Astadhyayi As A Case In Textual Criticism

Author(s): Ashok Aklujkar
Publisher: Ashok Aklujkar
Catalog link: https://jainqq.org/explore/269570/1

Page #1

Publications of the Centre of Advanced Study in Sanskrit, Class E, No. 9.
Proceedings of the International Seminar on Pämini : 1-10, 1982 (held in July 1981) Edited by S.D. Joshi and S.D. Laddu, Univ. of Poona, Pune, 1983.
The Astadhyāyi as a Case in Textual Criticism
(Part 2)
Ashok Aklujkar
1.1 In the first part of this paper, which will be presented elsewhere on account of its nature and length, I have pointed out that a comprehensive objective textual criticism of Panini's ("P", in abbreviation ) Astādhyāyi ( abbreviated henceforth to "A") is still a desideratum, and have delineated what this textual criticism would involve. In the second part being presented here, I wish to expand upon one aspect of the programme outlined in the first part. This aspect is that of secondary textual criticism.
1.2 In explaining the considerations of which a scholar attempting a secondary textual criticism of the A should be aware, I shall choose as examples those rules of P which have emerged, as a result of Hartmut SCHARFE's noteworthy study Panini's Metalanguage, as containing an anomaly of wording. The intention behind such a choice is twofold:
(a) To establish the claim I made in reviewing SCHARFE's study
in Orientalistische Literaturzeitung, Vol. 72, No. 3 (1977), p. 305 a, namely that "the possibility of textual corruption, when explored judiciously and rigorously, does yield straightforward and convincing explanations of some of the ano. malies that SCHARFE SO painstakingly brings 10 our atten
tion". (b) To illustrate how important a criterion consistency with P's
practice elsewhere would be in secondary criticism-how a diligent editor would derive considerable benefit from a
1. In using the adjective " secondary ", I am deliberately steering clear of the usual distioction between lower and higher textual criticism. While attempting a textual study of any work I find it convenient to distinguish between what is explicitly available in the sources (manuscripts, commentaries, etc.) of textual study and what is implicit and hence needs to be recovered. I think such a distinction does justice to the nature of materials available in the Indian tradition, which in most instances consist of individual works and text-based direct as well as indirect commentaries to them, and goes a long way towards adapting the science of textual criticism to the Indian scene. It has the additional merit of separating sharply what the editor supplies from what the editor is supplied with Secondary textual criticism, for example, includes emendation, commonly thought of as part of lower textual criticism, as well as tracing of the sources of the edited work, tracing which is generally held to be a constituent of higher textual criticism.

Page #2

2
Proceedings of the International Seminar on Panini
word-index2 of the A, once all that the manuscripts and other sources supply explicitly has been utilized in accordance with the principles of textual criticism.3
2.1 (SCHARFE 1971: 8b :) The form śnasoh in 6.4.111 (śnasor al-lopah) has been a puzzle to the students of P for a long time. As the rule pertains to śna, the class-sign of the seventh class of roots, and the root as, one expects śnásoh. The attempts made so far to render śnasoh acceptable are in effect an acknowledgement of its strangeness; they can hardly be called explanations. The Kašika remarks śnasor iti para-rupan sakandhradişu drastavyam. Vitthala's Prasada on the Prakri yakaumudi and the Balamanorama and Tattva-bodhini commentaries on the Siddhanta-kaumudi simply echo this line. Purusottama-deva merely protects himself from the possible charge of failing to notice the peculiarity. He has nothing more to say than śnasor iti para-rupatvam sautrat vat. Most other recognized Sanskrit commentators, as far as I can ascertain, do not even take the trouble of indicating that there is a problem here. Of the modern scholars, BÖHTLINGK observes after giving the forms 6.4.111 is supposed to derive: “Dieser Ausfall ist im Sutra schon durch śnasoh ausgedruckt; das a erleichtert hier nur die Aussprache.” SCHARFE ( 1971 : 8 b) follows suit and turns BÖHTLINGK's observation into a picce of evidence supporting his theory of an auxiliary vowel.* Yudhishthira Mimamsaka (1963/ saṁvat 2020: 224) thinks that the sin snasoh stands for as, as Apisali is known to have accepted s instead of as as the root.
2.2 What all these scholars do not seem to have realized is that all such comments make P a rather unpredictable whimsical fellow who would suddenly opt for an uncommon sandhi, decide to exercise unspecified prerogatives as a sutra author, or take fancy for someone else's lexical units.5 Needless to say, this is not the P that most of the A presents to us. Note
2. As is well-known to students of P, such word-indices are already available in the works of BÜHTLINGK (1887a, reprint 1964), PATHAK-CHITRAO (195), and KATRE
1968-69 ). They may have to be revised after the stage of primary textual criticism is completed, but, as the present paper will indicate, even in their present form they are very useful tools.
3. It is possible that the primary stage of the comprehensive textual study I have advocated in the first part of this contribution will make some of the following examples inappropriate. A principled consultation of manuscripts and other sources may establish that the emendation I have suggested in a certain rule is already available explicitly in the sources and is objectively the strong reading-that there is in fact no real change of reading involved, or that the objectively strong reading of sources is better than my emendation, or that P did not author the rule.
4. I intend to discuss the weaknesses of this theory elsewhere.
3. The few other cases in which, according to Yudhishthira Mimamsaka (Sauvat 2020 : 223-4), Puses a predecessor's technical or metalinguistic term, although he has a term of his own for the same, also need to be investigated.

Page #3

AKLUJKAR : THE ASTADHYAYI AS A CASE IN TEXTUAL CRITICISM
3
also that P does not shorten as to s in a rule of similar nature like ghy asor ed dhav abhyāsa-lopasca (6.4.119). In my view, the puzzle posed by śnasoh has a rather simple, usual kind of solution: The reading is a result of corruption. Is it not significant that Patañjali discusses, at least three times, the presence of t in al ( changed by sandhi to al) of 6.4.111. but not even once indicates that there is something peculiar about śnasoh? Obviously, the reading he knew was what we expect and need, namely Śnāsoh. Sometime between Patanjali and the Kasika authors this śnāsoh seems to have changed into śnasoh either in the A manuscripts or the Bhasya manuscripts.? Nor is such a change unattested in manuscripts. There are literally hundreds of instances of shortening of vowels at the hands of copyists for anyone who would care to look at or to compile critical editions. The shortening is all the more likely when the consonant sign (Tin this instance) includes a straight line and is followed by the straight line indicating a to 7 or HT to 5).
3.1 (SCHARFE 1971 : 15b :) The preceding was a case in which the awkwardness of their explanations and the possibility of a much simpler and probable explanation were not realized by scholars. In certain other cases, one's analysis must be of the more difficult stylistic kind. Consider, for example, the following rules in which the root dyś is mentioned :
1.3.57 : jñā-śru-smp-dršām sanah 3.1.47 : na dršah 3.1.137: på-ghra-dhma-dhet-dịšah sah 3.2.36 : asurya-lalatayor drsi-tapoh 3.2.60 : tyad-adişu drso 'nālocane kan ca 3.2.94 : drseh kvanip 3.4.29 : karmani drsi-vidoh sākalye 6.1.58 : srji-drśor jhaly am akiti 6.4.62 : spa-sic-siyul-tasisu bhava-karmanor upad se 'ijhana
graha-desăm và chạyad it ca 7.2.65 : vibhāşa spji-dršok
6. As stated in the first part of the present contribution, a critical edition of the Mahabhaşya based on a wider consultation of manuscripts than KIELHORN'S is a desideratum. Although the present Bhāşya editions read śnasoh in quoting or referring to P's rule, it is possible that śnosoh is the objectively stronger reading of Bhaşya manuscripts and the editors have not reported it because of the influence of other sources of the Paninian tradition on their minds.
7. The editions of Candra Vyakaraya read śnasol in 5.3.104 corresponding to P 6.4.111. If the available version of the Condra in fact antedates the Kasikii, as KIELHORN and others have argued, and if LIEBICH, OD whose edition the other editions of the Candra are based, is accurate in his reporting of the reading, then one could say that the change of sna' to śna took place in the period between Patañjali and Candra.

Page #4

Proceedings of the International Seminar on Panini
7.3.78: pā-ghrà-dhma-stha-mna-dan-dršy-arti-sarti-sada-sadam
piba-jighra-dhama-tistha-mana-yaccha-pasyarcha-dhau
siya-sidah 7.4.16 : rdso 'ni gunah
Here, 1.3.57, 3.1.47, 3.1.137, 3.2.60, 6.1.58, 6.4.62, 7.2.65 and 7.4.16 form one group, and 3.2.36, 3.4.29 and 7.3.78 form the other. The forms dịšām, dịšah and drśoh occurring in the first group clearly pre-suppose a stem drs. The stem evident in the other group, on the other hand, is drsi. Further, the first group can be divided into two sub-groups : one consisting of 1.3.57, 3.1.137, 6.1.58, 6.4.62, 7.2.65 and 7.4.16 in which dró is a member of a compound, and the other consisting of 3.1.47 and 3.2.60 in which drs is employed by itself.
3.2 Once such a stylistic analysis is attempted, two conclusions seem icevitable :
(a) P uses the stem dss at the end of a compound and the stem
drsi in a non-final position. (b) Tbe sutra 3.2.94, drseh kvanip, is odd in that its wording
does not stylistically agree with that of 3.1.47 and 3.2.60 as one would expect. Either the form drseh is later or the form dršah in 3.1.47 and 3.2.60 is later. I think the former is the case. When P has to decline a root-name or mention a single root, he uses the form ending in a consonant. For example, bhajo nvih (3.2.62), chandasi sahah (3.2.63 ). Thus, stylistic study will enable us to decide that probably the original form of 3.2.94 was drśah kvanip.
4.1 (SCHARFE 1971 : 13 b :) In some cases, a consideration of diction and the author's usual style will be a mere beginning. Textual issues that may arise will not be resolved unless observations of a subtler kind are made. I would like to clarify this statement with the examples of A 3.4.79, 4.4. 108, 6.4.67 etc. Let me begin with a general observation: P's practice is to refer to e and o by suffixing the time-marker t and to refer to ai and au without affixing any such marker. As evidence, I cite :
(a) 1.1.11 : id-ud-ed dvi-vacanam pragrhyam
3.4.90 : am etah 3.4.93 : eta ai
8. The obvious exceptions are roots which do not end in a consonant and to which no marker is affixed, e.g., i is mentioned as el in P 3.3.56.
9. From 1.1.11, 5.4.11, 6.4.119, and 7.2.103, in which simple e could have been employed but has not been employed, it is obvious that the addition of r is not motivated only by consideration of facility in inflection or of avoidance of obscuration through sandhi, etc.

Page #5

AKLUJKAR: THE AṢTĀDHYAYI AS A CASE IN TEXTUAL CRITICISM
3.4.96 vaito 'nyatra
4.2
6.4.119
7.2.5
7.3.103
7.4.52
8.2.81:
8.3.99
(b) 1.1.15
6.1.93:
6.3.112
7.3.71
8.2.45
8.3.20
8.4.28: (c) 3.4.93:
3.4.95
4.1.36
(d) 4.1.38 :
7.1.34:
7.2.107
(a) 3.4.79 6.4.67
kim-et-ti-avyaya-għad amv adravya-prakare ghv-asor ed dhav abhyasa-lopaś ca h-m-y-anta-kṣaṇa-svasa-jāgr-ni-sv-ed-itām bahu-vacane jhaly et
ha eti
era id bahu-vacane
(b) 4.4.108 (d) 7.1.84 7.3.118
eti sajyam aght of10
oto 'mśasoḥ
sahi-vahor od a-varnasya
otaḥ syani
od-itaś ca
oto gargyasya
upasargad anot-paraḥ11
eta ai
àta ai.
puta-krator ai ca12
manor au và
On the background of these sutras, the following sutras stand out as containing unexpected readings:
ata au nalaḥ
adasa au su-lopaś ca13
tita atmane-padanam ter e er lini
samanodare sayita o codattaḥ
diva aut
5
119 aud ac ca gheḥ14
10. Sutras 1.1.15 and 6.3.112 establish that is appended even in the nominative case and that it has not been introduced only for ease in inflection.
11. The Kasika reads upasargad bahulam. The reading in the Mahabhasya and BöHTLINGK's edition is as given here.
12. The form ais in 7.1.9 is due to the fact that a substitution for the whole grammatical element bhis is to be taught.
13. aut in 4.1.2. aus in 6.1.21 and aux in 7.1.18 are the result of special considerations. Hence they should be left out here.
14. The Bhasya seems to have read 7.3.116 to 119 (according to KIELHORN (1885: 192 and 1887: 180) 7.3.117 to 119) as one continuous sentence and then suggested yoga-vibhaga for it. In the Kasika, the yoga-vibhaga is accepted. Accordingly, aut is rule 7.3.118 for it and ac ca gheh 7.3.119.

Page #6

6
Proceedings of the International Seminar on Panini
It is obvious that a change in the wording of other rules to make the wording of these rules consistent would amount to an uneconomical and highly arbitrary procedure. Rather, it is in these rules that the possibility of emendation should be studied. Some context-specific considerations of a subtler kind indeed enable us to justify an emendation.
4.3 The rule which follows tita atmane-padānam ter e is thasah se. If the original wording of the former was-ter et, the combination in a continuous recitations of A rules would have been ef thasah Such a combination would rather easily simplify itself into e thasaḥ 16 unless a determined effort was made to pronounce the 1 of et and to follow it by a break ( virama) of some kind. We should not be surprised if in a rare instance a slip was made in the long tradition of the A and the first of the two dentals, which differ only in the feature of aspiration, was dropped.
4.4 Applying similar reasoning, we can conclude that it is probable that original reading of 4.4.108 was samanodare sayita oc codattah and that oc co° could have been inadvertently simplified into o coo.
4.5 As to 6.4.67, er lini, an original of the form el lini ( resulting by regular sandhi from et lini) should not be deemed improbable. The change of / tor and r to I noticed in the case of many sanskrit words is common knowledge.
4.6 In the case of dira aut (7.1.84), where consistency requires that t should be absent and the reading should be diva au, note the two rules which follow: pathi-mathy-rbhuk såm åt and ito 't sarva-nama-sthane. The taparakarana present in the three forms at, it, and at could have brought about a tapara-karana of au too.
4.7 Lastly, the text of 7.3.118-119, aud ac ca ghieh, seems to have been a matter of doubt at least from the time of Patanjali. It is significant that in Patañjali's improved formulation, au dic ca gheh, of the rule there is no 1 after au.17
15. By continuous recitation I mean the practice of reciting rules in the sequence in which they appear. Such recitation is done by Påniniyas even today. Given the importance of order in the application of P's rules, it follows that his rules were to be memorized in the order in which they appear. Continuous recitation would be a natural means of such memorization. Its acceptance as a practice expected by P, however, docs not imply acceptance also of the view that P foresaw freedom on the part of his readers in breaking up his sintra-patha. Although Katyayana and Patanjali have occasionally availed themselves of such freedom, it cannot be that P granted or even anticipated such freedom. .. 16. Prakrit ettha, for example, changed into ithe of modern Marathi.
17. This is according to KIELHORN's edition. In Kaiyața's comment on 7.3.118119' the reading is 'aud dic.... at least in the edition published by Motilal Banarsidass (p. 240 ).

Page #7

AKLUJKAR : THE AŞTĀDHYAYI AS A CASE IN TEXTUAL CRITICISM
7
5.1 (SCHARFE 1971 : 12 a :) Anyone who has worked on even a few Sanskrit manuscripts can testify to the inconsistency noticed in the application of sandhi rules. A scribe's not being particularly aware of the importance of consistency or being careless in his assignment is one obvious cause of this phenomenon. However, even in the case of conscientious and learned copyists, an editor has to be aware of two understandable yet mutually contrary factors. Since the common practice in Sanskrit sentence writing is to observe the rules of sandhi, scribes frequently effect sandhi where it is absent in their exemplar. On the other hand, since they are engaged in the activity of copying and can concentrate at a given moment on only the manageable units of the exemplar, they frequently end up dissolving the sandhi of the original. An awareness of this common phenomenon can be used to bring uniformity to the current sutra-patha. P's practice is to overlook the rules of sandhi when observance of sandhi will result in a misunderstanding of the grammatical element or operation being taught. This is as it should be according to common sense18 and the distinction between object language and metalanguage. Thus, when adeśas or substitutions are given, obscuring or misleading sandhi is not allowed to take place. Consequently, we read jaraya jaras anyatarasyām (7.2.101 ), not jaro 'nyatarasyām; bahuvacanasya vas-nasau (8.1.21), not vonasau. Given this procedure, the setra 6.1.63, pad-dan-no-mas-limi-nis-asan-yilsan-dosan-yakai-chakann-udannäsan chas-prabhrti su is suspect. It obscures the forms of adešas at a number of places. It does not indicate the lexical items for which substitutions are made, which is contrary to P's practice.19 It occurs after aci sirsah, which, while treated as a sutra in the Kašika, is a vårttika in the Bhasya. Furthermore, the order of points in the Bhasya comment on it is strange. The vårttika padădişu mans-prt-snunam upasankhyanam appears first, then a remark to the effect that śasprabhttisu is not necessary, and then the explanation of the varttika padădişu givep first. This indicates confusion in the Bhăşya tradition, which in turn gives rise to a doubt as to whether Patañjali looked upon 6.1.63 as a rule composed by P.
5.2 Applying the same consideration, we can determine that it could hardly have been P's intention to apply the rule 8 3.32, namo hrasvad oci namun nityam, in the metalanguage. The evidence collected by SCHARFE on p. Ila is sufficient to establish that the doubling of i in P 8.1.28 (tin atinah) found in certain editions (e.g. the editions of the Kašika by SHOBHITA MISHRA and by ARYENDRA SHARMA and others ) does not conform to P's practice. Similarly, the occurrences of trn in A 2.3.69, 3.2.135, and 6.4.11 clearly indi
18. If understanding occurs the accuracy, efficiency and efficacy of grammar will be sacrificed.
19. The rules like 2.3.64 ( Kytvo'rtha-) and 4.4.27 (ojah-saho'mbhasa), in which sandhi is effected, do not involve adešas. The change of sas to chap in 4.1.2 (svau-jas-am-aut-: SCHARFE 1971 : 12 ) and similar changes in rules listing or prescribing suffixes need further investigation.

Page #8

8
Proceedings of the International Seminar on Panini
cate that the doubling of n in 6.2.161 is due to the influence of anna which follows and could not have been a feature of P's own version of the rule.
6.1 Just as we should be aware of the role the predominant practice of sandhi plays, we should not lose sight of the fact that the A was maintained by a succession of scholars and scribes for whom memorization was a habit of the mind. Anyone who has worked with traditional Sanskrit scholars, no matter what their field of specialization, knows how much they are devoted to committing things to memory and how easily, after a certain stage of preparation, they retain and relate sentences scattered in different parts of a work. This attitude and this ability, well-cultivated in the Sanskrit tradition in general, are further emphasized in the study of P on account of the very nature of his wonderful work.20 One consequence of this fact is that a person copying a certain rule of the A is frequently likely to have a related rule at the back of his mind, which may affect the outcome.
6.2 An awareness of this possibility would prove helpful in the secondary textual criticism of the A. Thus, 4.4.136, matau ca, is problematic in that matau is used in the sense matv-arthe. 21 Since the rule teaches attaching of gha ( = iya ) suffix (to sahasra in order to derive sahasriya ), there can be no doubt that matau does not mean when matu or matup follows (i.e. when the situation is *sahasramar22 ); 'matau must meanwhere malu could have occurred when the sense of matu is present,' that is, matv-arthe 23 On the other hand, such a usage does not conform to P's practice. In several other rules in which a statement of the type Grammatical operation o' takes place, when the sense of x is present' is necessary or advantagecus, P employs compounds with x as the first member and artha as the second member. In fact, almost nowhere has P employed a technical element without the addition of artha to refer to the meaning of that technical element. Thus, we have lin-arthe ( 3.4.7). tum-arthåt (2.3.15), tum, arthe (3.4.9), krtyarthe ( 3.4.14), a-nyad-arthe ( 6.4.60), kȚtvo'rthe (2.3.64, 8.3.43 ), atasartha- (2.3.30), khal-artha- (2.3.69, 3.4.70), cvy-arthe ( 3.4.62 ), cvy-arthe su ( 3.2.56), lod-artha- (3.3.8) and nā-dhartha. (3.4.62). Moreover, matv-arthe itself has been twice used by P, once in 1.4.19 and once in the not too distant 4.4.128.24 The Kašika too seems to have read matv-arthe in 4.4.136, for
20. Since the derivation of object language forms takes place through a succes - sive application of rules and since conceptually related rules are scattered at many places in his work, one cannot understand or master P's grammar without extensive memorization.
21. Reasoning similar to the one given below would apply also to P 5.2.59, matau chah sukra-samnoh.
22. Such an expression is unlikely in the first place, since it goes against the general phenomenon in Sanskrit according to which one expects -vat after the stem sahasra ending in a.
23. Cf. Käsikā : matv-arthe ca sahasra-sabdad ghal pratyayo bhavati.
24. BÖHTLINGK's index somehow misses 4.4.128 under arthe. The rule is, however, commented upon in the Bhagya.

Page #9

AKLUJKAR THE ASTADHYAYI AS A CASE IN TEXTUAL CRITICISM
whereas it specifically comments on maye of the almost immediately following (4.4.138) rule with the remark maya iti mayaḍ-artho lakṣyate, it provides no similar gloss on matau; in fact, it contains no indication of matau, as it reads only matv-arthe. Nor is it likely that the authors of the Käsikà read matau in 4.4.136, but tried to suppress the problem such a reading presents, for in a later rule (5.2.59), the Käsikä does contain the remark matav iti matv-artha ucyate. Not providing such explanation for an earlier rule like 4.4.136 seems especially improbable. Thus, there is enough evidence to establish that matau ca was originally and up to the time of the Käsikä25 matv-arthe ca. Now, the question is: how could matv-arthe have changed to matau? It is not a change of the common transcriptional type. 26 I think the change has come about through a phenomenon described by one Sanskrit author as väsitantah-karanair lekhakais tatha likhitam. In matau bahv-aco 'n-ajiradinām (6.3.119), the form matau occurs appropriately. Under its influence, 27 first the matv-arthe of 5.2.59 seems to have been changed by scribes to matau (see note 25 above).28 Then the reading matv-arthe of 4.4.136 seems to have undergone a change to matau under the influence of both 6.3.119 and 5.2.59, or either.
7.1
Regarding 3.3.172, śaki lin ca, again, one can rationalize the use of saki for saky-arthe. One may argue, for example, as follows: P twice (6.1.81, 7.3.68) uses the expression sakyarthe meaning in the sense" is possible". Perhaps he did not wish to use a confusingly similar expression śaky-arthe meaning in the sense of [the root] sak', 'in the sense is capable". Also, the immediate context (3.3.169-174) of 3.3,172 is sufficient to indicate that saki signifies an upadhi, not an upapada. A literal interpretation of the rule when sak is upapada, lin also is used' would not have made sense anyway, as sak is not an upapada but a verbal root to which suffixes are added and as the occasion evidently was not one of deriving the optative forms of sak which have been provided for elsewhere.
7.2
However, this line of reasoning leaves P guilty of three things:
(a) He has used the locative of a technical element in a nontechnical sense, contrary to his usual procedure, (b) has relied upon contextual implication, and (c) has violated his own practice of adding artha.
25. On the other hand, P 5.2.59 (see Note 21 above) seems to have read matyarthe for matau only until the time of Patanjali.
26. I would account for maye of P 4.4.138, where one expects mayaḍarthe or mayarthe, in essentially the same manner as one outlined in the following lines.
27. The occurrence of matol in 4.2.72, 4.4.125, 5.3.65, 6.2.219 and 8.2.9 could have indirectly reinforced a form of matu in another case, namely the locative.
28. Note that matau occurs in the initial position in the rules as matv-arthe
does.

Page #10
Proceedings of the International Seminar on Panini As is evident from the expressions lin-arthe, tum-arthe, etc., P almost invariably adds artha when he intends to refer to the meaning of a grammatical element. That verbal roots are no exception to this procedure is clear frcm bhitrarthanam. (1.4.25), rucy-arthanam (1.4.33), avid arthasya (2.3.51 ), adhigartha- (2.3.52), krudha-mandarthebhyah (3.2.51), asty-arthesu (3.3.146, 3.4.65 ) and perhaps pasyarthaih (8.1.25). Besides, in 3.1.99 and 3.5.12, s aki stards for the root sak and does not mean in the meaning of (the root ] sak'. It is probable, furthermore, that the occurrence of saki in the initial position in the rules referred 10 influenced tbe initial saky-arthe or, cript tradition of the A. An explanation based on the influcnce of memo. rization thus seems better than the one amounting to rationalization given earlier. 8.1 The preceding discussion, I trust, bas served to establish the claim made at the beginning of this paper. A careful utilization of the word-indices of P's work, coupled with an awareness of the factors which are likely to bave played a role in the transmission of that work, will aid a critical student of P considerably in solving textual puzzles. A constant reference to wbat is implicit in the remarks of works such as the Bhasya and the Kasika will frequently provide corroborative evidence. It would be misdirected research if we lulled ourselves into believing that the sources and signals left to us by tbe tradition have been exhausted, if we did not realise that the traditional observations and solutions can be pressed into service in an alternative, text-critical, and historical way, and if we went on proposing idiosyncratic solutions or solutions that resemble only the cosmetic kind of traditional solutions. 29 BIBLIOGRAPHY For publication details of the works used in the preparation of this paper, see George CARDONA, Panini : A Survey of Research, the HagueParis; Mouton, 1976. Trends in Linguistics, State-of-the-Art Reports 6. trantuanang se visitadi titrasya kerin, which are i 29. By "cosmetic kind" I mean the solutions implicit in such remarks as sautrat vat or vicitra hi satrasya krtin, which are in effect statements of the problems not a real way of solving them.