Reviews Of Different Books

Catalog link: https://jainqq.org/explore/269337/1

Page #1
REVIEWS Hartmut Scharfe, Panini's Metalanguage (=Memoirs of the American Philosophical Society, volume 89). Philadelphia, American Philosophical Society, 1971, pp. 53. This monograph is "... intended as a grammar of the metalanguage used in Panini's grammar", in which "features in common with Sanskrit will receive a shorter presentation than those peculiar to this language (5b)". The work consists of eight chapters and two appendices. An introductory discussion on language and metalanguage (1-6a)1 is followed by seven sections dealing with the following subjects (the headings are Scharfe's): Phonology (II, 6b-12b, subdivided into: The sounds, Consonants in final position, Consonants in initial position, Phonetic changes within a compound and a sentence, Phonetic changes within the word), Inflection (III, 13a-18a), Composition (IV, 186-19b), Word formation (V, 20a-32b, subdivided into: The use of determinatives, The use of heterophones, A glance into Panini's workshop), Syntax (VI, 33a-43b), Quotation (VII, 44a-45b), Non-technical formulations (VIII, 46a-47b). The two appendices are: Katyayana and Patanjali on Panini's rule VI 1 67 (48), Is the text of the Astadhyayi younger than the word lists (gana)? (49-50). Lists of references (51a-52a) and abbreviations (52b), and an index locorum (53) complete the monograph. Panini used Sanskrit as the medium for stating rules which describe Sanskrit usage. The question arises, to what extent Panini found it necessary in formulating his grammatical rules to deviate from the Sanskrit usage for which he accounted and which he doubtless followed in his own speech. Scharfe attempts to answer this question. His major conclusion is the following. Although it is evident"... that the metalanguage has been modelled after the object language (Sanskrit)" and "even in its perfected state it has not severed all ties with the object language (56)", nevertheless this metalanguage is indeed only modelled on Sanskrit, from which it differs in important respects. I shall discuss three major parts of Scharfe's monograph: the evidence adduced to justify positing an indeterminate vowel a in the original text of Panini's grammar; the claims made regarding the sounds used as markers (it) by Panini; and the claims made concerning Panini's metalinguistic use of cases. Scharfe's treatment of these subjects is such that a fairly detailed discussion is called for. The phonology of forms in grammatical rules differs from that of normal Sanskrit in ways that both are obvious and have been noted before.2 Consonant clusters occur which are not allowed in Sanskrit, and the rules of Sanskrit morphophonemics - rules 1 This is translation of passages, not a theoretical discussion. References: Mahabhasya: Kiehorn's edition (volume, page, line); Ud(dyota): Rohatak edition; Kasika: critical edition (Hyderabad, 1967-70); Nyasa and P(ada)Manjari): Varanasi edition. 2 See Studies in Indian Grammarians, I (Philadelphia, 1969), 39b-40a. Henceforth this will be referred to by SIG-I.

Page #2
208 REVIEWS which Panini himself formulates in detail - are violated. For example, terms such as cphan (affix ayana), k-r-it which has a marker k,r,' n-n-it contain initial clusters which never occur at the beginning of Sanskrit words. k-r-it is a compound, and within a compound the normal morphophonemic replacements which occur across a pada4 boundary should apply: k-r should yield g-r.5 Similarly, n-n should yield n-n. A padafinal c should be replaced by k (cf. vak, nom. sg. of vac 'speech'). Yet the very first rule of the grammar (1.1.1: vyddhir ad aic) contains aic, with -c instead of -k. Scharfe mentions such features (9b-12b, especially 9b, 10a), but this is not new. He does, however, have something new to say in his section on sounds. In order to discuss what Scharfe says it is necessary briefly to describe Panini's use of markers, which he calls it and later Paniniyas also call anubandha. These are used, in general, essentially as are used superand subscripts in written grammars. A unit receives a special marker for the following reasons. (A) To indicate that it conditions or undergoes particular operations.or belongs to a certain class. For example, the nominal endings ne, nasi, nas (dat., abl., gen. sg.) are marked with ". Before such affixes the -i, -u of certain bases ending in these vowels are replaced by guna vowels (7.3.111: gher n-iti [gunah 108]): agnaye (Page #3
REVIEWS 209 on its first vowel regardless of the accent of the original base; audapani is the feminine of audapana 'who has come from Udapana'. The accentual difference is indicated by markers: 1, is ip, i, is in.? The nominative singular ending su is deleted after items containing i, or ig: bhavanti, audapani. In order to refer to both affixes at once with a single term, Panini marks them with n: nip, nin. He then refers to them by using the common referent ni (6.1.68: hal-ny-abbhyo dirghat su-ti-sy aprktam hal). There is another feminine affix, ig. This is high pitched, so that the derivative formed with it is oxytone; sarangt is the feminine of saranga 'spotted deer'. This affix is marked with for the same reason that nip and nin are so marked: so as to use si to refer to it and the other affixes i in rules such as 6.1.68. However, if iz were simply ni, a rule which stated that an operation applied after a unit ending in ni would now apply only where is occured. For there would now be a single unit ni distinct from nip, nin. In order to allow using ni to refer to all three affixes i, therefore, Panini marks is with a special it: nis. This s serves none of the purposes served by other markers attached to i. Its sole purpose is avoid eliminating the common term ni (samany-grahanavighatartha). (D) To prevent classing as it a final consonant which is part of a unit as used. The affix vat (as in brahmanavat 'in the manner of a Brahmana') is marked with i: vati. If the affix were introduced as vat, -t could be classed as it, hence deleted (see note 3). Panini could formulate special rules providing that the final consonants of certain units are not classed as it. He did just this for certain general classes (see note 3). In the case of vat, however, he would have been required to make a special statement for this unit, since vat is a taddhita affix and there are other taddhitas marked with t for accentual purposes (see note 9); he could not merely say that the -t of a taddhita is not classed as it. Hence, he marked vat with -i. The sole purpose of this marker is to protect the -t of vat from being classed as it and deleted (paritranartha). Markers of types (A)-(D) all serve specific purposes. In other cases units appear in grammatical statements containing appended sounds which, at first sight, do not seem to have any such purpose. According to commentators these serve two ends. (E) A sound is inserted in order to avoid doubt about what units are given in rules (asamdehartha). For example, it is claimed that the j- of the nominative plural ending jas (cf. acc. pl. sas) is a marker used to avoid confusion in 4.1.2 (sv-au-jas-am-aut ...). Ifj were not appended to as, the rule would be sv-av-as-am-aut..., with av. Hence, one could not be sure whether the second ending stated was au or ay. Similarly, the -t appended to i, u, e in 1.1.11 (id-ud-ed dvivacanam pragshyam) is supposed to have been used to avoid confusion resulting from sandhi alternations. 10 7 i and i, are affixes (3.1.1: pratyayah). An affix generally has high pitch on its first vowel (3.1.3: ady-udattas ca). However, nominal endings and affixes marked with p do not have high pitched vowels (3.1.4: anudattau sup-pitau). The first vowel of a unit containing an affix marked with n or n is high pitched (6.1.197: n-n-ity adir nityah). I have simplified for the sake of presentation, and I omit other details of accentual rules. 8 See, for example, Bhasya II.230.21-23 (regarding other affixes). 9 An affix marked with t has circumflex accent on its last vowel (6.1.185: t-it svaritam), contrary to 3.1.3 (note 7). However, a disyllabic item ending in yat has a high pitched first vowel (6.1.213: yato'navah), excepting navyd 'which can be crossed by boat'; e.g. osthya 'located at the lips'. 10 PM 4.1.2 (III.262): jaso jakaro'sminn eva sutre 'sandigdhoccaranarthah| anyatha hy aukarasyav-adese sandehah syat: kim aukarasyoccaranam athay ity asyeti. Kasika 1.1.11: ta-para-karanam asandehartham. It is probable that this is not wholly acceptable. A sound i denotes, in Panini's system, only itself (1.1.68: svam rupam sabdasyasabda-samjna). The term it, on the other hand, denotes i-vowels of all varieties (1.1.70: ta-paras tat-kalasya). In 1.1.11, then, it etc. may be used in order to refer to vowels of diffefent accentual varieties. I omit further details, noting only that whether the t of it

Page #4
210 REVIEWS owel which occurs with a unit is said to occur only to facilitate pronunciation (uccaranartha, mukha-sukhartha). Here two types are to be distinguished. (F1) A consonant followed by a is used to refer to the consonant itself: ka denotes the sound k. This vowel occurs also in forms of the type ka-kara "the sound k'. This is simply conventional Sanskrit usage, and there is no real question of considering a a marker. (F2) Vowels occur within grammatical units as they are introduced in the grammar, and such vowels can be considered used for ease of pronunciation. However, this is not the conventional usage of type (F1). The i of sic (s-aorist affix) and the u of suc (affix in forms such as dvis 'twice') are examples. Consonantal markers of type (E) can be interpreted as serving different purposes (see note 10), but there is no doubt expressed in any commentary about these consonants being it. It is about type (F2) that there is controversy in the commentatorial literature. of vowels to be classed as it (see note 3) was lost early in the Paniniya tradition 11 The controversy concerns the following points. First, is the vowel in question a marker which actually fills one of the functions noted under (A)-(D) or is it merely a vowel used to facilitate pronunciation? If the latter, does the vowel nevertheless have to be considered an it in order that it be deleted? Consider a few examples. In varttika 4 on 3.1.44 (cleh sic) the i of sic is considered a marker which serves a grammatical purpose. The third singular aorist of man 'think' is amamsta. The ending ta is a sarvadhatuka which is not marked with p. Therefore, it is treated as marked with (1.2.4: sarvadhatukam ap-it [n-it 1]). At the derivational stage amansta, then, the presuffixal base (anga) amans12 occurs before an affix treated as marked with n. Before an affix so marked, the penultimate n of a consontant-final base is deleted if the latter is not marked with i (6.4.24: anid-itam hala upadhayah k-n-iti (na-lopah 23]); e.g. sras-ya-te 'falls' (< srans-yan-te). If sic were not marked with i, amans-ta would yield *amasta. Once sic is marked with i, however, the presuffixal base containing it is so marked and 6.4.24 does not apply to delete its penultimate n.13 In vt. 5 on 3.1.44 this reasoning is rejected. Panini formulates a rule (1.2.14: hanah sic [k-it 5]) whereby sic is treated as marked with k after the root han 'strike, kill'. This is done so that the-n of han be deleted before sic (by 6.4.37: anudattopadesa-vanati-tanoty-. adinam anunasika-lopo jhali k-n-iti) in forms such as ahata (3rd sg. med.-pass. aor. Page #5
REVIEWS 211 of forms containing this affix. It adds that the -i of tasi (affix of the future type karta 'will make, do') also is pronounced without nasalization and serves only to facilitate pronunciation: tasi-sicor id-it-karyam nastity uccaranartho niranunasika ikarah pathyate. This is of interest because the Kasika on 3.1.33 (sya-tasi lr-lutoh) says that tasi is marked with i for the reason given in vt. 4 on 3.1.44 regarding amamsta; the deletion of n in manta 'will think' 3rd sg.) is thereby prevented: id-it-karanam anunasika-lopapratisedhartham,15 The examples given bring into focus the differences of opinion regarding type (F2).16 A particularly interesting case is the affix thamu, used in forms such as ittham 'thus' (5.3.24: idamas thamuh). Unlike vati considered earlier, thamu belongs to the class of affixes called vibhakti (5.3.1: prag diso vibhaktih). The final -m of a vibhakti is not classed as it (see note 3), so that the -u of thamu does not appear, at first sight, to be of type (D). Patanjali and later Paniniyas nevertheless argue that u is indeed here used to protect the m of tham from being classed as it and deleted. Patanjali claims that, by so marking tham, Panini indicates that 1.3.4 (see note 3) does not apply in the case of vibhaktis which are taddhita: acarya-pravsttir jnapayati na vibhaktau taddhite pratisedho bhavatiti yad ayam idamas thamur iti makarasyet-samjnaparitranartham ukaram anubandham karoti (Bh. 1.262.22-23, ad 1.3.4). I do not discuss here whether this is justified. As I have noted, the second point of controversy is whether vowels or consonants used only for the sake of pronunciation are to be classed as it in order to be deleted. The Kasika on 7.1.58 clearly states that the i of sic, tasi is not to be considered an it. The argument that a sound is not classed as it because there are no operations associated with this also appears in the Mahabhasya (e.g., 1.262.24). Nagesa, on the other hand, maintains that a sound used to facilitate pronunciation is to be classed as it: uccaranarthanam apittvam asty eva (Ud. II.215). There are good reasons for taking this stand. The sounds in question are not parts of items as actually used, and this is true of its. Further, markers are deleted by a rule of the grammar (see note 3), and there is no other rule whereby one could delete sounds used for ease of pronunciation; Ud. II.215: kim ca tad-abhave [i.e., ittvabhave] uccaritasya nivsttih katham syat ... The question now is whether, once the sounds in question are classed as it, operations apply which yield wrong results. These operations concern only one marker: u. The sound a is used as a marker with consonant-final roots (e.g., asa 'be seated', dupacas 'cook') in order to account for the proper distribution of parasmaipada and atmanepada endings: as is marked with a low pitched a, so that atmanepada endings occur after it (aste etc.) when an agent is signified; pac is marked with a circumflex vowel to show that both types of endings occur with it (pacati, pacate). a occurs also with affixes (e.g., akac, see below), where it is considered to be used for ease of pronunciation. However, there are no operations associated with a-marking which could apply here. This is true also of i. This anubandha is used with roots (e.g., dhivi 'nourish') to indicate that they receive the augment n after their last 15 Jinendrabuddhi notes that there is no internal conflict between the comments on 3.1.33 and 7.1.58; the former is Jayaditya's, the latter Vamana's. Nyasa 3.1.33 (II.410): nasti virodhah bhinna-kartykatvat/ idam hi jayaditya-vacanam tat punar vamanasya. 16 An interesting case concerns the unit asun, which replaces the final sound of pums 'man, male' before sarvanamasthana endings (7.1.89: pumso'sun). The replacement can be considered marked with u so that pumas be subject to n-infixation (7.1.70, see below): puman (nom. sg.). Nyasa 7.1.89 (V.633): ukara ug-it-karyarthah. However, if pums is itself considered derived from a verb through affixation with dumasun, itself marked with u, then the u of asun is considered used only to facilitate pronunciation; PM 7.1.89 (V.633): ukara uccaranarthah ... pums-sabdasyog-ittvan num/ sa hi puno dumasun iti vyut padyate.

Page #6
212 REVIEWS vowel in all contexts (7.1.58: id-ito num dhatoh): dhiny. The only places where i can possibly be considered attached to affixes in order to allow certain operations are sic, tasi, which were discussed above. In all these cases, as also in the type (D), we can speak of a neutral use of markers with respect to true grammatical operations. Now, u is attached to affixes and this marking is associated with grammatical operations. A derivative base which contains such an affix receives' the augment n after its last vowel before endings called sarvanamasthana (7.1.70: ug-id-acam sarvanamasthane'dhatoh (num 58]). If the affix ends in at, the penultimate a of the base containing it is replaced by a before the nominative singular ending su (6.4.14: atv-as-antasya cadhatoh (sau 13, upadhayah 7]).17 For example, matup is marked with u and ends in at: goman (nom. sg.) 'rich in cows', gomantam (acc. sg.), gomata (instr. sg. without n since the ending a is not sarvanamasthana). Concerning forms such as goman, the following is to be noted. The nominative singular ending su (gomat-s) is deleted after gomat, since the base ends in a consonant (6.1.68). Nevertheless, gomat-O is eligible for the operations stated in 6.4.14 (gomat -gomat) and 7.1.70 (-gomant). For 1.1.62 (pratyaya-lope pratyaya-laksanam) provides that, when an affix is deleted (pratyaya-lope 'when there is zero replacement of an affix'), the operation which it conditions still takes effect (pratyaya-laksanam '[the operation) which has an affix as a characteristic [i.e., condition]'). There is another rule (1.1.63: na lumatangasya) which negates this; if an affix is deleted and the zero replacement is denoted by a term containing lu (e.g., luk), the operation which the affix would have conditioned on its presuffixal base (anga) does not apply. The nominative and accusative singular endings su, am are replaced by luk after a neuter base (7.1.23: sv-amor napumsakasya (luk 22]). The base gomat does not, therefore, receive the augment n when it is neuter, since the zero in gomat-O (Page #7
REVIEWS 213 are members of the krt class (3.1.93; krd atin). A unit ending in a krt affix which itself n is classed as an indeclinable (1.1.39: krn m-ej-antah savyayam 371). The ending introduced after payam, kartum is therefore deleted by 2.4.82. Consequently, 1.1.63 applies, so that the augment n cannot be introduced. In such cases we may speak of the neutral use of a marker, whether this is considered necessary to protect a final sound from being classed as it (see note 18) or for the pronunciation of an item (namul, tumun): the sound in question is classed as it in order that it be deleted, but this does not provoke any grammatical operations. Particularly good examples of the neutral use of vowel its are the L-members (abstract symbols replaced by finite verb endings and participial affixes) lat, lan, lit, lin, lut, lun, Irt, Imo, let, lot. The vowels contained in these are classed as it, but there are no operations stated in the grammar associated with these particular markings in these units. Similarly, the u of inun is neutral. This is introduced after a root (3.3.44) to derive an item such as samravina 'shouting from all sides'. However, the item ending in -in is automatically followed by an (5.4.15), as shown. Since the item in -in never occurs immediately before a sarvanamasthana ending, 7.1.70 cannot apply. 19 Consider now a different type. A sound followed by the marker u is made a term which denotes a class of sounds (1.1.69); for example, ku denotes all velar stops (k, kh, 8, gh, n). Recall rule 7.1.54, which applies in the derivation of purusanam. This should introduce the initial augment n. If the u of nut is considered an it, one might now say that nu could denote all dental stops, so that 7.1.54 would allow *purusa-tam etc. 20 However, Panini himself shows that this is not so. His rule 6.4.3 (nami) states that the final vowel of a presuffixal base is replaced by a long vowel before nam (purusanam< purusa-nam). He could not have made this statement unless the term nut of 7.1.54 denoted only the initial augment n. From the above one can justifiably say the following. The markers i (nasi, vati), u (amu) are used to distinguish homophonous units and to protect final sounds from classification as it. These markers so used are neutral with respect to grammatical operations such as affixation and augmenting. The same can be said of sounds which are considered to have been used only to facilitate pronunciation. With this background, let us now consider Scharfe's contribution. He notes the "remarkable vacillation" (sa) of vowels in units such as sic, suc and also the occasional lack of an ending in forms such as ka used to refer to sounds. He says (7a), "I have come to believe that the metalanguage had yet another sound which could only unsatisfactorily be presented (sic) in the Devanagari script: a very short vowel of an unspecified character which I write as a." Of this vowel Scharfe says (76), "The question is whether we can call this sound a phoneme: its only function is to facilitate the pronunciation. It may well be that the author was not even aware that he used it". Scharfe continues (7b): "When Panini's grammar was put down in writing, scholars were apparently uncertain as to how they should denote this sound. In many instances they did not write any vowel at all: k-r-iti (1.1.5),21 n-d-rah (5.1.3), ch-voh (6.4.19), h-m-y-anta (7.2.5), r-voh (8.2.76), n-noh (8.3.28), s-tuna (8.4.41) etc. The reason is evident - these combinations containing consonants other than stops could be spoken tinguish hon used are name can be 19 The above discussion is not intended as complete (cf. note 17). I have not, for example, gone into problems associated with the u of the affix ghinun and discussed in the Bhasya on 3.2.141. 20 Scharfe (85) considers that if the u of tuk (final augment) were considered an it, this would allow adding an augment d, n etc. instead of t alone. He does not, however, fully discuss the question. 21 I have modified Scharfe's transliteration for typographic ease. I have also substituted Arabic for Roman numerals in references to rules given by him. These modifications will apply in subsequent citations from Scharfe.

Page #8
214 REVIEWS without an auxiliary vowel; it is likely that Panini himself pronounced them that way". Later (10a), Scharfe also mentions "some unusual initial combinations", including a "very peculiar feature": an initial cluster of stops, as in cphan. I think Scharfe's thesis can be summarized correctly as follows. To permit the pronunciation of certain items in certain environments an auxiliary vowel was used to break up consonant clusters. This vowel was of indeterminate quality, and Panini himself was probably unaware he uttered it. The vowel was later represented in script by a, i, u. Nevertheless, there are clusters such as cphan, where, to judge from the written text, no auxiliary vowel was used. This last point introduces some confusion into the argument. There are additional weaknesses in Scharfe's thesis. Scharfe considers (8b-9a) that the u of an affix such as namul represents a because, "None of the roles of the determinative u seems rightly applicable ..." to it: "That would leave us with determinatives without indicative function." He nevertheless has to consider that the vowels of the L-members lat, lit etc. are anubandhas, not representatives of original a. As he says (26b), "Such an interpretation is not only excluded by the nature of /e/ and /o/ as long vowels, but also and mainly by the systematic way in which these vowels are used to differentiate the technical names for tenses and moods." Since there are no particular grammatical operations associated with the anubandhas a, i, u etc. of these affixes, Scharfe has here accepted the neutral use of vocalic markers. This means his argument concerning the u of namul is not valid. Consider also the following. According to the thesis as presented, the auxiliary vowel a - which Scharfe claims Panini was possibly not aware of pronouncing - should occur only in very particular phonologic environments. Although Scharfe has not discussed the final vowels of units such as vati, I take it for granted he could not consider the -i of vati a representation of a. The genitive singular of vati is vateh (5.1.18), a regular i-stem genitive. It would be absurd to claim that Panini was not aware of the relation between the nominative vatih (5.1.115) and the genitive vateh. Now Scharfe (9a) considers the u of inun to represent original a. The ablative of inun is inunah (5.4.15), a regular consonant-stem ablative-genitive of the type -duhah. According to Scharfe's thesis, one would expect instead a form *innah, since this ablative occurs in a phonologic environment which does not require the auxiliary vowel @. To keep this thesis intact one would now have to argue that scribes carried the u of inun over to the ablative inuna). But this is begging the issue. The major problem lies in the claim that a was a totally indistinct vowel and that Panini probably was unaware he inserted it. I consider Scharfe's suggestion that there was such a mysterious vowel a recited in the text of Panini's grammar to be unproved. Scharfe has taken the path of least resistance with regard to the materials by simply positing a new entity. A reasonably close look at these materials raises serious doubts about the validity of this proposal and the method by which Scharfe arrived at it. I claim no finality for the suggestions I have made above; a review is hardly the place to set forth a detailed discussion of all the evidence. I do think that the answer lies in the direction indicated. Scharfe deals also with the organization of the sound list which preceeds Panini's grammar (the siva-sutras) and about the sounds used therein as anubandhas (20a, 286-30b). After listing the traditional catalog of stops in five five-membered groups (varga), Scharfe says (20a), "Panini adjusted this purely phonetic scheme to the needs of his grammar; instead of reading the panca-vargas horizontally and from left to right, he read them vertically from top to bottom, starting at the right side." Concerning the sounds used to close sets of sounds within the siva-sutras, Scharfe (following Breloer, see reference on p. 28b, note 45) accepts the following reasoning (29a). Many sounds were not used because of their susceptibility to contextual sandhi modifications: "He [Panini] found, that many sounds are more or less disqualified due to their changes in sandhi: aspirate and voiced stops could not, in final position, be told from non

Page #9
REVIEWS 215 ; cf. vit set avis). Yet Panin halam jas iho aspirate and unvoiced stops; // could not stand in any final position; vowels are too much subject to mergers in a sentence." The number of available sounds was further reduced, continues Scharfe, because t was chosen to form terms like at, at (denoting all short and long a-vowels respectively). Moreover (29a), "/n/ and /s/ are in Sanskrit subject to many sandhi changes". This marks a definite retrogression in our knowledge of Panini. The inquisitive scholar will ask why Panini had to "read" the traditional sound catalog in the upsidedown fashion noted. Scharfe does not enlighten us on this, preferring to refer to antiquated studies which simply said what he repeats. A reason for Panini's reorganizing the traditional sound catalog has been proposed (see SIG-I, 96), and the weaknesses of the suggestion which Scharfe accepts unquestioningly have been demonstrated (SIG-I, $5.1). Usual procedures of scientific discussion would require that Scharfe demonstrate in detail how the recent proposals are unacceptable and how the proposals he accepts avoid pitfalls encountered by others. This he does not do.22 He states baldly what he considers a truth. The arguments given concerning the sounds available to Panini for use as markers are also far from satisfactory, as has been demonstrated before.23 If Panini had avoided using certain sounds as anubandhas merely because of their contextual replacements, we should not expect him to use s or s; cf. vit 'settlement (nom. sg. of vis), dvit 'who hates' (nom. sg. of dvis). Yet Panini did use these sounds as final markers; e.g., 8.4.53: jhalam jas jhasi, 8.2.37: ekaco baso bhas jhas-antasya s-dhvoh. It is true that h cannot stand in final position in Sanskrit. It is also true that pada-final c is replaced by k. Yet Panini used cas a final marker in the siva-sutras and elsewhere. If Panini's using t as a special symbol elsewhere in the grammar is the basis for his not using this sound as a marker in the siva-sutras, it is remarkable that k, cand others are used both in the siva-sutras and elsewhere as markers. In sum, I think the approach adopted by Scharfe is totally unacceptable. Scharfe has accepted this approach without question. But he has failed to demonstrate that he was justified in doing so. The section on syntax is, in my opinion, the most important part of this monograph. It is in syntax that, according to Scharfe, Panini has most significantly departed from normal Sanskrit usage. The section begins (33a): "Panini has redefined the meanings of three cases, making them the basis of his syntax: the genitive (1.1.49), the locative (1.1.66) and the ablative (1.1.67). The nominative and the rare instrumental have values identical with those of the object language and, therefore, need no new definition. The meanings of the genitive, ablative and locative, though derived in some way from those these cases have in the object language, are so special and so technical, that new definitions were necessary." "The role of the genitive is defined" (33a) by 1.1.49 (sasthi sthane-yoga), which Scharfe translates (33a), "A genitive is to be construed with a [supplied) sthane ('instead [of ...]', 'in place (of ...]')". (brackets and parentheses Scharfe's). He goes on to note (33b), "the technical use of the genitive is skillfully derived from the object language". Scharfe then tells us (33b), "The new value of the ablative is established by Pan 1.1.67 tasmad ity uttarasya 'When something is enounced in the ablative, it means (i.e. the procedure refers to) that which follows'. The technical locative has the opposite meaning: Pan 1.1.66 tasminn iti nirdiste purvasya 'When something is enounced in the locative, it means (i.e. the pro 22 Scharfe does, however, refer to SIG-I in note 5 on page 20. Note that the explanation given by Scharfe (29b-30a) concerning the sounds used as final markers in the siva-sutras is remarkably similar - indeed almost identical - with the one given in SIG-I, 40-41, although Scharfe does not refer to this. 23 See SIG-I, $$ 7-7.1. I hasten to add that I apologize to the spirit of Breloer for having missed one of his articles on the question discussed. However, I must also add that Breloer's arguments do not, I think, differ essentially from those considered in SIG-I, although Scharfe uncritically accepts Breloer's position without question.

Page #10
216 REVIEWS cedure refers to) that which precedes'. It is mainly in the metalinguistic treatment of suffixes and augments, that these technical cases are important: a suffix A is added after a root X, an accretion B is added to a root Z before a suffix Y."24 Scharfe informs us (33b, note 11) that these translations are in accord with what is said in the Kasika,25 but that he himself has a different interpretation of these rules. Scharfe approaches this second interpretation of 1.1.66, 67 circuitously. On page 36b, he translates a small part of a Mahabhasya passage in which a particular view is espoused. The Bhasya discussion concerns rules 3.1.68 (kartari sap (sarvadhatuke 67]), 3.1.69 (div-adibhyah syan), and subsequent rules which introduce post-verbal affixes in the derivation of presential forms. 3.1.68 provides that sap is introduced after a root26 when this is followed by a sarvadhatuka ending which has been introduced to denote an agent. 3.1.69 introduces syan under the same condition after roots of the group div etc. The proposal is made27 that syan and other affixes introduced by subsequent rules be made replacements of sap.28 3.1.69 should then state: div-adibhyah sapah syan ('After diy etc., syan in the place of sap.'), with the substituend genitive (1.1.49) sapah. This genitive is obtained by letting the nominative sap of 3.1.68 carry into 3.1.69. In the latter rule, 1.1.67 applies to interpret the ablative div-adibhyah: 1.1.67 serves to convert the nominative sap to the genitive sapah. Under this interpretation, then, 1.1.66, 67 serve to convert nongenitive forms to genitives. This entails reading 1.1.66 tasminn iti nirdiste purvasya sasthi, 1.1.67 tasmad ity uttarasya sasthi, with sasthi understood as recurring from 1.1.49. Scharfe cites the Bhasya passage noted and approves of the proposal made therein: "This rule (tasmad ity uttarasya) states, that an element in the ablative determines the following element as the substituendum: properly marked by a genitive ending. This model changes the nominative sap to a genitive *sapah (36b)." Indeed, he takes pride in ostensibly following Patanjali and somewhat haughtily disagrees with another view (36b, note 24): "It is evident that 1, in following Patanjali, do not agree here with. ..." However, a bit later (38) Scharfe modifies his view. He notes that interpreting 1.1.66, 67 as containing sasthi carried over from 1.1.49 is difficult : "sasthi would continue not in direct line from 1.1.49 to 1.1.66, 67 - it would have to jump over the rules 1.1.50-65, in which it has no place. Besides, it creates difficulties for the suffix rules, which can be disposed of only with highly sophisticated argumentations (38b)." Scharfe then considers the Paniniya view, namely that in 1.1.66, 67 one is to understand karyam 24 Emphasis Scharfe's. I have substituted Italics for the boldface of the original. 25 The Kasika glosses 1.1.66 as follows: tasminn iti saptamy-artha-nirdese purvasya karyam bhavati nottarasya. According to the Kasika, then, tasminn iti refers to what is denoted (artha) by any locative form (tasmin, loc. sg. of tad used as a variable). Further, the Kasika emphasises that an operation (karya) applies to what precedes that which is denoted by a locative, not what follows it. This is in accord with what is said in the Bhasya. I do not think Scharfe has fully grasped what the Kasika is driving at. Certainly Scharfe's subsequent suggestions (see below) show a disregard for the subtleties of what the Paniniyas say. He has also not understood the import of Panini's using iti in 1.1.66, 67, as is clear from Scharfe's discussion of quotation (44a-45b, especially 45a). I reserve discussion of this for another occasion. 26 3.1.68, 69 come under the heading of 3.1.1. (see note 7), so that sap, syan are affixes. By virtue of the heading 3.1.2 (paras ca), all affixes occur after units unless a particular provision to the contrary is made (e.g., akac, see below). 27 Bhasya I.86.20 (and elsewhere): sab-adesah syann-adayah karisyante. Scharfe cites four lines beginning immediately after this. He also does not consider the full context in which this claim is made. This is discussed in some detail in the paper, 'On Panini's metalinguistic use of cases", appearing in the Charu Deva Shastri Felicitation Volume. 28 The finally acceptable view is that syan blocks the introduction of sap: 3.1.69 is an exception (apavada) to 3.1.68.

Page #11
REVIEWS 217 'operation'. He says (38b), "karyam offers no such problems (as does understanding sasthi to recur from 1.1.49), but it is arbitrary: it does not occur in any preceding rule, from where it could be taken to 'continue'. To me a metalinguistic genitive after 1.1.49 looks best for 1.1.66, 67, but this is not the place to go into so complex a problem." (bracketed suppletion mine, G.C.) In note 40 of page 38b, Scharfe suggests that 1.1.67 applies correctly in affixation rules, so that an affix is introduced as a replacement of zero. For example, 3.1.5 (gup-tij-kidbhyah san), which introduces san to form derived roots such as jugupsa 'shrink from', is translated by Scharfe as follows: "The suffix san (is added) after the [roots) ... [instead of what originally followed] (i.e. nothing)." He finds this "quite unobjectionable". If I understand Scharfe correctly, his conclusions are the following. 1.1.49, 66, 67 define metalinguistic values of the genitive, locative, and ablative which, though they derive in some unspecified way from normal Sanskrit usage, are totally distinct from the values of these forms in the object language. purvasya, uttarasya of 1.1.66, 67 are substituend genitives; Scharfe considers this the best interpretation although he shrinks from discussing "so complex a problem". Finally, 1.1.67 applies in affixation rules: these affixes are introduced as replacements of zero. A further conclusion reached by Scharfe is this: although Panini does indeed use the case forms mentioned with the values they have in Sanskrit, this usage is not correct with terms proper to the metalanguage. Scharfe says (34a), "Such non-technical employment of the cases should not occur with technical signs; in fact, they are extremely rare: the locative matau 'in the meaning (of the suffix] -mat occurs twice. ..." In addition (35a), "When technical terms and works taken from Sanskrit are joined in one sutra, we have a mixed syntax: cases following Panini's metarules and cases following Sanskrit practice. ..." In the case of nouns borrowed from Sanskrit and given technical meanings, "they follow in their inflected forms freely the syntactic rules of the object language or the metalanguage (35a)". I have noted that Scharfe has to admit of the nontechnical use of the locative in cases such as matau. His conclusions cause him other problems. I shall note a few of these with Scharfe's suggestions. 7.1.88 (bhasya ter lopah (pathi-mathy-rbhuksam 851) provides that the ti29 of the items pathin 'path', mathin "churning stick', rbhuksin (a name) is replaced by zero (lopa) if these units are bha;30 for example, patha (instr.sg.) Page #12
218 REVIEWS "Two explanations are possible for these two locatives: yusmady upapade and asmady [upapade) might be taken as imitations of the locativus absolutus of the object language ('when yusmad resp. asmad is an upapada), or the locatives denoting an upapada are extended incorrectly from that special metalanguage of the upapada section (3.1.92-3.4.117), where the locative has this special meaning ... - but then the word upapada would look rather superfluous." The whole of a later section (VIII, 46-47) is devoted to nontechnical formulations. I choose one example from this section, concerning which Scharfe says (46a), "It is more difficult to find the reason for the formulation of...." The rule in question is 5.3.71: avyaya-sarvanamnam akac prak teh "(The element) akac (, classed as affix, is added to) indeclinables and pronominals (, and it occurs) before the ti (of these units, not after them)' see notes 26, 29. Scharfe's comment on this rule is (46a): "Would *avyayasarvanamnam akac tau"/ak/ is added before the last vowel of indeclinables and pronouns' be materially different? To me it appears more as a question of style, which stresses Jak/ as an exception to /ka/ taught in a previous rule, occurring only in a limited field. The statement that this /ak/ is not suffixed but infixed before the last vowel, comes only as a corrective, an afterthought." These then are Scharfe's conclusions regarding Panini's use of the metalinguistic syntax he is supposed to have invented. According to Scharfe, Panini sometimes violates the style and spirit of this metalinguistic syntax and makes statements which include mere afterthoughts, statements which could as well be expressed otherwise in terms of his own metalanguage. At the same time Scharfe accepts (47b) L. Bloomfield's opinion that Panini's grammar is one of the greatest monuments of human intelligence. I find these two views hard to reconcile. I submit that Bloomfield was correct and that Panini did not commit the blunders Scharfe finds it necessary to attribute to him. Let us begin the evaluation of Scharfe's claims with a brief outline of some of the Sanskrit case uses for which Panini accounts.32 One use of the genitive is relevant in the present context. A genitive ending is introduced after a nominal base when there is to be denoted a relation which holds between what that base denotes and something else other than an action (2.3.50: sasthi sese); e.g., putrasya pita 'father of the son', vrksasya sakha branch of a tree'. For the locative two uses are to be noted. A locative ending is introduced after a nominal base if a locus is to be denoted (2.3.36: saptamy adhikarane ca); e.g., grame vasati 'is staying in the village'. Such an ending also occurs after a base if what is denoted by the latter is involved in an action which characterizes another action (2.3.37: yasya ca bhavena bhava-laksanam); e.g., brahmanesv adhiyanesv agatah 'He arrived while the Brahmanas were studying'. For the use of the ablative only one rule need be considered. An ablative ending is introduced after a nominal base construed with certain items, among them words which denote directions (2.3.29: ... dik-sabda ... yukte (pancami 28]); e.g., purvo gramat parvatah 'The mountain is east of the village'. Sanskrit is the language used by Panini in his grammar. The student of this grammar is assumed to know Sanskrit, to have a full knowledge of the uses accounted for by the rules noted. Panini does not have to describe to a student of his treatise the syntactic patterns the student knows. Now consider the following sentence types and examples. (a) tasya, tasya, ... tat (b) tasmin tat (C) tasmat ... tat where, however, kumbha + acc. should of course be kumbha+gen. and *kumbhan karah should be *kumbhanam karah. The reader is asked to correct these inadvertent errors. 32 The following is a brief exposition of what is dealt with fairly thoroughly in the paper alluded to in note 27. Full textual evidence is given in this paper, so that I do not think it necessary to give citations here.

Page #13
REVIEWS 219 (al) mama pituh samipe devadattah 'Devadatta is near my father'. (a2) mama pituh sthane'ham atra 'I am here in place of my father'. (b1) kupe devadattah 'D. is at the well'. (62) agate devadatte gato'ham 'I left when D. arrived'. (cl) gramat purvah parvatah (c2) gramad uttarah parvatah "The mountain is north of the village'. Note that two genitives occur in (a1), (a2): mama 'my', pituh 'of father'. Only the latter is construed with an item which fills the blank in (a). Let us call forms such as mama, pituh, which are respective values of tasya, and tasya, in (a), bound and unbound genitives respectively. Note also that (b1), when it has the meaning shown (and not ... is in the well'), does not convey where precisely Devadatta is relative to the well, whether to the east, west, north, or south of it. Consider now the following grammatical statements. (1) 1.4.80: te prag dhatoh (2) 1.4.105: yusmady upapade samanadhikarane sthaniny api madhyamah (3) 6.4.34: sasa [upadhaya 24] id an-haloh (4) 6.1.77: iko yan aci (5) 3.1.91: dhatoh (6) 8.1.28: tin atinah (1) is immediately understandable from one's knowledge of Sanskrit: they (the items classed as upasarga and gati by previous rules) occur before (prak) a root (dhatu). The ablative dhatoh is construed with prak in accord with the normal syntactic pattern accounted for by 2.3.29. (2) also is immediately understandable: the verb endings called madhyama occur when the L-member which has been replaced tentatively by any finite ending denotes the same thing (samanadhikarana) as a potentially used (sthaniny api) cooccurring item (upapada) yusmad. The other sentences are not immediately susceptible of a unique interpretation: they are not really full sentences. (3) is like (a1), (a2) in that it contains both a bound and an unbound genitive. The relation signified by the genitive ending of sasah is obvious: a penultimate sound (upadha) is part of a unit. However, (3)6.4.34 does not immediately tell the reader what relation holds between this penultimate sound of sas ('instruct') and i (denoted by it). (4) is similarly incomplete. The rule does not directly state what relation holds between the vowels i, u, r, ! (denoted by ik, gen. ikah) and their corresponding semivowels (denoted by yan). In both cases one could, from one's knowledge of Sanskrit, supply samipe, sthane as in (a1), (a2) or some other relational item in order to complete the sentences. There is an additional obscurity. The locative aci is immediately interpretable as denoting loci. It is also true that the only locatival relation which obtains between linguistic items is contiguity. However, one does not know from (4) whether the operation stated applies before or after contiguous vowels (ac). Rule (6) requires a suppletion in order to be fully understood. One could, then, understand either atinah purvas tin a form terminating in a verb ending (tin) and which preceedes a form terminating in a nonverbal ending (has no high pitched vowel)' or atinah paras tin 'a form terminating in a verb ending and which follows ...'; cf. (cl), (c2). Similarly, one does not directly understand from (5) whether the units introduced by subsequent rules are to occur after a root or before a root. Clearly, if a Sanskrit speaker were left to understand such rules without further rules to guide him in interpreting them, he would be at a loss. Panini supplies such rules. (5)3.1.91 is stated in a section of the grammar headed by 3.1.1. Within this section another heading is in force, 3.1.2 (see note 26): a unit introduced and classed as affix (pratyaya) occurs after (para) that to which it is introduced. The ablative dhatoh is thus construed with the direction word para according to normal Sanskrit syntax. A rule such as 3.1.97 (aco yat) together with the headings and 1.1.72 (yena vidhis tad OL.

Page #14
220 REVIEWS antasya) states that yat, classed as affix, is introduced after a vowel-final root (e.g., jeya 'to be conquered' Page #15
REVIEWS 221 ore only as a locative about pure and simple. nnot here have it t madhyama, uttama) are allowed to occur only in given contexts. There is no question of contiguity between yusmad and endings. Nor is there any question of a particular domain (visaya) in which the restriction applies: it applies in Sanskrit pure and simple. Hence, the locative yusmadi is interpretable only as a locative absolute (sat-saptami). Further, upapada cannot here have its technical meaning; it means simply a cooccurring item. As for 5.3.71, the wording is not merely a stylistic variant, as Scharfe claims. This rule introduces an affix (pratyaya) ak. Normally, an affix occurs after the unit to which it is introduced (see note 26). Hence, Panini had to formulate this rule as he did in order to have ak occur within indeclinables and pronominals; e.g., ucsakaih, not *uccaisak. Moreover, a bit of thought shows that a wording avyaya-sarvanamnam akac lau is absurd. A ti is a part of a unit. How then can one speak of a whole unit occurring in contiguity with part of itself? Finally, let us consider what Scharfe has to say about applying 1.1.67 in affixation rules. Not only is this not unobjectionable, as Scharfe would have us believe, it is absolutely undesirable. Once 1.1.67 is allowed to apply in affixation rules, 3.1.2 (see note 26) becomes superfluous: why state a particular heading whereby a unit is introduced after another unit if the same results can be gotten by applying a general metarule? Moreover, once one considers an affix a replacement of zero, as Scharfe suggests, a veritable Pandora's box of problems is opened. Nothing is obviously not a sound. Now, for reasons not considered here, Panini has to state a rule whereby a substitute (adesa) is treated as having the value of its substituend (sthanivat) except with respect to operations which would then be conditioned by an original sound (1.1.56: sthanivad adeso'nal-vidhau). Consider the derivation of divyati 'gambles'. The ending ti is allowed to occur after the root div (div-ti), which is then followed by syan (div-ya-ti); then the -j- of div is replaced by i; div-ya-ti. If, now, ya has replaced zero, div-ya-ti can be treated as though it were div-ti. ti is a sarvadhatuka affix and the base to which it has been introduced has a short penultimate vowel. Therefore, the rule (7.3.86: pug-antalaghupadhasya ca (sarvadhatukardhadhatukayoh 84, gunah 82]) can apply whereby i, u, r, I of a presuffixal base with a short penultimate vowel is replaced by guna before a sarvadhatuka or ardhadhatuka affix: div-ti - dev-ti. In addition, v is deleted before a consonant such as t (6.1.66: lopo v-yor vali): dev-ti-de-ti. This and other equally undesirable results cannot be avoided, except through extremely tricky arguments which are ultimately unacceptable, once one takes the step Scharfe has taken. I am, it is obvious, convinced that Scharfe's thesis regarding Panini's syntax is unacceptable and that Scharfe has not thought through the problems involved. His proposal involves so many problems, some of which he does not mention or consider in any depth, that it must be rejected. The Paniniya point of view, on the contrary, involves none of these problems. The Paniniya interpretation also does not assume, as Scharfe assumes erroneously, that the syntax of grammatical statements is radically different from that of the Sanskrit which Panini describes. I have considered three major parts of Scharfe's monograph. I have done this in order to present Scharfe's views in detail and as faithfully as possible and also to discuss questions which Scharfe has omitted although he should have taken them into consideration. There are other parts of Scharfe's monograph open to serious doubts. However, the sections dealt with in this review are sufficient both to present Scharfe's attitudes and conclusions and to justify my opinion of his work. Scharfe's conclusions are interesting and challenging. They are also based on an insufficient consideration of materials which call for a great deal more depth and subtlety than is demonstrated by Scharfe. It is to be hoped that in his next work on Paninian grammar Scharfe will be less amenable to allying himself with an ekadesin and that he will make far less use of the sthali-pulaka-nyaya. George Cardona

Page #16
222 REVIEWS A. D. H. Bivar, Catalogue of the Western Asiatic Seals in the British Museum. Stamp Seals II: The Sassanian Dynasty. London 1969. 145 S Mit diesem Buch hat B. eines der grossen Desideraten, die Publikation der bedeutenden Sammlung sasanidischer Siegel des British Museum erfullt und daruber hinaus wertvolle und anderweitig nicht leicht erhaltliche Informationen zum Gegenstand gegeben. In der Einleitung gibt B. namlich einen ausgezeichneten allgemeinen Uberblick uber die Studien sasanidischer Siegel, dies in wissenschaftsgeschichtlich sinnvoller Verbindung mit der Geschichte der Londoner Sammlung. Dann spricht er uber die Klassifikation, chronologische und palaographische Fragen, uber die Siegelformen, die ikonographischen Inhalte und uber literarische Nachrichten uber den Gebrauch der Siegel. Ein kurzer, aber sehr nutzlicher Part uber die mineralogische Seite und eine gute Bibliographie schliessen sich an. Der Katalog umfasst rund 800 Siegelsteine, gibt eine sorgfaltige Beschreibung mit allen notigen Grunddaten uber Steingattung (erstmals mineralogisch genau untersucht), Abmessungen und Literaturhinweise, dazu kurze Bildbeschreibung und die Transliteration der Legenden. Die 32 Tafeln sind von bester Qualitat fur den, der um die Schwierigkeiten gerade bei sasanidischen Siegeln weiss. B. fugt einen Index der Inschriften an, eine Tafel der Tamgas, die als Beizeichen figurieren, Konkordanzen zwischen seinem Katalog und den Inventarnummern des Museums, ferner einen Gesamtindex fur das Buch. Vier Zeichentafeln geben die Hauptformen der Siegel. Das Buch hat schon von seiner Anlage und seiner genauen Ausfuhrung her alles mitbekommen, was es unentbehrlich macht. Wenn verschiedene Dinge der Kritik offen sind, so ist das in erster Linie Schuld der Wissenschaftslage, kaum des Autors. Dass die Klassifizierung angreifbar ist, sieht jedermann; B. stutz sich auf die im British Museum seit Thomas (1852) ubliche. Da ich gleichzeitig an einem neuen naturlich rekonstruktiven Schema gearbeitet habe, das jetzt erschienen ist (Der sasanidische Siegelkanon, Braunschweig 1973) kann ich als befangen gelten, im ubrigen entfernt sich mein Schema nicht in allen Punkten, stimmt sogar oft mit jenem von B. uberein, versucht es vielmehr vom Inhalt her nach der Regelmassigkeit der Bildkomposition zu Grossbuchstaben und eine Nummer (z. B. BL 6) sind nicht unbedingt praktisch. Sehr wertvoll sind die Datierungen. Sie zeigen, dass bei allen mit den sasanidischen Siegeln Befassten im allgemeinen doch ein gutes gleiches Zeitgefuhl herrscht. Da genaue Angaben aus Grabungen fehlen, wird nur die Bildanalyse eines Tages weiterhelfen, wie ich in meinem eigenen Buch gezeigt habe. Die Aufteilung der Bilder auf die Tafeln ist ubersichtlich und splendid. Was stort, ist der "unendliche Rapport" der Anordnung, in der die verschiedenen Gruppen fast fugenlos ineinander ubergehen. Kleinliche Ausstellungen sind hier nicht am Platz, einige Anmerkungen, die vielleicht nutzlich sind, seien aber auszugsweise notiert. Tf.4, BA 7: ein Gryllus, hat nichts mit Hephthaliten zu tun!; Tf.11, DJ 5 muss um 90deg nach links gedreht werden: Tierkampf, wobei der Lowe uber dem gefallenen Buckelochsen steht. Tf.11, DM 3 ist kein Damonenkopf, sondern nach Parallelen mit Ubergangsstil ein frontaler Lowenkopf. Tf.13, EG 5 ist ein Flugellowe, kein Greif; EG 14: Flugelpferd, kein Greif; Tf.16, EO 3: Flugelzebu, kein einfacher Stier; Tf.20 GC 1 und 2: Eberkopf (nicht Barenkopf);. Von den unsicheren Tieren der Gruppe GI sind vermutlich: 1 Tapir, 2 Esel, 3 Wolf, 4 Bar?, 5 Flugellowe, 6 Lowe, 7 Leopard, 8 Lowe, 9 Flugellowe, 10 Zebu, 11 Pferd mit (Kunst-?)reiter; Tf.22, HE 2: Rabe, HE 4: Ente; Tf.23, HG 6: romisch, 3. Jh.: Adler mit gespreizten Flugeln, moglicherweise noch junger (byzantinisch?), aber nicht modern! Tf. 24, JA 3: Vogel uber Beute; muss um 180deg gedreht gesehen werden; Tf.25, KC 1: Flugelzebu mit Fischschwanz (nicht: "locust"); LA 13: keine Pflanze, sondern Blute auf Flugeln mit Tierkopfen

Page #17
REVIEWS 223 (links: Gazelle; rechts: Kranich o.a.); Tf. 27, NA 2: Altar zwischen zwei Kreuzen, NB 4 ist um 180deg zu drehen, desgl. NB 6, beide wegen des ublichen Tamga-Aufbaues; ND 1: um 180deg drehen! (Binden stehen aufwarts! Zur Bibliographie S. 38: Hans Henning sind Vornamen, von der Osten ist der Hauptname; ferner: Janichen (nicht Jannichen). Bilddeutungsdifferenzen sind besonders bei dem oft minutiosen, aber stark vereinfachenden Schnitt provinzieller Arbeit fast Selbstverstandlichkeiten, gegen die nur die Aufstellung langer Degenerationsreihen von bester hofischer Arbeit bis zum provinziellen Schnitt hilft. Auch das wird sich eines Tages einpendeln. Bestimmte Typen werden aber dennoch immer unklar bleiben. In den Lesungen ist B. genau und vorsichtig, er macht keine Versuche, einen Befund zu pressen oder in eine bestimmte Richtung zu dirigieren. Uber Falschungen sind wir nicht immer einer Meinung. So halte ich Tf.3 AE 5, 6, 7 fur sichere Falschungen, desgleichen BH 1, aber das verschlagt nichts. Das Buch ist ein ausserst wertvoller Beitrag zum Thema und tragt, wie ich inzwischen mit Freude sehe, als Anstoss zur Katalogisierung anderer Bestande beispielhaft bei. Auch den Sammlern wird es sehr willkommen sein. Auch wenn die erkennbare Dynamik des Themas und der Elan des Autors durch die britische Kuhle sichtlich gebremst werden, ist die Wissenschaft B. in jedem Fall zu grossem Dank verpflichtet. Vienna Robert Gobl J. Gonda, The Vedic God Mitra (= Orientalia Rheno-Traiectina XIII). Leiden. Brill, 1972. VIII + 147 pp. 1. For studying a single god of the Vedic pantheon various ways are open. In the case of Mitra the problem is in part lexicographical (where do the texts refer to god Mitra, where is mitra, m.n. an appellative noun and what exactly does this mean?) and in part mythological. In this new book of the distinguished Nestor of Vedic studies the lexicographical aspect has to a large extent been eliminated as this will be dealt with in a separate paper "Mitra and mitra". As for the mythological aspect at least two basically different approaches are possible. One can either start from the supposition that mythology in general has a systematic character, or study and interpret the data without any initial working hypothesis, collecting all textual evidence and building up, step by step, a coherent picture of the god. In both cases text-interpretation is (or should be the basis on which all conclusions must be founded. In both cases, too, one has to face the fundamental difficulty that there does not exist such a thing as "objective" text-interpretation. The mass of textual data has to be grouped and to be put into some order, which, again, is likely to reflect upon the interpretation of individual passages. The notion of system was introduced, as far as I am aware, by cultural anthropologists, and monographs such as, e.g., Scharer's book on the Dayak religion have shown how a religion can be described as a coherent system of notions. The semiotic analysis of texts, as applied to the Rigveda by, e.g., B. L. Ogibenin, Struktura mifologiceskix tekstov "Rigvedy" (Vedijskaja kosmogonija), Moscow, 1968, also operates with binary oppositions and the construction of a structural model of Vedic mythology, and leads to results which do not appear to be appreciably different from those of the structural approach in general. In case one rejects the idea of a mythological system as a startingpoint for the interpretation of individual gods, the coherence between the various utterances of the texts has to be constructed in a more intuitive way. In either case the risk of misinterpretation, inherent in any attempt of modern man to understand how the Indian theologians of the second and first millennia B.C. conceived their world,

Page #18
224 REVIEWS cannot be ignored. The history of Vedic Studies in the Western world, indeed, produces eloquent testimony to it. Gonda's book, which has grown out of a paper read on a congress of Mithraic studies, keeps in general the second course, although he fully recognizes the opposition between Varuna and Mitra. His general approach may be considered well-known from his cuvre. As he is wary of general working hypotheses, the emphasis in this work is on a complete collection of the references to Mitra culled from all Vedic texts. His conclusion can be summarized in the following quotation (p. 106): "After reconsidering all relevant Vedic - that is, not only the Rgvedic - texts I have arrived at the conclusion that the assumption of the mere meaning "Contrat (contract)" and the conclusion that Mitra essentially is the god in charge of contracts are nowhere self-evident or a necessity. At the risk of undue repetition I add that I am under the impression that there is no place where this translation is, by exclusion of all other interpretations, the only possibility. The texts moreover do not to my knowledge furnish us with a clue to the relations, historical or other, between a basic "contract" and the variety of the god's functions and activities". He further states (p. 109) 'that the Mitra of the Vedic texts - who "does not possess any individuality on the physical side" - rather is the god who, while maintaining the sta ... puts things right, regulates the contacts between men and between men and the divine powers, and exhibits benevolence and active interest. Whereas Varuna, the representative of the static aspects of kingship, is a guardian of that sta, his companion and complement Mitra, being no less concerned with it and no less its promoter, is rather its maintainer, the one who keeps its manifestations in the right condition, who redresses if something has gone wrong, who adjusts, restores, appeases, stabilizes, the god also who unites men'. Cf. further p. 112: "As far as the Veda is concerned there is a god Mitra and an appellative mitram which expresses the main idea the god stands for, viz. the maintenance, without wrath or vengeance, of right, orderly relations, manifestations of which were, first and foremost, the active benevolence and willingness to help and redress"; p. 113: "This essentially beneficent and benevolent power, energy and all-pervading essence makes its existence and influence also felt in a considerable variety of natural phenomena". Gonda, accordingly, rejects a theory which is nowadays widely accepted, viz. that Mitra is the personification of a social notion "contract", which is supposed to have been the primary meaning of the word mitram. For this meaning he substitutes "maintenance of orderly relations", "the active benevolence and willingness to help", etc., which actions and properties are considered to be functions of the god, rather than social phenomena. Every worker in the field of Vedic (and Old Iranian) religion will no doubt warmly welcome the appearance of this almost exhaustive and up-to-date collection of the relevant Vedic material. To those scholars who have up to now followed Meillet's lead, this book is a serious challenge. The circumstance that Gonda emphatically states (p. 98) that he has "found no Vedic texts which should or might put us on the scent of such a contract in connection with the name of the god Mitra" imposes on every scholar who endorses Meillet's view the obligation to reconsider the whole Vedic evidence in the light of the interpretation that Gonda here offers of it. 2. The very exhaustiveness of this collection, however, has also caused some inconveniences. As is well known (and stressed by Gonda, p.49ff.), it is in many Rigvedic passages hard to decide whether the poet means the appellative noun mitram or the name of the god Mitra. Besides, an exact determination of the meaning and the usage of mitram is of direct importance for the interpretation of the god. A discussion of all this material, however, has been reserved for a separate paper "Mitra and mitra", to which the author frequently refers. Considerations of space have, no doubt, made it necessary to split up this study into two publications. For the reader, however, this

Page #19
REVIEWS 225 separation has the inconvenience that he is often compelled to postpone his judgment until the justification of Gonda's view has been published. A second difficulty with which Gonda had to cope was the fact that "for reasons of space and in spite of the drawbacks of monographs dealing with one single member of the pantheon" (Preface, p. VIII) the character of Varuna could not systematically be discussed. This limitation must have been the more painful to him since he rightly states "that the pair of gods represents in a complementary way the two-sided aspect of the idea they stand for" (p. 40, cf. p. 15). In other words, it is in this case not even a single god but only one aspect of a dual deity that could be described in these 138 pages. Varuna is only mentioned "as far as his relations to Mitra may require" (Preface, p. VIII). In some cases, however, the complementary character of their relation may not explicitly have been stated in the text but nevertheless throw an instructive light on Mitra (see below). The limitation makes itself further felt in the fact that, although Varuna and Mitra are the most important members of the group of Adityas, this characteristic group of the Devas is but seldom mentioned in this monograph (p. 129). In this connection it may be remarked that Mitra's relation to the western quarter in ASPaipp. XVII.41.7 praticyai dise, varunaya 'dhipataye, prdakave raksitre, mitraye 'sumata etam pari dadhmah (quoted p. 83) is in line with the Vedic system of classification in which the Adityas as a group are constantly associated with the western quarter. The Saunaka reading (a)nnaye 'sumata (XII. 3.57) need not represent the original version. Also, the wider context of the whole pantheon might finally have been of some importance for the interpretation of such passages as MS. 1.8.6 (123,10f.), KS. VI.7 (57,9f.): yat sadhumam jyotis tad vaisvadevar, yal lohitam tad varunar, yat suvarnar tad barhaspatyar, yan na lohitam na suvarnam tan maitram "The light that is enveloped in smoke belongs to the Visve Devah, the blood-red light to Varuna, the goldencoloured to Bshaspati, that which is neither red nor golden-coloured belongs to Mitra". At any rate, in a more structural approach than Gonda's it might have been observed that Varuna and Bshaspati sometimes seem to stand for the cosmic moieties (e.g., JB. I. 180 and 181) and that the Mitra aspect of the dual deity, accordingly, was considered to belong to neither of the moieties. In passing it may be observed that this fits in very well with the tentative characteristic of Mitra given in IIJ, V. p. 53 as "the link that connects both cosmic moieties". As for Mitra's association with Bphaspati in the ritual (Gonda, p. 82f.), Mitra may here represent the dual godhead, Varuna being considered too inauspicious for this purpose. The general character of this book is clearly defined in the following words of the Preface (p. VIII): "It will mainly consist of a (nearly) complete collection and, as far as may be desirable, thorough interpretation of the relevant Vedic text places". "Speculation concerning that on which the texts are silent has wherever possible been dismissed". Again, on p. 105f., after quoting from a work with a characteristically structural approach, the author concludes with the words: "I for one would prefer relying on textual evidence to mere linguistic suppositions and prehistoric speculations, however natural they may seem to prejudiced minds". It is entirely in keeping with this basic principle that questions to which the texts do not give an answer are left open, although they are sometimes pointed out. Such a question, which has constantly been ignored by former scholars is: How can the fact be accounted for that in the Rigveda hardly any traces are found of the current identification of Mitra with day, and of Varuna with night, which is so well known from the brahmanas? On' p. 37 Renou's reference to "cas rarissimes ... ou le Rgveda distingue entre Mitra et Varuna" is quoted without comment but on p. 107 n. 1, in an incidental remark in a foot-note, this is stated to be, indeed, a fundamental problem. The author here sums up his objections to a book by Duchesne-Guillemin, the first of which are: "In my opinion the term Contract should in any case be avoided; the silence

Page #20
226 REVIEWS of the Rgveda about the "day-night opposition" etc. of Mitra and Varuna should first be explained. ..." Needless to say, I fully agree with Gonda (cf. IIJ. III, p. 211, V, p. 51). The crucial point is, however, how such an explanation can be given on the basis of the explicit textual evidence. If there is a silence on the part of the Rigvedic poets, the question arises as to what was the character of this collection of hymns. A century ago such sharp-sighted scholars as C. P. Thiele and Auguste Barth were fully aware of this basic problem and the latter concluded that "it is evident that a literature such as this will only embrace what is within the scope of a limited horizon, and will have authoritative weight only in regard to things in a more or less special reference, and the negative conclusions especially which may be deduced from such documents must be received with not a little reservation" (The Religions of India, 3d ed., Preface, p. xv; cf. p. 18 = Quarante ans d'indianisme I, pp. 5, 29. Cf. also, e.g., Ogibenin's recent remark on the "positivistic" character of the Rigveda in Struktura mifologiceskix tekstov "Rigvedy", p. 77 n. 1). Many students of the Veda have been struck by the "one-sided" or deviating character of the theology as presented by the Rigvedic poets. Attempts to explain it in terms of an evolution in post-Rigvedic religious thought fail to convince (see IIJ, III, p. 211). On the other hand, stating that the testimony of the Rigvedic hymns has a special character is one thing, explaining it another. The need for a better insight into the special nature of the Rigveda as a whole, in order to understand the reasons of its reticence in certain matters, has more than once been stressed in recent times but the progress made since Barth wrote the words quoted above is not impressive. It will be the task of a future generation of Vedic scholars to tackle this problem. Since Gonda does not take up a position in such disputes there is sometimes inevitably some vagueness in this book. Thus he writes on p. 42f. as follows: "As I cannot enter into a discussion of Varuna's nature I must limit myself to the remark that I do not feel inclined to think, on the strength of texts such as KB. 18,9 "the sun, having entered the waters, becomes Varuna", that there was a more or less exclusive, original or fundamental association of that god with night or the nocturnal heaven. It would rather appear to me that this relation, like that of Mitra with light, is one of the expressions of the complementary character of the duality Mitra-Varuna. It is, to wind up with, perfectly clear, on the one hand, that both gods, guardians of the sta, are quite naturally conjointly concerned also with the light of heaven, which is a manifestation of universal Order, and, on the other, that their functions do not coincide". In general I agree with Gonda. However, a foot-note to the words "association of that god with night or the nocturnal heaven" attributes a different view to me. I may, therefore, be permitted to observe that I am afraid that Renou's statement (Etudes vediques et panineennes XV, p. 8), to the effect that I am inclined "a relier Varuna au ciel nocturne comme conception fondamentale", does not give a quite accurate view of what I attempted to do in my paper in IIJ, VIII. What I there tried to demonstrate is that Varuna is one of the gods of the nether world and, as such, god of the waters and the nocturnal heaven. Anyway, if the functions of the two gods do not coincide (which is beyond question), it may be asked in which respect Varuna's function differed from Mitra's with regard to, e.g., sunrise and sunset. If they actually were two aspects of one and the same deity, one is led to surmise that the function of the one was determined by the other's, and that Mitra's function cannot be fully understood unless it is contrasted with Varuna's. Gonda, indeed, describes their functions in terms of contrast on p. 34, where he follows Luders in considering Varuna a denizen of heaven, while suggesting that Mitra may have been associated with earth. If so, Varuna would have been a god of both day-time and night-time sky. On p. 41 the admittedly difficult stanza RS. V.62.8 is, indeed, explained as saying that Varuna in day-time ascends to the top of the cosmic axis. This, however, is not the author's final conclusion for on p. 113 n. 4 he returns to the

Page #21
REVIEWS 227 problem of Varuna's relations to the night in the words: 'I have no doubts about the correctness of Bergaigne's formulation (o.c. III, p. 119): the role of the 'president of the night "semble donc bien n'avoir ete assigne au couple qu'a cause de Varuna auquel il appartient en propre". Since, then, Varuna was a denizen of heaven and a "president" of the night, he must apparently have been a god of the night-sky. As for V.62.8 quoted above, here Mitra's function may just as well have been transferred to Varuna as Varuna's was to Mitra in the passage commented upon by Bergaigne (viz. V.62.1). In the same way the fact that, as a rule, Varuna is only conjointly with Mitra said to give rain (Luders, Varuna, pp. 715, 719) may indicate that this was properly a function of Mitra alone (see IIJ, V, p. 47f., where attention is drawn to Varuna's role as detainer of water). With regard to Gonda's book, however, it should be stressed that, although the reader may sometimes be in some uncertainty as to the author's views about Varuna, this is only a peripheral point in his approach and does not confuse the issue of his study of Mitra. In strict adherence to his principle Gonda does not try to make light of features of Mitra which are seemingly abstruse and do not fit in with his picture of a beneficial god. One of the most striking is "the curious relation of the god with secretion, the anus and excrements" discussed on p. 124f. It is, indeed, frankly stated that within the framework of this study no explanation of this aspect can be given. Nor can the theory that Mitra is "Contract" account for it. It may here be added that in a review-article on the Indo-Iranian god published in this journal this very aspect of Mitra was quoted as a proof of the correctness of the theory that Mitra is basically the god of deliverance from the bonds of the nether world. See IIJ, V. p. 50f. Since this theory was not a mere speculation on the "original character" of the god but an interpretation of the textual evidence, the question may be raised whether the later Puranic equation of Mitra to Visnu's fontanelle (Gonda, p. 124) was still based upon an association of Mitra with the notion of moksa. As for the connection of Mitra with defecation it may be noted that the prototype of the words avaggatir apanah quoted from Mhbh. XIV.42.34 can be found in avan pranah of the Satapatha Brahmana (see A. Minard, Trois Enigmes II, p. 335f.). For the word apana- G. W. Brown's almost forgotten paper in JAOS, 39 (1919), especially p. 111, is still of importance. 3. For few Vedic gods the theory of a personification of an abstract idea may seem more attractive than for Mitra. This theory, first proposed by Meillet in the Journal Asiatique 1907, II, pp. 143-159, has been a central motif in discussions of the last few decades on Mitra. Gonda's objections to the manner in which this personification is sometimes conceived (p. 104) would seem fully justified. What philological research can achieve and, at the same time, what its limitations are, is most clearly demonstrated by the case of Vstra. As Renou has pointed out in detail (Vrtra et Vroragna, p. 97), it can be shown how the Vedic neuter word vstram, plural vstrani "obstruction, resistance" is only in the latest portions of the Rigveda used as 'a masculine nom. sing. Vrtrah. From a philological point of view it is important to establish this shift from a neuter word to a masculine. For the historian of religion, however, nothing has changed except that the nameless old cosmogonical dragon (ahi-) who impersonated the power of resistance has now received a proper name "Resistance". Gonda (p. 104) objects to Meillet's "attempt to vindicate, on the strength of sociological considerations, the basic meaning "contract" which I cannot find in the texts, and in the second place, the suggestion that this explanation, again under the influence of sociological considerations of a general character, should be based on an etymology which linguistically speaking is only a possibility and from the philological point of 1 Similarly Gonda, p. 34.

Page #22
228 REVIEWS : view not convincing". See also his remark (p. 105 n. 1) on Meillet's "preposterous line of reasoning". Here a few words, of a strictly philological character, in defense not so much of Meillet's theory as of Meillet himself, may not be out of place. Unlike Gonda, Meillet dealt with "Le dieu indo-iranien Mitra". Three years before the publication of his article Bartholomae had completed his Altiranisches Worterbuch, in which (col. 1183) he gave for the masculine word midra-in Avestan the meanings "Vertrag, Abmachung, Kontrakt". In Zarathustra's Gathas, where the god Midra is never mentioned, it occurs only once in Y. 46.5b urvatois va huzantus miroibyo va, where all translators from Bartholomae to Humbach have translated "Freundschaftsbundnisse", "(Freundschafts) Vertrage", etc. Lommel's warning, however, should be noted: "Die rechtlichen und soziologischen Begriffe sind schwer genau zu bestimmen" (Die Gathas des Zarathustra, 1971, p. 136). Although in the late Vendidad mira-is unquestionably a masculine word, the gender of the Gathic word mioroibyo is open to doubt. Its plural form has a counterpart in RS. I. 170.5b mitranam, which is a neuter word. It was not illegitimate to assume, on the basis of the purely philological evidence of the Avesta and the Veda, a Proto-Indo-Iranian neuter word * mitram "contract" and a proper name * Mitras derived from it, whatever the merits or demerits of the linguistic and other arguments which Meillet put forward in support of his theory. It was, however, a prehistoric reconstruction and as such open to revision. Gonda combats it in the following words (p. 114f.): "The masculine name should not, in my opinion, be regarded as having developed, in India, from the neuter - which is foreign to the Avesta -, but rather as a counterpart of the Greek daitpos which does not denote "any man who carves and portions out" but "der Vorschneider", beside Saitpov "one's portion", from which it has in all probability not developed. That means that, if I am right, Mitra, bearing the masculine name, was the dispenser of active benevolence par excellence". This line of argument is perfectly clear. On the one hand, there can be no reasonable doubt about the Proto-Indo-Iranian origin of the god *Mitras. His occurrence in Kurtiwaza's treaty of c. 1380 B.C. testifies to his early worship. On the other hand, if mitram is limited to Old Indian the supposed personification of mitram to Mitras could only have taken place in India - which is plainly unacceptable. Therefore, * Mitras must have been an independent formation of Proto-Indo-Iranian. : For any prehistoric reconstruction the crucial point is obviously how to evaluate the masculine gender of midra- "contract" in the Vendidad. If one considers this inherited, the question arises how in Avestan the meaning "dispenser of active benevolence" has developed to "contract" or "religiose Verpflichtung, Bundnis". If, on the other hand, one does not attach such an importance to the gender in this comparatively late text, it is possible to derive the name *Mitras from *mitram, the more so as in the Rigveda the neuter again shifts to a masculine (see below). These, however, are considerations which lie outside the scope of Gonda's book, from which the discussion of prehistoric reconstructions is strictly excluded (see p. 4 n. 2, p. 115 n. 1). 4. There can be no doubt about Mitra being a beneficent god. The only questions that remain open are whether this was also a specific function of his, and whether the texts actually call him so. As far as the appellative noun mitram is concerned, there is ample room for different interpretations. On the one hand this is due to the general uncertainty as to where the texts refer to the god and where to a "friend" (as Geldner translated the word in many passages). On the other hand this is the result of fundamental difficulties inherent in the process of "Deutung" (interpretation) so eloquently described by Oldenberg, Vedaforschung (1905), p. 30. In general that procedure is the better applicable, the less there are cultural factors involved. This is particularly true of words relating to religion, where in the interpretation of a single word a general view of the religion in its entirety can be more or less implied. As for the appellative noun

Page #23
REVIEWS 229 mitram, it is obvious that a decision about its meaning (e.g., "contract" or "friendship") is bound to have consequences for one's ideas about Mitra. Gonda's study of this problem is to appear separately. What follows are personal reflections of this reviewer, here inserted as an excursus. If one limits oneself to the Vedic evidence, leaving aside all etymological speculations and the Avesta, there is, apart from interpreting all occurrences of mitra, a possibility of attempting a more formal analysis of the ways in which mitra is used, as distinct from, e.g., sakhya "friendship". The compounds and derivatives mitrin and mitreru (Mayrhofer, s.v.) do not prove much, but mitradruh is interesting. It occurs in MS. IV.3.4 (43,12) apam phenena siro 'chinat, tad va enam anvavartata mitradrug asfti, TB. 1.7.1.7 apam phenena sira udavartayat, tad enam anvavartata mitradhrug iti. It has long been compared to Avestan midro.zyam(ca) mioro. drujimca (Y. 61.3, Yt. 10.82) "the infringer of the contract and the man false to the contract", cf. Parthian drwxtmyhr "committing a breach of contract" and Pahlavi miorandruzan "covenantbreakers" (Gershevitch, Hymn to Midra, pp. 113, 153). More instructive is mitram dha, without Old Iranian parallels and mostly occurring in the later portions of the Rigveda. It shows that, although a god can be said to be a sakha as well as a mitra of the devotee, the abstract noun mitram differs from sakhyam "friendship" in that it is established. It is said to be concluded with another person or god in VIII.96.6c indrena mitram didhisema girbhih "we wish to make a mitram with Indra by means of words of praise", X. 108.3c mitram ena dadhama "we will make a mitram with him" (viz. Indra)". Cf. also IV.33.10cd te rayas posam dravinany asme dhatta rbhavah ksemayanto na mitram "(give) us increase of wealth, riches, make, O Rbhus, a mitram like people who are longing for peace!" In these passages "to conclude an alliance" and "to contract a (specific form of) friendship" are both possible. Renou translates "contracter un pacte"; Gonda prefers the latter meaning and has his doubts about other translations (p. 106, n. 6). The incidental use of a plural form in I.170.5ab tvam isise vasupate vasunam tvam mitranam mitrapate dhesthah (to Indra) "Thou, O lord of the goods, hast power over the goods (and) over the mitras, O lord of the mitra(s), being the best establisher (of them)" gives no clue to the exact meaning. In view of the repetition of tvam and the normal construction of dhestha and destha with an accusative the genitive mitranam can (but need not) primarily be construed with isise. Renou, EVP, X, p. 56 renders: "O maitre de pactes, tu es celui qui conclut au mieux les pactes". Otherwise Gonda, p. 106 n. 6: "O lord of friendship, thou givest most friendships". It should be noted that this plural, a hapax in the Rigveda, is reminiscent of midroibyo in Zarathustra's Gathas but that, on the other hand, also sakhya- is once used in the plural. In my interpretation of I.1.70.5 an accusative mitrani must be understood before dhesthah, cf. the parallel passages IV.41.3a indra ha ratnam varuna dhestha "ye, O Indra and Varuna, are the best givers of wealth", VII.93.1d ta vajam sadya usate dhestha (to Indragni) "ye, the best bringers, at once, of prizes to him that desires them". This parallelism is complete in the first case: just as we find in the Rigveda ratnadha and ratnadheya by the side of ratnam ... dhestha, so mitradha and mitradheya are found in the Atharvaveda. Cf. AS. II.6.4 mitrena 'gne mitradha yatasva "O Agni, occupy thy proper position as a establisher of mitram, together with Mitra" (and the curious and isolated passage KS. XXXVII.10: 91,6 mitradha no mitre dadhatai "the establisher of mitram must establish us in mitram"; corruption of no mitrar ?), ASPaipp. III.33.5 mitrena 'gne mitradheyam yatasva, for which the Yajurvedic recensions read mitradheye yatasva "occupy thy proper place in the establishing of a mitram". This much would seem apparent from these passages that mitram had a cosmic significance: Indra presides over them as over the wealth he produces, and he and Agni establish the mitra(s), alone or with Mitra. When surveying all the Rigvedic passages where mitra is used as a (masculine or neuter) appellative, one is led to conclude, first, that this had something to do with a

Page #24
230 REVIEWS peaceful existence. Cf. IV.33.10 (to the Rbhus) te rayas posam dravinany asme dhatta rbhavah ksemayanto na mitram "Give ye us, O Rbhus, increase of wealth, riches, and establish mitram like people longing for peace", II.4.3 agnim devaso manusisu viksu priyam dhuh ksesyanto na mitram "The gods have installed Agni (as a friend among the clans of men, just as those who wish to live in peace (establish) a mitra". In the latter passage a person may be meant (see below for the first verse of this hymn). Geldner translates "Mittler". It may, indeed, be surmised that the constantly recurring phrase mitran na (mitro na, mitra iva) "like a mitram/Mitra" is due to Agni's function as an intermediary between god and men (see also Gonda, p. 48). In this connection the expression jane mitro na (see, e.g., Geldner ad II.6.7, X.68.7) might deserve a closer examination in the light of jana "foreign people" (Caland, translation of PB. XVI.6.8, K. Hoffmann, MSS. XI, p. 89 n. 13, Renou, Etudes sur le vocabulaire, p. 34, Minoru Hara, Pratidanam, p. 256ff.). Be that as it may the formal parallelism between agnim dha "to install the fire" and mitram dha may have been the direct cause for the introduction of the simile mitras na in those passages where agnim dha was used. In these passages, to which the following observations will be limited, the god Agni gave the abstract noun mitram, too, a personal character. The fact should be stressed, however, also in view of Thieme's valuable remarks in Der Fremdling im Rgveda, p. 141, that this shift, from the abstract relation that was established to the person (mitrah) with whom it was established, is an exclusive peculiarity of the Vedic poetic language. Only within the circles of these poets mitram has been re-interpreted as the accusative of a masculine noun mitrah (cf. vstram > Vrtrah). As for the meaning, here the same dilemma recurs that was noted above: was the mitrah a "friend" (thus Geldner) or rather an "ally" (thus, e.g., Oldenberg, SBE, 46, pp. 202, 209, Renou)? The association of mitram with peace (see above) and the fact that in classical Sanskrit this word is the technical term for a king's ally (e.g., Hillebrandt, Altindische Politik, p. 145f.) are in favour of the second alternative. So is also the expression mitram dha which shows that a formal act was needed to establish this relation, whereas for sakhyam "friendship" no corresponding term occurs (otherwise Gonda, p. 112). The shift referred to above is clear in II.4.1 huve ... agnim..., mitra iva yo didhisayyo bhud deva adeve jane jatavedah "I call on Agni ..., who desires to be installed just as an ally (desires an alliance to be concluded with him), the god among godly people, the wise one". With this verse cf. stanza 3 quoted above and, e.g., VIII.23.7a-8d agnis vah purvyar huve ... mitrar na jane sudhitam stavani "I first call on your Agni ... who is well-installed among people living in accordance with the Universal Order, like an ally (with whom an alliance has been well-concluded)". In the latter verse the interpretations vacillate between the abstract and concrete meaning of mitra/ Mitra (e.g., Renou, EVP, XIII, pp. 68, 151). In several other passages the agni sudhita "bien place", "bien mis en place" (Renou) is likened to a mitra conceived as a person, the tertium comparationis being sudhita "well-installed"/"well-concluded". Cf. IV.6.7cd adha mitro na sudhitah pavako 'gnir didaya manusisu viksu "The pure Agni is now burning among the clans of men, wellinstalled like an ally (with whom an alliance has been firmly concluded)". Geldner's translation "wohl aufgenommen wie ein Freund" disregards the technical meanings both of agnis dha and of mitram dha (cf. Renou, EVP, XIII, pp. 11, 98). Similarly V.3.2cd anjanti mitram sudhitam na gobhir yad dampati samanasa krnosi "They anoint thee, who art well-installed like an ally (with whom an alliance has been firmly concluded), with milk as thou makest husband and wife unanimous" (otherwise Thieme, Der Fremdling, p. 139, Mitra and Aryaman, p. 85, Renou, EVP, XIII, pp. 20, 106, Gonda, p. 48), VI.15.2ab mitram na yar sudhitam bhrgavo dadhur "Whom the Bhtgus have well installed (sudhitam dadhur) like an ally (firmly connected by an alliance)", VIII.23.8bc (yam krpa sudayanta it), mitrar na jane sudhitam stavani (see above). Only . Fremdling, p. 139 Milanest husband and wife undance has been firmly co

Page #25
REVIEWS 231 once, in the last book, is the conventional phrase used in a different way, cf. X.115.7cd (hymn to Agni) eva 'gnir ... vasu stave ... nrbhih, mitraso na ye sudhita stayavo dyavo na dyumnair abhi santi manusan "Ainsi Agni est-il loue ..., lui le Vasu, ... par les seigneurs, , qui, tels des allies au contrat) bien conclu, fideles a l'Ordre, depassent en eclats les humains comme les cieux (depassent la terre)" (Renou). That in all these passages the phrase mitram dha actually underlies the simile is shown by a few other instances where the non-compound hita is used alone in a simile to denote a firm, reliable person. It proves that the procedure expressed by dha was an indispensable condition for someone to become a mitrah. Cf. I.166.3cd uksanty asmai maruto hita iva puru rajamsi payasa mayobhuvah "the refreshing Maruts, like firm (mitras), sprinkle for him the many regions with milk", IV.57.1ab ksetrasya patina vayam hitene 'va jayamasi"We will be victorious with the lord of the soil as with a firm (ally)". Therefore, X.7.5 dyubhir hitam mitram iva ..., bahubhyam agnim ayavo 'jananta can hardly mean anything else but "Day after day the Ayus have generated Agni, like a firm ally, with their arms" (otherwise Gonda, p. 50). If so, Renou's translation of hitamitra as "avec qui on forme alliance" (see Gonda, p. 106 n. 6) is unobjectionable. This excursus was an attempt to arrive at some greater certainty about the meaning of mitram by starting from mitram dha. Its result confirms Renou's translation "conclure une alliance", and "allie (par pacte)" for mitra (e.g., EVP, X, pp. 21, 70). A few words may finally be added on sumitradha. There is a vacillation among scholars as to whether Rigvedic sumitra, durmitra and amitra are karmadharayas (pw.: "ein guter Freund") or rather bahuvrihis. Debrunner, following Thieme, finally decided upon bahuvrihis (Nachtrage on Altind. Gramm. II/1, p. 268, 19 and II/2, p. 814), which is indeed most probable (cf. Old Persian hamiciya- "rebellious, coniuratus"). A quite different case is, anyway, the Yajurvedic compound sumitradha. It occurs in a formula addressed to Soma during the ritual of its purchase, viz. mitro na ehi sumitradhah (KS KKS TS) or sumitradhah (MS VS). As is apparent from the accent, this cannot mean "gute Freunde machend" (pw.; not recorded in PW. and Wackernagel-Debrunner), as this would have been *sumitradha. It is, accordingly, a determinative compound of su- and mitradha (also attested in KS., see above) and the formula means: "Come to us as Mitra, well-concluding alliances" (Otherwise, Keith, transl. of TS. I.2.7.1, VI.1:11.1 and Gonda, p. 60). Soma, when purchased, is at first the inauspicious Varuna: like Agni, Soma comes from the nether world to the world of men and thus fulfills, as an intermediary, the characteristic Mitra-function (see IIJ, V, p. 53). This, however, involves the necessity of converting him into Varuna's auspicious aspect, impersonated by Mitra. 5. A few marginal notes may conclude these reflections on the problem of Mitra, which were inspired by Gonda's thought-provoking book. On p. 52 there is a brief reference to Varunic features of Agni. Cf. also p. 6 and p. 46 n. 2 with the reference to Renou's distinction between "Agni varunien" and "Agni solaire" in EVP, XIII, p. 106. In this connection Agni upanaddha is of some importance, because he is identified, with Varuna (see Hillebrandt, Vedische Mythologie II, p. 60). On p. 68 this is rightly paralleled with the soma upanaddha. As for the relevant feature of the rite described by Gonda, I do not think this is Mitra's conjuring effect on the wrath of another deity. As is apparent from the ritual of the somakrayana, the central motif is that Soma is Varuna (that is, respresents the Varuna-aspect of the world he comes from) as long as he is tied up. Cf. MS. III.7.8 (85,18) varuno va esa etarhi varunadevatyo yarhy upanaddhah, AB. I.30.26 varunadevatyo va esa tavad yavad upanaddhah. He must first, by being untied, be transformed into the Mitra-aspect of that world. Cf., e.g., MS III.7.8 (86,7) varunam va enam etat santam mitram akah, KS, XXIV.6 (96,2), 7 (97,9), KKS, XXXVII.7 (201, 21/234, 21), 8 (202, 11/236,10) mitram evainam kurute, KS. XXIV.6 (96,17) yan mitra karoti, KKS. XXXVII.7 (201,21/

Page #26
232 REVIEWS 235,16) yan maitram karoti, TS. VI.1.11.2 yad varunar santam maitram karoti. The same is true of Agni, cf., e.g., TS. v.1.5.3; 6.1.1 varuno va agnir upanaddhah, AB. III.4.6 sa yad Agnir ghorasamsparsas tad asya varunam rupam, tam yad ghorasamsparsam santam mitrakrtye 'vo 'pasate tad asya maitram rupam "Agni's being dread of contact constitutes his Varunic form. That they make him who is dread of contact to Mitra as it were and meditate on him constitutes his form as Mitra". The form mitrakrtya in this quotation is probably a gerund like pumskrtya; see Debrunner, Altindische Grammatik II/2, p. 786 (also Gonda, p. 45. Otherwise Wackernagel-Debrunner, op. cit., III, p. 117). It should be noted that the mythological idea which underlies this ritual may have been expressed by the Rigvedic poets in, e.g., 1.75.4b agne mitro asi priyah "O Agni, thou art the dear Mitra" and perhaps in I.91.3c (hymn to Soma) sucis tvam asi priyo na mitro "Thou art pure like the dear Mitra". See also above, p. 230. In conclusion the question may be raised what alliance Soma was considered to conclude when during the ritual of the somakrayana he was addressed with the words mitra na ehi sumitradhah "As Mitra, effectively concluding an alliance, come to us". It may be called to mind that W. Brede Kristensen has stressed the cosmic aspects of the alliance (see IIJ, V. p. 53 with references) and that in the Rigveda the waters and Dhisana are said to make the alliance effective (1.96.lc apas ca mitram dhisana ca sadhan, cf. III.5.3d? Otherwise Gonda, p. 112 n. 7). Does the Yajurvedic formula refer to an alliance between the Varunic nether world and the upperworld? See above, p. 225. Needless to say that such a suggestion is alien to the spirit of Gonda's book. P. 117: Gonda is, I think, fully right in questioning Zaehner's contention that in the tenth book of the Rigveda Yama has replaced Mitra. Such evolutionistic explanations often miss the point. It would seem to be a natural consequence of Mitra's nature that he has no place in the realm of the dead and the dark side of the nether world. Yama, as the deified ruler over the dead, here complements Varuna. In later times he even takes over the place that was Varuna's. In this connection it is interesting to observe that in Bharata's Natyasastra III.26, 38, 61 Mitra is called upon together with Yama and is localized in the South. The only explanation for this grouping would seem to be that here Yama is, indeed, the successor to Varuna. To the same conclusion points the parallelism between Mitra and Vivasvat in the Vastupurusa as summarized by V. S. Agrawala, Matsyapurana - A Study, p. 347 (with diagram). Elsewhere in the Natyasastra, however, as in the adhivasana of the theatre (III.6), Mitra is mentioned conjointly with Agni. P. 131: In connection with Gonda's suggestion that the cakra mentioned in Mhbh. 1.218.34 as Mitra's weapon (mitras ca ksuraparyantam cakram glhya vyatisthata), is the sun, the well-known passage of Garuda's stealing the Some may be of some importance: Mhbh. I.29.2 sa cakram ksuraparyantam apasyad amstantike, paribhramantam anisam tiksnadharam ayasmayam. A rare slip of the pen occurs on p. 54 n. 5, where the words mitranam pujayitah (Sayana's gloss on the vocative mitramahah X.37.7) is translated "worshipped (honoured) by friends" instead of "O worshipper of friends!" All in all, this exhaustive collection of all the relevant material will be welcomed by every worker in this field. It will no doubt stimulate the discussion and may even ultimately lead to a more profound reflection on the basic methodological differences which are the cause of the somewhat chaotic character of present-day disputes on the nature of god Mitra. F. B. J. Kuiper

Page #27
REVIEWS 233 A. J. Alston (tr.), The Realization of the Absolute. The "Naiskarmyasiddhi" of Sri Suresvara. Translated by A.J. A. London, Shanti Sadan, 1971. xvi +269 pp. PS2.00. Suresvara's Naiskarmyasiddhi is one of the most popular Vedanta texts. Potter's Bibliography of Indian Philosophies lists five editions (Nos. 2393, 2394, 2396, 2398 and 2399) to which one must add a recent edition by Swami Satchidanandendra Saraswati: Naishkarmya-siddhi of Sri Suresvaracarya with the Klesapaharini by Swami Satchidanandendra Saraswati (Adhyatma Granthavali), Holenarsipur, Adhyatma Prakasha Karyalaya, 1968. The edition generally used is Hiriyanna's revision of Colonel G. A. Jacob's edition (Bombay Sanskrit and Prakrit Series XXXVIII, Bombay, 1925). It contains Jnanottama's Candrika, a very good introduction and many useful notes at the end. Hiriyanna has consulted four unpublished commentaries: Citsukha's Bhavatattvaprakasika, Jnanamsta's Vidyasurabhi, Akhilatman's Naiskarmyasiddhivivarana and Ramadatta's Sarartha. Hiriyanna's notes often quote from these commentaries. Especially interesting are his quotations from the Sarartha. As far as I know, neither this commentary nor the three others have as yet been edited. The Naiskarmyasiddhi has only recently been translated into English. In 1933 Ras-Vihari Das published his Essentials of Advaitism, Suresvara's Naiskarmyasiddhi explained in English (Punjab Or. Series No. 21, Lahore). This publication, which is only known to me from F. Otto Schrader's review of Hacker's Untersuchungen uber Texte des fruhen Advaitavada (ZDMG, 101, 1951, pp. 418-424), seems to give an analysis rather than a complete translation. Alston's translation appeared first in typescript form in 1959. In 1965 S. S. Raghavachar published a translation accompanied by the Sanskrit text (Naiskarmyasiddhi of Sri Suresvaracarya. English translation by S. S. Raghavachar, University of Mysore, Mysore). In his Untersuchungen uber Texte des fruhen Advaitavada 1. Die Schuler Sarkaras (Wiesbaden, 1951), Paul Hacker has made a very thorough analysis of the doctrines of the Naiskarmyasiddhi. In the course of his study he translated many verses of the text and one can only regret that he has not published a complete translation. Hacker's work is of fundamental importance for a better understanding of the Naiskarmyasiddhi. Alston has made much use of Hacker's study, to which he often refers in his notes. In his introduction Alston also mentions Saccidanandendra Svamin's 'profound and critical treatment of the Naiskarmyasiddhi in his Vedanta Prakriya Pratyabhijna (Holenarsipur, 1964), pp. 210-276. I have not been able to consult a copy of this work. The Naiskarmyasiddhi is a very interesting text. For the non-specialist in Advaita Vedanta it has the great advantage of being a short work and not a commentary. Indian scholars have praised its style, but not without reservations. Hiriyanna has noticed two defects: repetition of thought and argument, and the use of a rather large number of un-Papinean forms. Satchidanandendra Saraswati mentions the ruggedness of the style and the occasional grammatical and metrical difficulties. However, Jnanottama's commentary and Hiriyanna's notes are very helpful in understanding the text. Raghavachar's translation is very useful, although from a stylistic point of view it is much inferior to Alston's supple English. Moreover, Raghavachar has closely followed Jnanottama's interpretation, so much so that his translation is sometimes no more than a paraphrase of the latter's commentary. 'Alston's revised translation has taken Raghavachar's interpretation into account, but one has the impression that he could have derived more help from it. His translation is extremely readable but sometimes too free. In several instances there is no doubt that Alston has misunderstood the text. His revised version is a great improvement. However, there are a few places in which his original translation seems preferable. One must hope that a Vedanta specialist will critically examine Raghavachar's and Alston's translations. I have read Alston's translation with great pleasure and I only venture to make a few criticisms in

Page #28
234 REVIEWS order to draw the attention of more qualified readers to this new translation of a famous work. The remarks which follow enumerate a number of passages which I believe have not been entirely correctly translated by Alston. Wherever Raghavachar's translation seems to me to be essentially correct, I have refrained from further comment. In each case I have given first the Sanskrit text and, following that, Alston's revised translation. R. refers to Raghavachar's translation, Alston-I to the first edition of his translation, and Hacker to his Untersuchungen. 1.63 karmaprakaranakanksi jnanam - "The kind of knowledge (i.e. erroneous) which leads to action". R. "The knowledge that belongs to the context of action". 1.86 na hy ekatra pramanatam / vastuny atanti manani - "Fot the different texts do not attain authority in relation to one and the same object". Manani = pramanani. R. "The several modes of valid knowledge ...". 1.96b karakadyatmadarsinam - "the one ... possessed of the means to action (a body etc.)." R. "one who ... sees factors involved in action as the Self itself". II.14c drstva sadharanam deham - "Seeing what a contemptible object the body is". R. "Seeing that the body is the common possession of you and the dogs". II.20 Intr. dehadikaryakaranasamghata - "the individual personality, with all its effects and instruments from the body up". R. "the complex aggregate of effects and instruments that is the body etc.". For the expression karyakaranasamghata see Olivier Lacombe, L'absolu selon le Vedanta (Paris, 1937), p. 125, n. 1 and p. 129. II.40 drastapi yadi drsyaya atmeyat karmatam dhiyah / yaugapadyam adrsyatvam vaiyarthyam capnuyac chrutih - "If the Self, which is the seer, could become the object of the intellect, which is already an object for the Self, neither could really be an object (because there would then be no subject). And, the statement in the Veda 'there is no break in the sight of the Seer would be rendered vain". Alston-I translates 40c: "both would simultaneously be subject, and hence neither could really be an object". R. follows the commentary: "If the seer were to be objectified by the intellect, which itself is the seen in relation to the seer, then both the seeing self and the seen intellect, should simultaneously be both seeing and seen. As both are seers, there should be no object of seeing. And sruti would be useless in that case". The difficulty arises from the fact that the statement in 40d does not follow logically from 40c:"they would simultaneously not be an object (both being subjects)". One must add before d, as does the commentary drsyadrsyatve catmanah "the self being the object of (his own) object". By becoming an object the self ceases to be a seer and, consequently, the sacred text (: na hi drastur drster viparilopo vidyate) would become meaningless. I suggest translating c as follows: "at the same time (the Self and the intellect) would be subject (and the Self would be the object of its own object, the intellect)". II.41 naluptadyster drsyatvam drsyatve drastrta kutah / syac ced drg ekam nirdrsyam (or eka nirdrsya, cf. Hiriyanna's ed., pp. 239-240) jagad va syad asaksikam' - "The one of uninterrupted vision cannot ever be an object; if it were an object, how could it be the Seer? If it were to become an object) then the Seer would be the one seen or else the world would be devoid of a Witness". Much better is Alston-I: "How can the one of uninterrupted vision ever become the object, or how can the object become a seer? In the latter case only a seer would exist with no object to see; in the former case the world would be without a witness". For ab I propose: "The one of uninterrupted vision cannot be an object, for how can the object be a seer?" II.42 Alston's translation omits artam which refers to BIhadaranyaka Up. III.7.23: nanyo 'to 'sti drasta ... ato 'nyad artam. II.58 esa sarvadhiyam nyttam aviluptaikadarsanah / viksate 'viksamano 'pi nimisat tad dhruvo 'dhruvam - "This (Self) is the one unbroken witness of the dance of every intellect. Verily, itself eternal, it views the passing without the act of looking - as if through half-closed eyes." Nimisat qualifies nyttam: R. 'the insentient dance'. (cf.

Page #29
REVIEWS 235 Jnanottama: nyttam nimisaj jadam). Hiriyanna quotes from the Sarartha the gloss nimisat = parinami. II.63 nagesu "mountains"; R. "trees on the banks". 11.69 sabdadyakaranirbhasah ksanapradhvamsinir drsa / nityo 'kramadrg atmaiko vyapnotiva dhiyo 'nisam - "The one motionless Self, not subject to sequence or succession in time, the eternal Seer, constantly pervades, as it were, the flickering modifications of the mind, which in turn illumine the forms of material objects". R. "The one, eternal and non-successive seer, through his consciousness, pervades as it were, all the perishing functions of the mind always which take up forms of objects like sound and colour". Alston does not translate drsa. Hiriyanna quotes from the Sarartha: sabdadyakarena nirbhasanta iti sabdadyakaranirbhasah tattadvisayakara vrttih - "They appear in the form of sound, etc. The functions of the mind appear in the form of sound, etc., they take the form of their objects". This gloss shows clearly that nirbhasa means 'appearance', not 'illumination' or 'reflection'. According to Hacker the relation of the Self to the mind is similar to that of the mind to the objects: "der die Formen der Objekte erleuchtende oder reflektierende Innere Sinn ist seinerseits Objekt der Erhellung durch den Atman (sabdady-akara-nirbhasa ... bhasya II 91)". (p. 45). However, the mind transforms itself in the forms of the objects but the Self does not transform itself into the mind. 11.94 yadyad visesanam drstam natmanas tad ananvayat - "No qualification seen anywhere belongs to the Self, for it conforms to no limitations". R. "Nothing that appears as qualitatively determining the Self, does really belong to the Self". Neither translation renders ananvayat correctly. Hacker explains very well: "er [i.e. der Atman) ist eigenschaftslos (nirvisesana), denn alle individuellen Qualitaten haften ihm nicht dauernd und durchweg an (ananvaya)" (p. 39). One may suggest the following translation: "No qualifications seen anywhere belong to the Self, for they are not permanently associated with it". 111.6 Intr. nihsamdhibandhanam mithyajnanam - "erroneous knowledge which consists in connecting what are not connected". R. "false knowledge, attaching itself fast". Hacker renders nihsandhibandhana as "feste Verknupfung" (p. 58). III.6d jnatmana tyajayed vacah - "One should give up the words through awakening to the Self as knowledge". R. "The words of the scripture set it aside by focussing on the Self of the nature of the pure consciousness". Alston has not seen that vacas is the subject: "The words of the scripture make one give up (ignorance) by means of the Self, the knower". See also Hiriyanna's quotation from the Sarartha (p. 257). 111.73 bhrantiprasiddhyanudyartham tat tattvam bhrantibadhaya / ayam nety upadisyeta tathaivam tat tvam ity api - "When a man wishes to dispel the erroneous notion of another, he first conforms his speech to that erroneous notion as if it were a fact, and then he says it is not so'. The same is the case with 'that' and 'thou'." R. "When something is the object of an illusory apprehension, it is first of all referred to as it is known under the illusion and then its real nature is taught by the removal of the illusion by saying 'This is not so'. So is the case in 'That thou art' also". 111.79 samsaritadvitiyena paroksyam catmana saha / prasangikam viruddhatvat tattvambhyam badhanam tayoh - "Being the transmigrant' and 'not-being-immediately-evident' are only accidental characteristics of the Self, since they are in mutual contradiction. They are therefore both negated, (the first) by 'that', and (the second) by 'thou'." - R. "Being in bondage is contradicted by being one without a second. Being mediate and indirectly presented is contradicted by self-hood. The two terms 'That' and 'Thou' intending a unitary import sublate by implication these contradicted meanings". Alston follows Hacker in interpreting prasangika as 'accidental (cf. Hacker p. 81: 'zufallig'). However, Jnanottama explains prasangika by nantariyaka, 'inherent', a meaning which is not to be found in the Petersburg Dictionaries. However, it is given by Monier-Williams with a reference to Wilson's dictionary: `inherent',

Page #30
236 REVIEWS 'innate'. See also Apte's The Student Sanskrit-English Dictionary, 2nd ed., 1922: prasangika 2) 'connected with', 'innate'. Jnanottama mentions a variant reading prasangikaviruddhatvat which seems preferable. III.88 Intr. abhyupagame 'pi ca prasamkhyanasatenapi naiva tvam sambhavitadosan mucyase - "And even if it were admitted (that the Self was really known to be connected with pain), then not even by a hundred acts of symbolic meditation could you escape from the defect that would arise". R. "If the fact of misery etc. are admitted to be established by other means of knowledge, a hundred meditations even cannot bring about emancipation from evil". Alston renders sambhavita wrongly and Raghavachar omits it. Sambhavitadosa is the defect imputed to the Self by considering it as duhkhita. III.90 Intr. tac [i.e. prasamkhyanam] canusthiyamanam pramitivardhanaya paripurnam pramitim janayati na punar aikagryavardhanayeti / yathasesasucinide strikunape kaminiti nirvastukah purusayasamatrajanitah pratyaya iti - "When properly performed it [i.e. meditation) generates perfect knowledge and does so by improving on such knowledge as already exists, and not through merely improving the mind's powers of concentration. It is not to be compared with the purely imaginary notion, arising through subjective mental activity, that the body of a woman, that receptacle of every impurity is a charming object of desire". Alston follows R. in taking the sentence beginning with yatha as illustrating aikagryavardhanaya. Alston-I, however, agrees with Hacker, p. 99:"fes erzeugt durch Vermehrung der Erkenntnis vollendete Erkenntnis, nicht aber bloss durch Vermehrung der Konzentration'. Eine solche Wiederholung von Denkakten ist nach Ansicht des Opponenten sogar imstande, einen festen Glauben zu erzeugen, dem uberhaupt keine Wirklichkeit entspricht, z. B. entsteht durch intensives Denken eines Mannes an eine Frau, die doch in Wirklichkeit nur ein Leichnam, voll von aller Unreinheit' ist, die Vorstellung, sie sei begehrenswert". Hacker is undoubtedly right. The opponent maintains that prasamkhyana leads to a more perfect knowledge but not to increased concentration. It is even capable of producing a wrong notion, not based on reality, as for instance the notion of a 'charming woman' with regard to a corpse full of filth. Alston-I correctly remarked that "prasamkhyana on the holy sentences yields 'improved' knowledge of reality in the same way, with the difference that this variety of 'improved' knowledge happens to be true". See also Jnanottama: loke bhavanapracayasyavastuny api pratyayadardhyahetutvadarsanat - "one sees that in ordinary life repeated acts of imagination produce a firm notion even with regard to something devoid of reality". The Klesapaharini relates yathasesasucinide to aikagryavardhanaya: aikagryavardhanayam nidarsanam 'yathasesasucinide'. IV.56 vastavenaiva vsttena nirunaddhi yato bhavam / nivsttim api modnati samyagbodhah pravrttivat - "Through knowledge of reality he brings empirical being to an end. Right-knowledge destroys the path of renunciation as surely as it destroys the path of action". R. translates vastavenaiva vrttena by 'in reality'. Hacker is more precise: "Weil die rechte Erkenntnis bloss dadurch, dass sie Wirklichkeit geworden ist, das Werden verhindert, zerdruckt sie die Werkenthaltung ebenso wie die Tatigkeit" (p. 105). Australian National University .J. W. de Jong A.S. Acharya, Barkur Kannada (= Publications of the Centre of Advanced Study in Linguistics, University of Poona, series major, no. 1; Linguistic Survey of India series, 6). Poona, Deccan College, 1971. 102 pp. Rs. 10/-. The present volume is a study of a Dravidian dialect spoken around the small town of Barkur, in Udipi taluk, South Kanara district, Mysore state, on the west coast of India.

Page #31
REVIEWS 237 Although Kannada is the main language of Mysore state, Tulu is the principal mothertongue in South Kanara district; and the Barkur area, Acharya tells us in his preface, is something of a linguistic island. Data was collected during a two-month period, principally from a single educated Brahmin speaker. The data presumably reflect informal speech; but since the informant no doubt knows formal Literary Kannada too, we might wish that Acharya had given us a little sociolinguistic commentary. One wonders, too, what the non-Brahmin speech of Barkur may be like. However, Acharya's description is purely descriptive, and follows a classic pattern of taxonomic structuralist linguistics: the divisions are (1) Phonology, (2) Morphology, (3) Texts, and (4) Vocabulary. The focus is on the word, rather than the sentence, to such an extent that even the rules of external sandhi are given, not under Phonology, but as a section introducing the Texts; and there is no explicit analysis of syntax. It appears that more monographs on this same model may be forthcoming: nine titles are announced in the "Linguistic Survey of India series", including two others by Acharya himself (although the present work is no. 6 in the series, it is apparently the first one published). This is a little discouraging: Acharya is careful, systematic, and thoroughly competent within the framework of his model; but this is a model which was taught at Deccan College in the 1950's. Since then, linguistics has changed, in India and around the world. Considering the skill and energy of the authors in this series, it would be a pity if we were to receive from them only relatively superficial descriptions. A look at the Barkur vowels, as presented by Acharya (pp. 1-5) will illustrate what I mean. The system is said by him to have short and long vowels in eight qualities: ie a oui. Is this really an important departure from the five-vowel system of the Dravidian literary languages? An examination of Acharya's data suggests that it is not: thus, most of his occurrences of & occur before a syllable containing a low vowel (ele 'leaf', pe:te 'market), and e generally occurs elsewhere (eru 'ant', e:ti 'blow'). There are indeed some surface contrasts of e with e, e.g. herge 'outside' vs. herge childbirth'; but the alternative form herige and the stem heri 'to give birth' (pp. 100-01) suggest underlying forms/herage/ 'outside' vs. /herige/ 'childbirth', with lowering of e to a when a low vowel follows. Similarly, o generally has a low vowel in the next syllable (@le oven', ko:te 'fort'), as compared with o (koai give', o:li 'palm leaf'). Of the exceptions, some are like toli 'to wash', cf. toli 'to step'; but it is noteworthy that although No Barkur verbs end in e, Literary Kannada has such verbs, including tole 'to wash'. It is evident that Barkur shows the same metaphony, lowering e o to respectively, as do many other Kannada dialects: first tole becomes tole, then a separate change of final e to i (only in verbs, for Barkur) results in toli 'wash'. Other surface occurrences of Barkur o in examples like ko]ke "third crop' vs. kolke 'a hook can be understood in the light of the corresponding Literary forms kolake and kolike, kolike respectively. And metaphony is not merely a historical explanation here: Literary Kannada, showing the underlying 5-vowel system, is undoubtedly in the active repertory of Acharya's educated informant, and in the passive repertory of even illiterate Barkureans.3 What of Barkur i ti, also absent in Literary Kannada? Acharya notes that i never 1 An example like etta 'where' (p. 1) evidently requires a special explanation; but it should not obscure the marginal nature of the e-e contrast in this dialect. 3 Cf. W. Bright, "Dravidian metaphony", Language 42 (1966), pp. 311-22, esp. pp. 317. 8 Cf. W. Bright, "Phonological rules in Literary and Colloquial Kannada", JAOS, 90 (1970), pp. 140-44. An example like Barkur koda 'he cannot give' (cf. koda "pot) requires a special explanation: I suspect that low vowels in the inflectional endings of verbs do not cause metaphony in this dialect.

Page #32
238 REVIEWS occurs initially, but an inspection of his data reveals even stronger restrictions. In word-initial syllables, short u, e.g. in uri 'fire', occurs to the exclusion of short i (except in some Sanskrit loanwords with r- in the original). In medial syllables, short i occurs to the near exclusion of short u, as the counterpart of Lit. Ka. u (Barkur elibi 'bone', Lit. elu(bu); of Lit. Ka. i, with which it sometimes varies freely (he:lida~ he:lida 'he said', Lit. he:lida); and of Lit. Ka. a (entidi'what', Lit. enthadu). In a few cases, however, medial u occurs after an initial syllable containing u (kurule 'young plantain plant'), and in future forms like ma:d-uv-a 'doing' (participle). Finally, short s occurs as the wellknown Dravidian "enunciative vowel", corresponding to Lit. Ka. u (no:vi pain', Lit. no:vu), but Barkur has final u in two kinds of examples - namely in nouns of the shape CVCV (suru 'beginning', magu 'child'), and in 3rd person neuter verb forms of the future tense (ma:du 'it may do', cf. ma:dido!'). All this is to say that the superficial difference between i and u is largely predictable if Lit. Ka. forms are taken as basic, as is the case in many other Kannada dialects. Acharya's description of the morphology of nouns and verbs is as rigorously taxonomic as his phonology, with abundant use of zero allomorphs. E.g. ma:du 'it may do' is analysed as ma:di 'do', plus a zero variant of the future (otherwise -p, -6, -v), plus -u *3sg. neuter'; one wonders if a preferable analysis would not be as ma:d-uv-u with a rule contracting uvu to u. Much reference is made to free variation of allomorphs; thus the accusative of adi'it is said to be adinnar adanna- adi; little attempt is made to describe such phenomena in terms of general phonological rules (such as a frequent change of medial a toi), and no clue is given as to the possible sociolinguistic background of such variation. Apart from such matters of presentation, Acharya's description presents some interesting data. Although noun morphology seems much like that of other colloquial Kannada dialects, here Archarya muddies the waters by positing a comparative case morpheme' -iginta (with other allomorphs, p. 25) which is clearly a combination of the dative case suffix with a postposition inta. The verb morphology shows full paradigms for present, past, future (usually translated with 'may') and negative. In the future, there are some interesting stem-formations in -p, -b which are not used in Lit. Ka.: e.g., from bart 'come', bappa 'he may come'; from tinni 'eat', timba 'he may eat'. The corresponding 3sg. neut. forms, where one would expect *bar-uv-u etc., are bakki: 'it may come', tingi'it may eat'. Various inconsistencies are discoverable in Acharya's description. Thus verbal nouns like ma:di:di 'doing' are treated along with derivative and compound nouns (p. 16), though they are best understood as a type of non-finite verbal form, built in fact on participles (pp. 42-3). In describing the participles themselves, Acharya holds (p. 43) that they are formed by adding -a to a past stem (ma:did-a 'done'), to a future stem (ma:duv-a 'doing', evidently taking the place of a PRESENT participle), and to a negative GERUNDIAL; but why not recognize a negative STEM in ma:dad-a? An "Assertive" form like tarki 'one should bring' is treated as a "finite verb form" though it makes no distinction of persons (pp. 42-3); but this looks like a reduced combination of verb plus impersonal "modal", perhaps indeed be:ki 'it is required' (p. 47). Misprints are more numerous in this volume than they have been in some other linguistic publications from Deccan College; but most will be self-correcting for the 4 Cf. W. Bright, "The enunciative vowel", International Journal of Dravidian Linguistics, 1 (1972), pp. 26-55, esp. pp. 44-5. The long i of Barkur is more unusual. It apparently results from two phonological rules: (1) contraction of uva to i: in forms like ma:duva 'doing' (participle) +-di + ma:di:di 'doing' (verbal noun); and (2) contraction of i with the a of anni 'to say', e.g. e:li 'get up!', eli:ndre 'if one says "get up!"

Page #33
REVIEWS 239 specialists who will use this monograph. The work stands as a valuable collection of data from a Kannada dialect which, though spoken in the interestingly multilingual South Kanara district, retains most of the characteristic features of colloquial Kannada as used elsewhere. University of California, Los Angeles William Bright