss the idea of a Tri-kuta understood to be associated with Ceylon, we cannot easily dismiss the Tri-kuta associated with Ravana's Lanka. The sika author mentions Ravana and raksas immediately after the statement of identification we are discussing. It is probable, therefore, that the Tri-kuta he had in mind was the one that figured in the descriptions of Ravana's kingdom. This observation leaves only two alternatives open to us: (a) that the sika author is correct in his statement of identification about parvata, but wrong in connecting the identified (Tri-kuta) parvata with Ravana; and (b) the sika author does not understand Ravana's Lanka to be Ceylon but a region near an historical Tri-kuta. Both these alternatives dictate that we determine independently which mountain or mountains can historically claim the name Tri-kuta. 4.9 Our attempt to determine this will of course be confined to mountains (c) and (d) of the list given in 4.7, for the available evidence does not lead us in any other direction. Of them, (d) is a suspect candidate for its claim to the name Tri-kuta has not generally been accepted by specialists of ancient and medieval Indian geography. Gupta (1973:246-47), agreeing with V. S. Ramachandra Murty's article in volume II (p. 45) of the Journal of the Andhra Historical Research Society, concludes that there is little evidence to support the identification of Kotappakonda with the Tri-kuta mentioned in historical records. 13

Page #14
4.10 Mountain (c) was known as Tri-kuta from at least the third century A.D., as the evidence collected by Mirashi and others establishes (4.7c). Our acceptance of it as the mountain intended, or originally intended, in the tradition preserved by the sika will not, therefore, be incompatible with the date of the Tika (Aklujkar 1982:3, 6, n. 4). It is also a mountain of the Daksina-patha, so its acceptance will not go against the expression daksinatyesu of verse 485 (see note 3). We know that Northern scholars, litterateurs, and their works reached the southern part of India through the territory adjacent to this mountain. Also, if we assume that in the sika author's perception Lanka was situated near this mountain (4.8, note 27), we will have accounted not only for the presence of Ravana by the side of Tri-kuta in his statement but we will also have arrived at a hypothesis that scholars such as Shah and Sankalia, interested in . determining the location of Valmiki's Lanka, have independently developed (cf. Cardona 1978:97). 5.1 Taking parvata in its usual sense of 'mountain', we thus arrive at two likely locations for the retrieval of vyakaranagama: Sri-parvata and Tri-kuta. Both of these agree with contextual indications such as daksinatyesu; were well known in the period (the early centuries of the Christian era) to which the concluding verses of the Vakyakanda belong; are likely to have been visited by Candra whether he started from Kashmir (as in the RT account) or returned from Simhala (as in the Tibetan accounts); had excellent potential for preserving the vyakaranagama manuscripts as areas of pilgrimage (Katre 1954:25-26); and were frequented by followers of both Brahmanism and Buddhism, 14

Page #15
* which jointly preserve Candra's memory. Besides, both are supported by a piece of information recorded as early as the sika (see note 13 for the value of this piece as evidence). 5.2 We have now reached a point in our discussion at which we can choose between the following positions: (a) The uncertainty regarding which of the two locations was actually the place of Candracarya's discovery belongs to the period of the sika; and (b) The uncertainty goes back to the period of the author of 486 (4.4). If we accept (b), it would be pointless to press this discussion further in the hope of being able to determine one location exclusively, for the author of 486 is our earliest source on what Candracarya achieved. If he is deemed to be undecided, there is practically no hope of our being able to decide what the location of the recovery of vyakaranagama was in the original account. However, before we accept alternative (b), let us remind ourselves that we have thus far proceeded on the basis of two assumptions: that tri-kutaika-desa and tilingaika-desa are separable parts of the sika phrase, and that the word tri-kuta can stand only for a mountain (see note 24). Both of these assumptions deserve further examination. 5.3 As is well known, readings in manuscripts should, as far as possible, be interpreted as they are. Assumption of loss or change of text matter should be resorted to only if the available reading cannot make contextually acceptable sense without such an assumption. Such is not the case with the sika phrase. It can be understood to mean 'from a part of Tilinga which (in turn) exists in a part of Tri-kuta'. There is nothing contextually incompatible in this meaning. 15

Page #16
Mere repetition of eka-desa would not be adequate justification for imagining a gap in the sika phrase or for emending it. If Tri-kuta refers to a mountain, it is not at all improbable that it would refer also to the region around that mountain." In fact, as Mirashi (1955:xli-xliv) notes, a name like Purva-tri-kuta-visaya 'East Tri-kuta district is found in the Anjaneri plates of Bhoga-sakti, and Tri-kuta is mentioned in a list of countries in the Ajanta inscrip- . tion concerning Harisena (475-510 A.D.). A dynasty of Traikutakas, who most probably derived their name from association with the Tri-kuta region, 33 is also clearly attested in historical records (Mirashi 1955:xli-xliv; Mulay 1972:16, 29; Gupta 1973:246). True, Mirashi observes that "The Traikutaka kingdom at its largest extent seems to have extended from the Kim in the north to the Krishna in the south, and to have comprised South Gujarat, North Konkan, and the Nasik, Poona and Satara Districts of Maharashtra," and thus it is not certain that the Traikutaka country included Tilinga. However, as the word "seems" in Mirashi's statement indicates, one cannot rule out the possibility that, the borders of the Traikutaka country could have at times been different.84 Evidence available for determining boundaries of political units in as early a period as the second to the sixth centuries A.D. rarely allows one to draw precise and final conclusions. As the Traikutakas emerged after the decline of the Satavahana empire, which included Tilinga, it is not unlikely that Tri-kuta, as the region of the Traikutakas, included or was thought to include Tilinga. Such a situation is especially likely to have prevailed in 16

Page #17
the early period of the Traikutaka dynasty, for which, as Mirashi observes, we have no records. 5.4 Thus, whether we accept a tradition of two locations for Candracarya's achievement or a tradition of one location would depend on whether we are willing to concede the possibility of the inclusion of Tilinga in Tri-kuta and on how natural we find the repetition of eka-desa in the Tika statement of identification. Acceptance of a tradition of one location does not involve assumption of textual corruption in the sika but it does involve the assumption that Tri-kuta, as a region, at one time included Tilinga. As such an acceptance leads us to Sri-parvata, it would better explain why the author of 486 used a general expression like parvata (4.5) in the singular number. But it would also enhance the possibility that even the author of 486 viewed Candracarya's achievement as a miracles and that the story of the recovery of vyakaranagama is simply one of the myths associated with Sri-parvata (Aklujkar 1982:6-7), thus having no specific historical value. Acceptance of a tradition of two locations, on the other hand, allows us to take Tri-kuta, in conformity with the indisputable part of the available evidence, as a region contiguous to the mountain range Tri-kuta. Such an acceptance leaves room for one further interesting connection. If B was a Maitrayaniya and if the Maitrayaniyas enjoyed a prominent presence in the Nasik area (see references collected in Bronkhorst 1983:396), the author of 486 probably belonged to the Nasik (i.e., the Tri-kuta) area. He could have then recorded in his composition a piece of local history and intended to refer to Tri-kuta by the word parvatat. 17

Page #18
6.1 At this point, I would like to return to the possibility entertained at the beginning (1.3) of this exploration of parvata, the one in which parvata could be taken to refer to a person. Although the specific identifications proposed by Varma and Laddu are unacceptable (2.1-2), I do not consider the possibility itself a weak alternative. According to the Brhacchankara-vijaya of Vidyaranya-svamin, quoted by Rama-tosana Bhattacarya in his Prana-tosini (p. 956), parvata is employed in the sense of a certain type of ascetic. Furthermore, Bhattoji Diksita, in his Siddhanta-kaumudi on the Paninian (2.2.34) rule alpactaram and the varttika abhyarhitam ca, gives as an example the expression tapasa-parvatau. This example is sensible only if parvata is comparable in some way to ta pasa 'ascetic'. Thus, it is not improbable at all that parvata carried a meaning of the form 'an ascetic who is found in a mountain area, a wild ascetic, an ascetic given to extreme modes of ascetic conduct'.39 If this meaning is assigned to parvata in 486, one can explain at least two things: the occurrence of the brahma-raksas 'brahmin ogre' element in the sika and Patanjali-carita accounts of Candra's achievement (see note 21a), and the use of a very general term like parvata by the author of 486 when we expect him to tell us how or where the vyakaranagama was recovered. If parvata is taken to be expressive of location, then the author has been unexpectedly imprecise; he has not specified anything beyond a mountain in the South. On the other hand, if parvata is viewed as a word used in the rather specific asceticdirected sense, the author cannot be faulted for having used too general or too wide a term. It is not his expression 18

Page #19
that lacks precision; rather, it is our knowledge of the expression he uses that needs to be widened. Similarly, if parvata stands for an ascetic haunting mountains or caves and having a wild, frightening, or repulsive appearance, we can see how a brahma-raksas came to be associated with the verse as a benefactor of Candra. I should also point out that the use of the ablative parvatat in 486 somehow seems more natural in the case of a sentient source of acquisition than in the case of a location. If the author of 486 intended to give us the location of the acquisition made by Candracarya, would he not have used the locative form parvate?40 These observations should explain why we cannot absolutely exclude the possibility of parvata being a reference to a person. 6.2 The word parvata in the sense elucidated just now could be a variant or a corrupt form of a secondary (taddhita) derivative parvata. Indications available in the Kasika, Nyasa, and Pada-manjari on Panini 2.4.23, 3.2.53, and 4.2.144 suggest that there was an older commentarial tradition according to which the word amanusya occurring in Panini did not convey a literal or etymological (yaugika) meaning of the form 'anything or anyone other than a human being but a specific conventional (rudha) meaning like 'a raksas, pisaca, etc.', that is, 'a member of a species thought to be similar to humans, but having strange capacities'. If this is so, one early, if not exclusive, understanding of parvata (and parvatiya) derived by Panini 4.2.144, vibhasamanusye, must have been something like 'a person of extraordinary or miraculous capacities associated with a mountain'. The probability that the pre-Kasika 19

Page #20
Paniniyas such as B and his disciples were aware of this meaning of parvata/parvata and used the word in that sense in their writings is thus strong." 42 6.3 One consequence of attaching the meaning 'mountain ascetic' to the word parvata, as in the case of identifying parvata with Sri-parvata (5.4), will be that the probability of tracing back the miraculous element in the accounts of Candra to B's time will be strengthened. This element is indeed widespread, as will incidentally become clear in part 4 (under preparation) of the present study. Moreover, since B or his students (1.2) were not totally against the possibility of knowledge through extraordinary means (Aklujkar 1970b:42-50), it is not incongruent to associate them with the miraculous. While this realization will deprive the Vakya-kanda verses of some of their value as a historical statement, it cannot be avoided by any historian who, rightly, evaluates his evidence before allowing it to lead him to definite conclusions. 7.1 There is a way in which both lines of interpretation, taking parvata as a reference to an individual and taking parvata as a reference to a mountain, come together. If Candracarya and others recovered, in the eyes of the author of 486, the agama from a siddha-like ascetic on a Southern mountain (most probably Sri-parvata, which was famous for such ascetics), then, while the first-level or initial interpretation of parvata would differ as 'a mountain ascetic' and 'a mountain', the ultimate import will be the same. We will have satisfied both expectations to which the context of 486 gives rise: (a) Who among the daksinatyas had the vyakaranagama? (b) Where in the South was the 20

Page #21
vyakaranagama found? The Tika explanation, in effect, accomplishes this, although it speaks of a brahma-raksas in the place of a parvata ascetic. 7.2 Parvatat is the only "down-to-earth" detail in VP 2.486. It is unfortunate that we are required to keep two lines of interpretation (mountain and person) open and to entertain two possible identifications (Sri-parvata and Tri-kuta) in the first line of interpretation. However, I believe such a course will serve the interests of future research better than a rushed, precise identification. Our ignorance of what actually happened may be disturbing but now it will at least have well-defined boundaries. If a preference must be indicated, I would rank as follows the identifications defended above. In the present state of our knowledge, the strongest identification appears to be 'parvata = an ascetic or brahma-raksas belonging to Sri-parvata'. Next in strength seems to be 'parvata = Sri-parvata', with the attendant assumption that some scholar, or scholarly community, on Sri-parvata had in his, or its, possession the manuscripts in which the agama of the MB was preserved. Last in terms of acceptability is the equation 'parvata = Tri-kuta (as mountain or region)'. This rank-ordering is based (a) on a consideration of how many independent lines of evidence point in the same direction, and (b) on whether an identification forces us to assume something which is not in the evidence and thus to sacrifice economy of explanation. The elements of what I consider to be the strongest identification can be related to the evidence as shown in table 1. 21

Page #22
a person, grammatical considerations who is associated with parvata, Kasika (Nyasa, Pada-manjari), Siddhanta-kaumudi <-- i.e., a mountain or its equivalent: sika, (possibly also the) Chinese travelogues a tall structurel temple, Chinese travelogues, Tibetan legends strongly associated with the word parvata, and sika, Chinese travelogues, Dhere well inside South India Tika, Tibetan legends, - Chinese travelogues who is seen as transcending the ordinary human condition, and Kasika (Nyasa, Pada-manjari), Siddhanta-kaumudi, Tika, Tibetan legends, Patanjalicarita who serves as a source well inside South India sika, Tibetan legends, Patanjali-carita Table 1 22

Page #23
It can be seen from this table how almost all the relevant evidence converges in the case of, or can be accommodated in the frame of, the first identification. For what I consider to be the next best identification, only the boxed part of the convergence holds good. On the other hand, the 'parvata = Tri-kuta' identification rests only on the statement in the sika and this, too, when the original form of that statement is assumed to be different from what it is now. Notes * Conclusions reached in this part were reported in the form of a short paper on 21 March 1983 at the 193rd meeting of the American Oriental Society held at Baltimore. The author was able to study many of the sources used in the article because of financial assistance provided at various times since 1969 by the University of B.C. Humanities and Social Sciences Research Committee, the Social Sciences and Humanities Research Council of Canada, the Shastri Indo-Canadian Institute, the American Council of Learned Societies, and the Alexander von Humboldt Stiftung of the Federal Republic of Germany. I also wish to acknowledge the dispassionate response to my criticism of his position that Professor Hartmut Scharfe gave in his letter of 6 November 1987. The response based on the 26 October 1987 version of the paper helped me in clarifying some points and correcting a few errors. 23

Page #24
1. The implicit refusal to take parvata as 'mountain' is especially remarkable in the case of Weber and later researchers mentioned here. Weber (1862:161n) and Kielhorn (1974b) knew at the time of their relevant writings that Tara-natha Tarka-vacaspati had given a specification of parvata as Citra-kuta or Tri-kuta. An earlier edition of Tara-natha's Sabdartha-ratna (the reference according to the third edition available to me is 1902:2) is mentioned by Weber, and Tara-natha's edition of the Siddhanta-kaumudi (1864:2) and Weber's article are mentioned by Kielhorn. Kielhorn also knew that Punya-raja, an ancient commentator of VP 2.486, had understood parvatat to be a reference to a region. Finally, it is beyond doubt that Bhandarkar and most of the later scholars mentioned here exhibit an awareness of Kielhorn's writings. 2. Varma (1971a:350) contradicts himself when he also takes parvata in its sense of 'mountain' and identifies it with Kashmir (as mountainous country) of King Abhimanyu's time. 3. (a) Any directional adjective like daksinatya 'southern is relative in its application, for what is southern to one user can very well be northern to another. However, it is reasonable to assume that in most Classical Indian uses of daksinatya the reference would be to persons or objects of the Daksina-patha unless a further specification is available in the context (Sircar 1960:172, n. 2). Another consideration one should keep in mind in the case of VP 2.486 is that, as its author belonged to the Paninian tradition of Sanskrit grammar, his under 24

Page #25
standing of daksinatya is not likely to be very different from that of Patanjali (Kielhorn's ed., vol. 1, p. 8), who indicates that the author of the varttika yatha laukika-vaidikesu was a daksinatya (Cardona 1976: 268-69). The third helpful consideration in ascertaining the common meaning of daksinatya in the Classical period is the remark by several authors to the effect that the word cora/caura in the language of the daksinatyas means odana 'cooked rice' not 'thief'; cf. Prabha-candra, Nyaya-kumuda-candra, p. 547; Jayantabhatta, Nyaya-man jari, p. 242; and Abhinava-gupta, Tattva-viveka on Para-trimsika 5-9ab, p. 125, who distinguishes between saindhavas and daksinatyas. With reference to the meaning of the same cora, Vadi-devasuri (Syad-vada-ratnakara, p. 703) distinguishes between gurjaras and dravidas; the latter must obviously be daksinatyas in his view. Probably Sridhara (Nyayakandali, p. 215) also has a relevant observation to offer in this case, although I cannot verify the reference at present. (b) For a determination of the reference of Daksina-patha, see Law 1932/1979:xv, xix, 3-4, 44, 48, 60, 66; Raychaudhuri in Yazdani 1960:3-4; Sircar 1960: 14-15, 52, 57, 73, 172-73, 178, 242; Gupta 1973:8-9; and Simha 1974:85. 4. (a) parvatat tri-kutaika-desa-varti-tilingaika-desad iti. (b) Variant readings available for the sika phrase in (a) are given in note 15. The possibility of its 25

Page #26
having suffered some damage in manuscript transmission will be discussed in 4.2, note 24, and 5.3. 5. Peterson does not specify how, when, or where Muller's suggestion about parvata was made. 6. (a) Tara-natha Tarka-vacaspati offers the same identifi cation as Scharfe's without giving any reasons to support it, as if he was simply explaining or replacing the reading tri-kuta of the sika with citra-kuta. (b) Scharfe does not indicate awareness of the information given in (a), which, one expects, would have been known to him through Weber 1862. (c) Scharfe's presentation differs also in that he further identifies Citra-kuta with Rama-giri, presumably the same Rama-giri as the one mentioned by Kalidasa in his Megha-duta. It should, therefore, be noted that the identification of Citra-kuta with Rama-giri is not universally accepted (cf. Gupta 1973:101-3) as Scharfe seems to have thought. (d) Bronkhorst (1983), who also does not indicate awareness of the information in (a), follows Scharfe in a strange way. On pages 393 and 395, he has Candra going to the Himalaya to acquire "[correct] traditional knowledge" or "The Patanjalian oral tradition." On the other hand, on page 397, he has Candra at least contemplating a journey through Kausambi, as in Scharfe's view, but not necessarily residing at Citra-kuta, as is Scharfe's view. In other words, Bronkhorst uses Scharfe's discussion to assign Candra to Gujarat or north Maharashtra but not to explain Candra's retrieval of the agama. In so doing, he severs the direct connec 26

Page #27
tion between VP 2.486 and Candra Vrtti 1.3.106-7 that Scharfe seeks to establish. Thus, his view is similar to Varma's (2.1) and is partly based on what Scharfe thought to be justified. As my discussion shows, both Varma and Scharfe have put forward views that violate contextual considerations and go against the available evidence about Candra's achievement. Bronkhorst's composite view, therefore, stands doubly refuted and need not be discussed separately. 7. (a) Note that the option of considering Citra-kuta as relatively southern is not open to Scharfe because in his view the verses deal with the withering of the MB tradition in the South and hence the location of agama retrieval must be in the North; cf. Bronkhorst 1983:395. (b) In his letter of 6 November 1987, Scharfe makes the following remarks, which serve to clarify his 1976 interpretation of verse 485: "... the VP stanza says that the southern tradition existed only in manuscripts; that implies that it existed in the north also orally, and that would be the place to get this tradition. If B had wanted to say that these manuscripts were all that existed of the Mbh [= Maha-bhasya] tradition, it would be odd to refer to South India at all. He would rather have said: "The tradition existed only in manuscripts." My preceding remarks should suffice to establish that this interpretation is contextually improbable. It is also flawed by lack of logical implication. There is no incompatibility between (i) saying that the MB tradition existed only in the form of manuscripts, and (ii) saying that the MB tradition existed only in the South. Third, 27

Page #28
there is no justification for taking agamah daksinatyesu grantha-matre vyavasthitah as synonymous with daksinatyah agamah grantha-matre vyavasthitah, as Scharfe does. 8. (a) See S. K. De 1938:258-59, and Simha 1969:242-43, for a discussion of Candra's/Candra-gomin's province of birth. (b) The presence of Gaudas in early Kashmir is proved by RT 4.323-35 and Ksemendra, Desopadesa, adhyaya 6. The latter refers to students from Bengal who had come to Kashmir to study Sanskrit and Sanskritic branches of knowledge. The grandfather of the remarkable author Jayanta-bhatta was also a Gauda who migrated to Kashmir. (c) Bronkhorst (1983:396-97) actually accepts Candra's residence in the South, i.e., in Gujarat or north Maharashtra. His discussion of Candra's locality shows no awareness of the literature referred to in (a). He obviously assumes that Candracarya is definitely the same as Candra-gomin, that a Buddhist like Candra-gomin must have lived in an area inhabited by followers of the Maitrayaniya recension of the Veda, and that the possibility of the localities of Candracarya and Candra-gomin being different need not be entertained! 9. There are some minor but complex problems, in addition to the major problems indicated here, associated with Scharfe's use of the examples in the Candra Vrtti and other related texts. It would be better to take them up for discussion in a separate publication. 28

Page #29
10. The problems connected with the acceptance of Candragomin's authorship of the available Candra Vrtti are discussed. in Birwe 1968. 11. It is not clear to me exactly how Scharfe deduces from "We shall eat (bhoksyamahe) rice" that Candra-gomin's was a one-day journey to Kausambi. Since Scharfe contrasts this example with others meaning "We shall eat twice," I have assumed that in his view "We shall eat rice" implies 'We shall eat rice once' and one rice meal implies one day's journey. However, who is likely to eat only one rice meal or only one meal per day? Does the example presuppose a Buddhist monk as the speaker of the sentence? In that case, why is there nothing in the context of the example that would suggest restriction of the example to a monk's situation? If, on the other hand, the inference that the distance was to be covered in one day is to be drawn only from the grammatical form bhoksyamahe, why do other grammarians specify the number of times they would carry out a particular act in a similar situation? Why does the Kasika (3.3.136, 138) see nothing wrong in changing odanam bhoksyamahel bhoktasmahe to dvir odanam bhoksyamahe/bhoktasmahe, although it, too, speaks of a journey through Kausambi to Pataliputra? Scharfe (1976:275) glosses over the problem the Kasika examples pose by stating that they "appear to be compounded renderings of the sentences found in the Candra-vrtti and Abhaya-nandin's Mahavrtti ... the example must be regarded as one of those standard examples handed down through the times and used without respect for its actual [factual?] accuracy." Before 29

Page #30
blaming the evidence that inconveniences his speculation, however, Scharfe should have asked himself: Why would the Kasika opt for compounded renderings? Is it certain that the Kasika was written later than the Candra Vrtti and the Maha-vrtti? I am aware that, since Kielhorn's 1886 article, it has generally been held that the Kasika is influenced by the Candra vyakarana. However, as I pointed out at the time of the 192d meeting of the American Oriental Society in 1982, this view may not be valid. Recently, Bronkhorst (1983:368) has argued that there is no Candra influence on at least the sutra text in the Kasika. As for Abhaya-nandin, he has been placed in the post-Kasika period by the historians of Sanskrit grammar on very good grounds (Yudhisthira Mimamsaka, 1973/samvat 2030, vol. 1, 3d ed., pp. 458-71, 584-86; also pp. 9, 24, and 51-52 of the introductions. contributed by Vasudeva-sarana Agravala, Nathu-ramaji Premi, and Yudhisthira Mimamsaka to Sambhu-natha Tripathi's edition of Abhaya-nandin's Maha-vrtti, Kashi: Bharatiya Jnana-pitha, 1956). Moreover, two of the Kasika phrases that could be said to have been inspired by the Maha-vrtti are not as precise as their counterparts (compare tatra saktun pasyamah and yukta adhyaimahi with tatra... dvih saktun pasyamah and yukta dvir adhyaimahi). If the Kasika were at the borrowing end, why would it sacrifice the existing precision? That the Kasika elsewhere copies examples from such works as the MB hardly constitutes sufficient proof for the present purpose. 30

Page #31
12. The questions I have about this particular assumption are: Was travel in the range of fifty miles generally viewed by Candra's contemporaries as possible in one day? Would an ordinary man eat only once during a fifty-mile journey when the journey is to be completed with ancient means? Would it be correct to hold that the Candra Vrtti example is from the perspective of a monk who eats only one meal per day? 13. The attitude that may be implicit in Peterson's and Scharfe's attempts at identification also needs comment. Both these scholars write as if the Tika identification does not exist or need not be taken seriously. This is hardly a justifiable view to take of a piece of information that is about one thousand years old; that is unlikely to have been given unless it was known to earlier students and commentators of the VP (cf. anusruyate in the Tika); and that is in itself very plausible (especially because it contains the names Tri-kuta and Tilinga attested in a number of inscriptions, etc.). If Peterson or Scharfe had pointed out that the Tika detail offends common sense (has a mythic character like some other details in the Tika) or is contradicted by another equally old source, they would have been justified in setting it aside. By not following any such procedure they give the impression that information preserved by scholars directly in the line of transmission deserves little or no credence. This is an unwarranted and unprofitable position to take. Moreover, since their own views are based on the use of some indirectly related tradition, they are open to a charge 31

Page #32
of self-serving eclecticism. The same comment can be extended to Bronkhorst's discussion. 14. Mirashi (1955:xl) follows a reading "varti-tri kalingaikao not found in Manavalli's edition (1887), which Mirashi apparently used. As note 15 establishes, there is some trace of this reading in the generally inferior manuscript E7 and its transcript E14 but there is no clear support for it in any of the available manuscripts. Even in E7, the presence of ka is due to the copyist's inability to decide whether his exemplar reads talti or ka. It would appear that Mirashi read the name Tri-kalinga in his source through an oversight caused by the memory of that name, which occurs in other histori cal records handled by him. 15. Following the designations given in Rau 1971, the devia tions noticed in manuscripts from the sika text accepted here can be specified as follows: (a) trikutekadesadeg E1, trikutaidesao E25, trikutaiddesao E5, trikatadesao E7, trikute desadeg E14.. (b)' Ovarttinetilimgai E8, varttinotilimgaikadeg E1, 'vartinolimgaikadeg E3, E9, E9a, E22, "varttinolimgaikao with no changed to ti in the margin E10, ovartitakalimgaikadeg E7, E14, 'varttitaligaukadeg E5, 'vartiti limgaikadeg E20, 'varttine (or te] ligaikadeg E23, varttitelimgaikadeg E4, E21, E25, ovavartitilimgaikadeg E6. The difference of one t between varti and vartti is of no consequence, as anyone conversant with common features of Sanskrit orthography knows. The generally reliable manuscripts of the sika are E4, E6, E11, E13, E15, E16, E21, E24, and E25, as my 32

Page #33
planned critical edition of the Tika will establish. They agree in reading the Tika phrase as I have accepted it in this study, except for the redundant writing of va in E6 and the presence of te in the place of ti in E4, E21, and E25. The last variation does not make any significant difference. Tilinga and Telinga are wellattested names of the same Telangana region. The manuscripts designated E2, E17, E18, E19, and E26 by Rau do not contain the required portion. Rau's E12 is a modern transcript of E2. E9a did not somehow come to Rau's notice but was made available to me at the library of the Sanskrit College in Calcutta. 16. (a) I have not yet had time to determine which, if any, of the manuscripts accessible to me Manavalli followed in his editio princeps of the Vakya-kanda-tika. It is possible that, as a nineteenth-century pandit, he showed preference for the Sanskrit-sounding form trilinga over the Prakrit or desya-sounding forms tilinga and telinga. (b) trilinga is noticed elsewhere as a variant of tilinga (see note 18cd). Its acceptance would not affect the subsequent discussion in this study. 17. For a critical examination of the other aspects of Thieme's view, see Aklujkar (1986). 18. (a) This spelling is according to the National Atlas of India. The spellings "Telingana," "Telengana," etc., are also noticed. (b) For information on Tilinga or Telangana, see Raychaudhuri in Yazdani 1960:27-29; Sircar 1960:16, 71, 75-76, 88, 89; Gupta 1973:37; and Schwartzberg 1978:137, maps a and b; 38, map b; 39, map c; and 100, map a. 33

Page #34
(c) tri-linga, tirilinga, etc., are attested as variants of tilinga. A secondary derivative tailanga is also found. (d) "The name Tirilinga (Skt. Tri-linga) is traditionally derived from three shrines (lingas), namely, those at Kalesvaram, Srisailam and Draksarama...". (Gupta 1973:37). Bhimesvara in the place of Draksarama is given as the third linga leading to the formation of Tri-linga by Luders in Epigraphia Indica 6, no. 10 (1900-1): 93. 19. Parvata is also taken as a proper name of a region in northwestern India. According to Schwartzberg (1978: 184a), it refers to Jammu. It does not seem likely, however, that this reference was current in the days of B and his disciples. Besides, this Parvata was not likely to be thought of as southern in the period with. which we are concerned (see note 3). 20. (a) For references to Sri-parvata in Sanskrit litera ture, inscriptions, etc., see Aklujkar 1982:8. For the importance of Sri-parvata in Indian religious history, see Dhere 1977:105-9, 179-80, 194-95.. (b) The geographical location of Sri-parvata is shown in Schwartzberg 1978:22, map a; 27, map b; and 32, map a. (c) Sri-parvata and Sri-saila have been treated as practically identical in the publication mentioned in (b), as well as in some Sanskrit works and modern research publications. Sri-parvata would seem to be the older of the two names and has a general as well as a specific reference. In its former role, it seems to 34

Page #35
stand for the mountain range toward the southwestern extremity of which the Brahmanical site of Sri-saila took shape (cf. Sircar 1965:235, n. 3; Schwartzberg 1978:27, map. b). In the latter role, Sri-parvata appears to be a predominantly Buddhist site near Nagarjunakonda in the northeastern part of the same Nallamala (alternative spellings "Nallamalur" and "Nalamalai") mountain range. (d) The location of Sri-saila is shown in plate 35 of the National Atlas of India and in Schwartzberg 1978:21, map a; 25, map a; 26, map a; 27, map b; 31, map a; 34, maps a and b; 38, maps a and b; 41, map a; 47, map a; 99, map a; and 140, map a. 21. (a) Accounts of Candra's discovery, including those in Tibetan, will be compared in part 4 of this study. (b) Fa-hsien's report of his travels in India between 399 and 414 A.D. contains a hearsay description of a five-tiered, or five-storied, mountain monastery (Po-lo-yu or Po-lo-yue) called approximately parava in the local language. At least some specialists of the travelogues of Chinese visitors to India have come to the conclusion that this monastery is identical with the one (po-lo-mo-lo-ki-li) which Yuan Chwang (=Hiuen Tsiang) visited about two hundred years later and which was on Sri-parvata, associated with Nagarjuna. Accordingly, Watters (1905:200-1; cf. Yazdani 1960:146-47) sees in Fa-hsien's parava the Sanskrit word parvata (not paravata 'pigeon' or 'columbarium' as Fa-hsien thought) standing for Sri-parvata. 35

Page #36
22. The tentative assumption here is that Candra-gomin is the same individual as Candracarya or was confused with Candracarya by this time. 23. Thus, Ramakrishna Kavi was on the right track when he included the guess "somewhere near Sri Parvata, probably in the Andhra country" in his 1930 article (p. 239). However, he did not give any evidence or reasoning in support of his remark or provide a geographical specifi cation of Sri-parvata. 24. (a) Another possible identification of Tri-kuta, not as a mountain but as a region or country, will be taken up in 5.2. The assumption in my discussion at this point is that the sika phrase tri-kutaika-desa-varti-tilingaikadesad could originally have been different (a suspicion based on the repetition of eka-desa, as on Thieme's part, but not resolved in the same way as Thieme proposes). The original form of the phrase was perhaps. as follows: tri-kutaika-desa-[varti...ity eke. anye tv ahuh . . .-/varti-tilingaika-desad iti, i.e., the text available at present could be a result of haplography caused by the recurrence of varti. Resting on this assumption is the point of view that two possible lines of identification, one based on the name Tri-kuta and the other based on the name Tilinga, should be kept apart at this stage of the discussion. (b) As the Sabda-kal pa-druma (p. 655) notes, Sanskrit lexicons give the names Tri-kakud, Suvela, Tri-mukuta, Tri-srnga, and Citra-kuta in the same group as Tri-kuta. Except for Suvela, which may be helpful in 36

Page #37
determining the location of the Tri-kuta mentioned in Rama narratives, the other mountains seem to have been grouped with Tri-kuta only on account of the structural similarity of the names. 25. This mountain would be very close to Sri-parvata (see note 20) if not identical with a part of Sri-parvata understood as a mountain range. 26. Mirashi (1975:205-19) establishes that, in the percep tion of many Classical authors, Lanka was located in Simhala. 27. (a) It would be simplistic to hold that only one tradi tion regarding the location of Lanka was current in as diverse and large a country as India and throughout as long a period as that of Classical Sanskrit literature. (b) The view that, in Valmiki's perception, Ravana's Lanka was situated in a place other than Ceylon, probably near the Vindhyas, has been expressed for many years. Shah (1976) and Sankalia (1982) are the most recent exponents of that view known to me. 28. If, however, the identification is correct with refer ence to a part of the historical evidence available (see note 34) and we have in fact two Tri-kutas (approximately on the western and eastern borders of the Satavahana empire), then we will have open to us the alternative 'parvata = Tri-kuta = Kotappakonda', but it will not materially be very different from the alterna tive 'parvata = Sri-parvata' (see note 25). 29. For understandable reasons, the ancient traveller in India had to move along those areas where the height of mountains like the Vindhyas was either not to be met . Wan 37

Page #38
with or was negotiable. A sort of western corridor for the movement of men and goods, therefore, seems to have developed near the Nasik area. 0. (a) Tri-kuta is near such ancient Buddhist sites as Kanheri and Ajanta, and near Brahmin holy places like Tryambakesvara. The appeal that Sri-parvata had for Buddhists and Brahmins can be judged from the information and sources given in note 20. (b) I intend to discuss the problem of the religious affiliation of (B and) Candracarya in a separate public cation. One's conclusions in this regard will naturally depend on whether Candracarya is held to be identical with Candra-gomin and, to some extent, on what view one takes of the religious affiliation of B and his disciples. 31. Subramania Iyer's (1977:li, 204) renderings of the sika phrase, "Trilinga country, near the Trikata mountain" and "Trikata," are not as precise as they should have been. 32. The sika author could also be said to have a region rather than a mountain in mind. In the latter part of his identifying phrase, eka-desa is compounded with tilinga, the name of a region. The same is, therefore, likely to be true of eka-desa in tri-kutaika-desa. 33. Thus, Schwartzberg (1978:137, map a) is right in showing Tri-kuta as a region. 34. As possible evidence to the effect that Tri-kuta could have at times denoted a larger area than the one determined by Mirashi, note the following: 38

Page #39
(a) The word tri-kuta, probably referring to an area, occurs in a third-century A.D. plate discovered in the Vizagapatnam district; cf. Hultzch, Epigraphia Indica 3, no. 3 (1894-95):19-20. (b) A temple of Tri-kutesvara is said to be near the village Kondakavuru, which is eight miles south of Narasaravupela; cf. Luders, Epigraphia Indica 6, no. 12 (1900-1):116, 127. (c) In a Visnukundin inscription, Prince Madhavavarman II, whose headquarters were at Amara-pura (= Amaravati near Sri-parvata?) has been called trikuta-malayadhi pati; cf. Sircar 1960:189. (d) If there were in fact two historical Tri-kutas, (c) and (d) of our list in 4.7, it is possible that the area between them governed by rulers such as the Satava. hanas was known as the country of the Tri-kutesvaras or as Tri-kuta. (f) There is a temple of Tri-kutesvara existing from at least 1191 A.D. at Gadag in Dharwar district; cf. Kielhorn, Epigraphia Indica 3, no. 30 (1894):217-20. (g) Mishra (1973:139), on the basis of Sumpa MkhanPo's Pag-Sam-Jon-Zang (p. lxvi) edited by S. C. Das, points out the probability that there once was a Traikutaka vihara in West Bengal. 35. There are no expressions indicating a miracle in 486 (Aklujkar 1981:600), although a miracle is at least suggested in all the other known accounts except the one in the RT." 36. Under this possibility, what the author of 486 would, in effect, be saying would be this: "Candracarya and others 39

Page #40
came from the North and acquired from the mountain the agama that had survived only in a written form among the daksinatyas, that is, among us'. B, as the author's teacher, would then be the person who built on the foundation prepared by Candracarya (verse 487) and revived the tradition, at least in the Nasik area. 37. (a) 'vaset parvata-mulesu praudho yo dhyana-dharanat / sarat saram vijanati parvatah parikirtitah // 'He who lives in the foot-areas of mountains, is advanced/ mature on account of practice of meditation, and knows the quintessence, is said to be parvata'. (b) According to the Hindi Sabda-sagara (vol. 6, p. 2883), parvata is a samnyasin who belongs to the Dasanami sect and lives at the foot of a mountain, devoting himself to meditation. This information is corroborated by the Brhacchankara-vijaya passage quoted in the. Prana-tosini. The names of the ten samnyasins are given in that passage in this order: tirtha, asrama, vana, aranya, giri, parvata, sagara, sarasvati, bharati, and puri. (c) Prior to quoting the verse in (a) as a definition found in the Avadhuta chapter of the Prana-tosini, the Sabda-kalpa-druma (III, pp. 77-78) informs us that parvata in this sense is "a specific kind of disciple/ follower of Mandana-misra, who was the disciple of Samkaracarya." I do not know the basis for the specific association with Mandana-misra. 38. The author of the Bala-manorama commentary is puzzled by Bhattoji's example, as can be seen from his comment: parvatasya sthavara-janmataya tapasasya tad-apeksaya 40

Page #41
bhyarhitatvam bodhyam. bhasye tu mata-pitarav ity udahrtam. 'Since a parvata is born of a stationary entity (or is stationary by birth), a tapasa should be understood as worthy of respect in comparison to him. The example in the Bhasya is, however, mata-pitarau'. The Tattva-bodhini commentary does not even indicate awareness of the example ta pasa-parvatau. 39. (a) See note 37. (b) It might be asked if this sense of parvata could be as old as the author of 486. I am not aware of any evidence that would conclusively prove this sense to be so old. However, since Vidyaranya's list of the ten classes of monks is not very logical in the first seven location is the criterion, whereas in the last three it is not; there is also overlapping between asrama, vana, and aranya, and between giri and parvata), it may be inferred that it has old as well as new terms simply put together. See also 6.2. 40. (a) Or has he used parvatat for parvate simply because, in the phrase parvate agamam labdhva, the word parvate, reduced to parvata by samdhi, would have disturbed the meter? (b) Scharfe (1976:276, n. 21) simply dismisses the view of "an eminent Indian Sanskritist" that parvate would have been proper if a mountain was intended and that the use of parvatat indicates that a person was probably intended. Scharfe does not point out, as he should have, any instances in which an 'ablative + labh construction is used to give the location (as distinct from a conscious source) of an acquisition. 41

Page #42
Note that, for the resolution of the grammatical point involved here, it is not necessary that the acquisition be that of an agama. I am not making the unreasonable expectation that another occurrence strictly of the type 'ablative of location + agamam + labdhva' be pointed out, for I do realize that there would be very few occasions requiring one to speak of the acquisition of an agama. Similarly, any form of labh or a synonymous root would do. The point is that Scharfe should have cited at least one parallel instance favoring the 'location' interpretation before setting aside so casually the observation he attributes to an eminent Indian Sanskritist. In his letter of 6 November 1987, Scharfe informs me that the observation was made by V. Raghavan at the time of the American Oriental Society meeting held at Santa Barbara. (c) As the correct reading of RT 1.176 is candracaryadibhir labdhvadesam tasmat (Aklujkar 1986), the source in that 'ablative + labh' construction is King Abhimanyu, i.e., the source is sentient. 41. (a) Since the etymological or wider meaning of amanusya is something anyone knowing Sanskrit would think of right at the outset, it does not need a tradition to come into existence, but the narrow meaning is so specific and unexpected that it is not likely to be found in commentaries like the Kasika unless a tradition to the effect that it was intended by Panini existed. (b) From Patanjali's comments on 3.2.52-3, it seems that commentators of Panini had begun by Patanjali's 42

Page #43
time to entertain the possibility of taking amanusya in a literal sense. The Kasika does not appear to be the first commentary to do so. 42. As the secondary derivatives tend to acquire a general meaning (such as 'belonging to, pertaining to, associated with') in the history of Sanskrit, parvata (and parvatiya) probably expanded beyond the sense 'a parvata-dwelling ogre/ghost/ascetic' to include items associated with mountains in general; i.e., parvata (and parvatiya) could also be used as adjectives of things such as water and fruit found on a mountain. It is likely that because of this semantic development the Kasika decided not to follow consistently the narrow meaning of amanusya it advocated in the case of Panini 2.4.23. Select Bibliography Items for which publication details have been specified in the earlier parts of this study or can be gathered from Cardona (1976) have not been included here. The same applies to well-known works like the Ramayana and well-known series such as Epigraphia Indica. Aklujkar, Ashok. 1987. "Raja-tarangini 1.176." In Ancient Indian History, Philosophy and Culture, Essays in Memory of Professor Radhagovinda Basak Vidyavacaspati. 43

Page #44
(Eds.) Pratap Bandyopadyay and Manabendu Banerjee, 224-45. Calcutta: Sanskrit Pustak Bhandar. Bronkhorst, Johannes. 1983. "On the History of Paninian Grammar in the Early Centuries Following Patanjali." Journal of Indian Philosophy 11:357-412. Cardona, George. 1976. Panini: a Survey of Research. The Hague, Paris: Mouton Indian edition. Delhi: Motilal Banarsidass. De, S. K. 1938. "Candra-Gomin." Indian Historical Quarterly 14:256-60. Dhere, Rama-candra Cintamana. 1977. Cakra-pani (adya marathi vanmayaci samskrtika parsva-bhumi). Poona: Visva-karma Sahityalaya. In Marathi. . Gupta, Parmanand. 1973. Geography in Ancient Indian Inscriptions (up to 650 A.D.). Delhi: D. K. Publishing House. Hindi Sabda-sagara. Revised ed. 1969. Kasi Nagari Pracarini Sabha. Prayag: India Press, Ltd. In Hindi. Joshi, S. D. 1976. "Sanskrit Grammar." In Ramakrishna Gopal Bhandarkar as an Indologist, a Symposium. (Ed.) Dandekar, R. N., 113-42. Poona: Bhandarkar Oriental Research Institute. Joshi, S. D., and Roodbergen, J. A. F. 1976. Patanjali's Vyakarana-Mahabhasya Anabhihitahnika (P. 2.3.1-2.3.17). Poona: University of Poona. Publications of the Centre of Advanced Study in Sanskrit, class C, no. 11. Katre, S. M. 1954. Introduction to Indian Textual Criti cism. Poona: Deccan College Post-graduate and Research Institute. Deccan College Handbook Series, no. 5. 44

Page #45
Kirfel, Willibald. 1954. Das Purana vom Weltgebaude. Bonn: Selbstverlag des Orientalischen Seminars der Universitat Bonn. Laddu, S. D. 1981. "A Reconsideration of the History of the Mahabhasya." In C.A.S.S. (Centre of Advanced Study in Sanskrit] Studies, no. 6, pp. 187-97. Poona: University of Poona. Law, Bimala Churn. 1932. Geography of Early Buddhism. London: Kegan Paul, Trench, Trubner & Co. Reprint 1979. New Delhi: Oriental Books Reprint Corp. Manavalli, Gangadhara Sastri (Ed.). 1887. Vakya padiyam ... sri-bhartrhari-... viracitam sri-punyaraja-krta- ... tika-yutam. Benares: Braj B. Das & Co.' Benares Sanskrit Series, nos. 11, 19, 24. Mirashi, Vasudev Vishnu. 1955. Inscriptions of the Kalachuri-Chedi Era. Corpus Inscriptionum Indicarum, vol. 4, pt. 1. Ootacamund: Government Epigraphist for India. - 1963. Inscriptions of the Vakatakas. Corpus Inscriptionum Indicarum, vol. 5. Ootacamund: Government Epigraphist for India. - 1975. Literary and Historical Studies in Indol ogy. Delhi: Motilal Banarsidass. Mishra, Vibhuti Bhushan. 1973. Religious Beliefs and Prac tices of North India During the Early Mediaeval Period. The Hague: E. J. Brill. Moticandra. 1953/1966. Sarthavaha. Patna: Bihara-rastra bhasa-parisad. In Hindi. Mulay, Sumati. 1972. Studies in the Historical and Cultural Geography and Ethnography of the Deccan. ... 45

Page #46
Poona: Deccan College Post-graduate and Research Insti tute. Deccan College Dissertation Series D, no. 68. Rama-tosana Bhattacarya. Prana-tosini. (Ed.) Jivananda Vidyasagara Bhattacarya. 3d ed. Calcutta: n.p., 1898. Dates of earlier editions not known. Sankalia, Hasmukhlal Dhirajlal. 1982. The Ramayana in the Historical Perspective. Delhi: Macmillan. Scharfe, Hartmut. 1976. "A Second 'Index Fossil' of Sanskrit Grammarians." Journal of the American Oriental Society 96:274-77. Schwartzberg, Joseph E. (Ed.). 1978. A Historical Atlas of South Asia. Chicago, London: University of Chicago Press. Association for Asian Studies Reference Series, no. 2. Shah, U. P. 1976. "The Salakatankatas and Lanka." Journal of the American Oriental Society 96:109-13. Simha, Raghu-natha (Ed., trans.). 1969. Kalhana's Raja tarangini, vol. 1. Varanasi: Hindi Pracaraka Samsthana. In Hindi. ------- (Ed., trans.). 1974. Kalhana's Rajatarangina, vol. 2. Varanasi: Hindi Pracaraka Samsthana. In Hindi. Sircar, D. C. 1960. Studies in the Geography of Ancient and Medieval India. Delhi: Motilal Banarsidass. Subramania Iyer, K. A. 1977. The Vakya padiya of Bhartyhari, Chapter II, English Translation. Delhi: Motilal Banarsidass. Tara-natha Tarka-vacaspati (Ed., comment.). 1864 (samvatsara 1921). Siddhanta-kaumudi, pararddham [second half). Calcutta: Samskrta-yantra (press of the Sanskrit College?]. 46

Page #47
-------, 1902. Sabdartha-ratnam. 3d ed. Calcutta: Daksina-carana Cakravartin at V. L. Press. Dates of the first two editions not known. Watters, Thomas. * 1905. On Yuan Chwang's Travels in India (A.D. 629-645). London: Royal Asiatic Society. First Indian ed./rpt. 1961, two vols. in one. Delhi: Munshi Ram Manohar Lal. Wilson, H. H. 1894 (Trans., annot.). The Visnu Purana. ... Edition with introduction by R. C. Hazra, 1972. Calcutta: Punthi Pustak. Yazdani, G. (Ed.). 1960. The Early History of the Deccan. Parts 1-6 in one volume. London: Oxford University Press. 47