Você está na página 1de 11

M.

WITZEL

ON T H E A R C H E T Y P E OF P A T A ~ J A L I ' S MAH]tBH/~.SYA

W. Rau 1 recently indicated that all available MSS of the Mah~bh~.sya go back to
ONE single archetype from which all have been copied. 2
This is evident, first of all, because of some lacunae in Vedic quotations: certain
words or syllables had been forgotten by the scribe of the archetype but were
later on restored in the margin and still later consequently entered into the text
at the wrong place (a procedure quite common in Indian text transmission): 3
Other lacunae, not expressively mentioned by W. Rau are: #044 pdpma hi >
pdpma, cir. MS S 44 p. 167; #719 su.sup~a indriyam ~ ~ MSS 44, p. 167;
#184 rk.salomakd >.rk.sakd.h alomakd.h ; ~P540yadaho r ~ ~ ahor ~
Vedic quotations lend themselves very well to such an investigation, as they
frequently could not be understood by the scribes, nor could many of them be
checked by a scholar (except for those of schools prominent in his own area).
This is especially the case with such comparatively rare schools as the Ka.tha or
Maitr~iya.ni S~kh~s: such quotations tend to live a life of their own in the written
transmission. 4
One can add that the character of writing mistakes found in the Vedic quotations
of the Mah~bh~.sya permits a first attempt to trace both time and place of its
archetype MS. s

1. For this brief investigation,6 the mere slips of writing, which could occur at
any time or place are of little consequence. 7 A number of mistakes indicate a
certain degree of interpretative correction by the scribes viz. scholars who used
the MS(S) preceding the Archetype. s
A closer look at some of the mistakes reported in Rau's monograph shows that
most of them tend to have occurred in one of the Northern post-Gupta scripts,
like the "Siddham~t.rk~", Licchavi (in Nepal), early S~rad~, etc. 9
Some peculiarities, however, may be of South Indian origin: #694 sasavdrhso
vi ~m. vire > sas.rvd.mso vi~rg,vire could be compared to the frequent interchange
in S.MSS ofra/.r; .d/t in Mbh. could be connected with the Dravidian interchange of
tenuis and media, but cf. author, 'The Paippal~da hymn to Takman' (forthcoming);
#290 tilpiajam dan.da(nam) na.dam (AV 12.2.54, AVP 17.3 5.2) > t ~ dan danatam
is probably due to interpretation (or, rather a confusion of the letters .d/t see below
w 5).

Indo-lranian Journal 29 (1986) 249-259.


9 1986 by D. Reidel Publishing Company.
250 M. WITZEL

The rest of the mistakes seems to be of Northern origin. First of all, a number
of them belong to the post-Gupta period:

2. d > o is a typical mistake for the p.rsthamdtra type of writingthe vowels: -N)61
~ ~ see notes (appendix) for the representation in Indian scripts: (1),
cf. MSS 44, p. 164; #265 *rohavai? > rohavehi (2). Similarly, the method of
writing the Anusv~ra circle (not yet a dot!) next to the Ak.sara is responsible
for the confusion, frequently found e.g. in Nepalese (Buddhist) texts (as well
as in some S. scripts), with the two Visarga circles: #024 ajah pO > ayam me p~
(3); #513 -a.h > a.rn; #633 carum. / > caruh/; cf. also: - h / > -k/: #511 md no
abhi sr~ > md no asrdk (4): perhaps because of the form of .h, but more probable
because of the following adrdk.

3. A number of cases is possible in early post-Gupta scripts, and hardly before or


after this period:
#739 ajigdrhsan > sam.]igdm,sat: a-/sa (5); the confusion of tin (6), however,
is possible in all periods;- similarly, srna/nu (7) at a time when the subscribed
-u still looked somewhat like a modern Virima: #157 ihaiva bhava ma nu gah. >
~ sma gah. (AV 5.30.1 and AVPK + Or. 9.13.1); sva > su (8) belongs to the same
time: #276 tanu.sva > tanus.u; it requires the post-Gupta form of -u/-va. Again,
pra/a- (9): #443 pratikulam > ati~ ; khya/ ]m : #329 pari khy an ? > parijman (10).

4. Another group of mistakes seems to be possible only at a slightly later period,


ca. 800 A.D:
fin occurs in Gilgit, S~rad~, Nepal in the 9th cent. cf. also Pallava (738 A.D.):
#505 pra min?] jan ~ (thus corr.) > prami.nfmi ]an~ (because of ni-j]/.nan, then >
ninjan, somehow understood as correction, > nimi]an). Similarly, the mistake
h/v (11) is possible only before 1000 A.D. in the North, (and otherwise in modern
scripts of the South): #586 harinasya raghu.syada.h > vam nasya raghus.yadah.
A confusion of m/p (12) is perhaps possible at all times, but more probable at,
e.g. ca. 800 A.D. in Nepal: #494 parur > marur. - The similarity of t/g (13) is
possible in the 9th cent. (e.g. in Nepal): #704 sutarmd.nam > sugarmd.nam; similarly
the one of]In (14) is possible, e.g. about 810 A.D. in Nepal: #365 ]agat navagat >o
]anagat. The following example is interesting: #028 atrgtsata.~ayah sapta sdkam >
san(n) uta sapta.r.saya.h sapta sdkam; here, a-/sa and tra> n(n)u point to a period
where N. post-Gupta script changed into N~gari; the remaining sata is "restored"
to sapta (15). - Similarly, the interchange of-t/-tu (16) (andau/o): #751 hiranyayf
naur acarat (PS 7.10.7 hiranyaye!) > hira.nyay~ no nayatu! (16) - v/dh: The old
dh is similar to v (17) in N.MSS before the N~gari script developped: #624 gatavdro ;
~ataa~dro; equally, e- : va (18) in #113 dtmann eva > dtmana eva. If indeed e- is
involved here, this mistake cannot have occurred much later than 1000 A.D., (i.e.
ON THE ARCHETYPE OF PATAIqJALI'S MAHABHAS.YA 251

before e- got its small downwards stroke). Finally, a change th/dh (19) is possible
during this period but not later than, e.g. a Tirhut MS of 1026 A.D., already written
in early Devan~gari: #699 samithesu >sdmidhenyo.
Most interestingly, a number of the mistakes evident from W. Rau's collection
are possible only after the development of early N~gari:

5. bh > y cannot be of Southern origin, and can occur in the North only after the
triparite type ofya had disappeared in the 7th century and had changed to the two
bar type ofya. lo This change is more probable when bh also had changed to its
Dev. type, i.e. with the upwards turned bha's, like in 1144 A.D. Benares, 1026 A.D.
Tirhut MSS. (as opposed to the forms of bha with its first stroke pointing downward
see: #461 ba-h[kam astu yadrapah. (21): (sic AVP 9.7.5) > bdh[karn astu bhadram
va0 (20).
Similarly, m/bh is possible only after ca. 1000 A.D. in N. N~gari scripts (again
with an upward bh), see: #730 stoman ]o > stobhair]O. Equally, ca > i is possible
only in N./W. N~gari (and not in S. scripts) after ca. 1000 A.D. (22): see candro >
indro in: #508 mddbhis .tvd candro v.rtrah~ > m~dbhir is..tvdindro v ~
Finally there is another peculiarity which may support this: the change from
.d> t (23): #'290 tilpi~]am, da.mJa(nam) na.dam (AV 12.2.54, AVP 17.35.2) >
t~ Mbh. For this phenomenon, cf. the Siddham~t.rkfi, S~rad~, (Bengali,
Oriya) forms. The change will have occurred at a time when post-Gupta ta changed
to its N~gari form 12 (which did not occur in S~rad~, Bengali, Oriya).
These few clear examples seem to indicate that the archetype MS was written
fairly soon after the development of the N~gari script. This could lead to a probable
homeland of the archetype in Benares, Nepal, or Gujarat. .3
More striking, however, are a few peculiarities which seem to indicate a more
western origin of the archetype MS:

6. The re-interpretation necessary to arrive at #541 yacla-hy asau matte < KS


14.5:205.3 anrtam, hi so "matte . . . . cf. MSS 44, p. 167, corresponds exactly
to one frequently found in the Kashmir MS of the Paipp. Sarhhit~ of the AV
which, in my opinion, goes back to a Devan~gari MS of ca. 1350 A.D. coming
from Gujarat; for the Sandhi, cf. Suppl. ZDMG 1985, p. 258, w (cf. also #486
mattasya; #044 a.nrtam hi).
Strangely enough, the archetype of Patai~jali's Mah~bh~.sya agrees another
time with the Kashmir version of PS, and only with the Kashmir version: In #368
nabha.., p.rthivyd.., is found while AVP (Or) 20.8.6 and AV (Saun.) 7.6.21 read
p rthivydm. IftAVP (K) indeed comes from W. India/N. Maharastra (perhaps via the
the Vijayanagara court and s then these two cases could be indicators of an
origin of the archetype of Mbh. in the area ofW. Nagaff (Gujarat, Maharastra etc.)) 4
252 M. W I T Z E L

7. The preceding paragraphs give some indication already about the time of the
archetype: The N]gari peculiarities point to a time of about 1000 A.D. The changes
ca > i, bh > ya, also bh > m, .el > t are possible only after the change from Northern
post-Gupta to early N]gari scripts.
However, the predecessor of the archetype is much older; there are two cases
where this can be shown definitely: #431 peciran which stands for AV 5.18.11,
AVP (Or) 9.18.5: apeciran, cf. MSS 44, p. 165. However, the K~w (ed. Ary.
Sharma et al., Hydarabad 1970, p. 758, w already reads peciran, with the
commentary: paceran ity etasya chdnclasam, hrasvatvam. The same is found in
Kaiyya.ta (1 lth cent., Kashmir; cf. A. Wezler, ParibbZ.sfi, p. 156), who reads in
his Pradipa: pecirann iti / pacerann iti prdpte chandase etvahrasvate kriyete (Rau,
p. 166). The change to peciran therefore must have taken place - at least in N.
India (including Kashmir) in/before the 7th century.IS
Similarly, #045 anegan menaketi > anegam ~ must have taken place before/in
the K~ik~ 6.4.120 (ed. Ary. Sharma et al., Hydarabad 1970, p., 757, w which
already reads: anegam/ menakA/anegam iti nega.h lini pus.dditvdd ah/. Haradatta,
however, still reads anegann iti . . . anegam iti tu prdyena pdt.ha.h. - The quotation
comes from MS 2.9.2:122.4, KS, TS, VS etc. anegann asya yd igava.h (while menakd ~
is from MS 2.8.10, KS 17.9 etc.). 16
The mistake existed already in/before the K~ika, i.e. in/before the 7th century.
However, if it is correct that Haradatta was a S. Indian (and not a Kashmiri) of ca.
the 10th century, then this agrees with the identification of the archetype as based
on a Northern tradition. Haradatta may have had access to other, unknown or lost
SOHrCes. 17

8. As for the place where the archetype was copied, the above notes indicate a
Northern area: The post-Gupta type script on which the mistakes in the Vedic
quotations are based must have been a Northern alphabet. I mention especially:
bh > y is not Southern but Northern (only possible in the 7th century and later),
see #46; equally: rn > bh, see; #730;ca > i, see #508. The strange correspondences
with the sub-archetype of the Paippal~da Sarhhit~ (MS *D, in Western N~gari,
from Gujarat/Maharastra) copied from earlier MSS around 1350 A.D. (see above,
w may further define the area of the archetype as belonging to Western India.
This is, however, only a tentative proposal: more evidence will have to be
brought forward. If the suggestion is correct, one could think of the multifarious
activities of king Bhoja of DhS_ra, under whose reign various commentaries were
written, like the comm. on the VS and the RV and V~j. Prat. by Uvata or the
Kau~ika-Sfitra comm. by Ke~ava.18

9. This result can now be compared with the tradition that the MBh. was transmitted
ON THE ARCHETYPE OF PATAlqJALI'S MAHA.BHA.SYA 253

in one area only at Candragomin's time (5th cent. A.D.), as is reported in a verse
from the second book of Bhartrhari's V~kyapadiya. 19
The South Indian tradition had been interrupted and was revived only by
Candragomin who relied on the N. tradition "from the mountain", which Scharfe
takes as the Citrakfi.ta/R~magiri, S. of Benares. Candra is the one who starts a
new Mahfibh~.sya tradition in Kashmir as well, at least according to Kalha.na
(1148/9 A.D.), R~jataraflgi.ni 1.176: where he speaks about a reintroduction under
the vague king Abhimanyu I, (who reigned right after the Kus~nas Hu.ska, Ju.ska,
Kani.ska); then a new grammar was composed by Candra and others. - 4.488:
another restoration of an interrupted study under King Jayapi.da (and cf. the
comparison of Jaypapi.da to Patafijali 4.6.36 (see comm. by Sir M. A. Stein).
It probably will be difficult to show that some changes in the tradition before
a late/post Gupta hyper-archetype go back to one or the other form of Brahmi
or Gupta script prevalent in these areas, but the material presented by W. Rau
and discussed here will perhaps draw someone's attention to this possibility?~

10. The discussion of the materials presented here presents implications for several
areas of Indian philology:
First of all, it shows that we could make a much more progress in the tracing
the history of a particular text is we only had:
(a) a proper edition, preferably with sterna eddieum, or at least all the variants
of the MSS available to an editor;
(b) a paleography of the various post-Gupta 'Siddham~trikfi', early N~gari,
and the regional S. Indian scripts (preferably until ca. 1600 A.D.). The problem
is, of course, that we have only a limited amount of manuscript material from
the late 1st and the early 2nd millenium A.D., i.e. except for Sinkiang (which
can be excluded here for all practical purposes), mostly from Nepal, Benares
and surroundings (found in the Nepal National Archives), as well as a few from
Gujarat (Jaina texts). For additional evidence one has to take into account the
inscriptions - which usually are much more conservative in the form of the letters
they use than the MSS.
(c) more knowledge about local habits of pronunciation, preferably collected
from Vedic and other recitation. Local pronunciation frequently is the cause of
writing mistakes.
Ultimately we will need an atlas of the phonetic and scriptural mistakes which
will allow to trace exactly the history of transmission of a particular text.2x
However, even now a few statements can be made:
(1) One should not accept Patafijali's text as it stands now as one of ca. 150 B.C.
but as one of ca. 1000 A.D. It has to be counterchecked against evidence from
other grammatical texts (which are, nora bene, themselves not critically edited, or
254 M. WITZEL

have been "corrected" according to the Mbh. tradition the commentator belonged
to or had before him in MS form; modern editions of such (sub)commentaries
frequently follow Kielhorn's Mah~bhfi.sya edition!)
(2). The situation is similar in the field of Vedic studies: the history of each
Vedic quotation in grammatical or philosophical literature has to be investigated,
as per (sub)school and/or local tradition. (Otherwise, there will be a danger of
circular argumentation.) Obviously, this is best done by compiling the material
by computer. 22
(3) It is important for the reconstruction of old texts, like theiPaipp. S., that
Pat.'s quotations cannot be accepted as primafacie evidence when reconstructing
a text for a period of ca. 150 B.C. Each quotation has to be counterchecked,
if possible, against other evidence.
(4) It is my suspiction that most of our well-known "classical" texts go back
to manuscript(s) (archetypes) of ca. 1000 A.D. only, i.e. to MSS written after
the emergence of the various (sub)regional scripts at ca. 1000 A.D.: The earlier
post-Gupta MSS must have been rewritten sooner or later after ca. 1000 A.D. in
the new type(s) of script; cf. already my lecture at the TObingen Or. Conf., in:
Suppl. ZDMG 1985, p. 256 sqq., or cf. for example a Manu-Smrti MS from Benares,
written under Jayacandradeva in 1181 A.D. (Kesar Library, Kathmandu), which
already follows the Vulgate version (and not Bh~ravi's text).
The situation may vary slightly for texts which rest on a purely local tradition
only, as for example many of the Kashmirian texts: there, the slow changes in
the development of the S~radg script do not create the same amount of mistakes
nor the same degree of corruption as found in the "mainstream" N~gari traditions
in which a large number of mistakes, if not the majority of the corruptions,!occurred
during the transcribing process from post-Gupta scripts to N~gari. - In such cases,
it depends very much on the degree and speed of change the script in question
shows when compared to its Gupta time parent. Investigations into this process
are necessary and would not only provide useful tools for the philologist but,
indirectly, also better editions.

APPENDIX

11. The collection of Patafijali's Vedic quotations compiled by W. Rau allows


to add a few remarks on his probable home. However, in order to do so, an
investigation into the homelands of the various Vedic tests and schools is necessary
as a precondition. This has been done elsewhere.23
The quotations expressively mentioned by W. Rau as being "hapax", i.e.
traceable, according to our present state of/knowledge, in one Vedic text only
(see Rau, p. 102), provide the following picture: 24
ON THE ARCHETYPE OF PATANJALI'S MAH.A_BHA.SYA 255

WESTERN texts CENTRAL EASTERN

AB 3
KS 39 {RV51} TS 17 KB 1 -VS 3

Patafijali
KpS3 ~ ' ~ 3 {RVkh3} TB2 = 23 KA 1 (SBK 2) =8
/ SB1 {AVP 16} TA4 ~s 2 ~B 5
MS 13 Chal {GB 2}
MU 1 {AVS 4}
JB 2

The texts have been roughly arranged here in their geographical position. The
RV and AV were, of course, known everywhere. But, as is well known since P.
Thieme'sPdniniand the Veda (1935, p. 63, cf. Rau p. 103), Pat. I:1.4 quotes the
beginning of the PS, and this text is, at least in the early period, apparently more
Western than the Vulgate (AV-Saunaka). 2s
The rest of the quotations shows, as Thieme and Rau (p. 102-3) also state,
the Patafijali concentrates on KS, TS/TB/T.~ and also MS; (especially, Rau adds,
on KS and PS). 26
The home of the Ka.tha school is the Eastern Panjab ("where the rivers flow
westwards"), that of the Maitr. school the S. Kuruksetra/R~jasthfin/Madhya
Pradesh area (perhaps stretching up to Gujarat), and that of the Taittiriya school
is the Paficfila country (modern U.P.). The areh where these three schools intersect,
e.g. Mathur~, has been identified as Pata~jali's home by P. Thieme (in his
classes, 1966, according to Patafijali's treatment of sibilants in popular words);
similarly, H. Scharfe, Gramm. Lit. (Wiesbaden 1977), p. 153.27
This investigation could be further extended by comparing such quotations
as #123 dmbdndm carum where Pat. sides with KS, TS, SBK as against MS, SBM,
cf. MSS 44, p. 164. This would, however, lead too far here.
An interesting item, finally, is the long quotation from TS 2.5.1.6, given by
Pat. at 2.3.62:466.10 (#556, cf. MSS 44, p. 161) which can indicate that in Pat.'s
time TS perhaps was recited without anuvdka divisions still: the border between
anuv. 5 and 6 has been eliminated by Sandhi in Patafijah.

NOTES

1 Die vedischen Zitate ira Vydkaran.a.Mahdbh(sya, Akad. d. Wiss. Mainz, 1985 No. 4, p. 101
sq. - All quotations marked # are from Rau's book. Cf. also: W. Rau, MSS 44, p. 162 sqq.
where a number of quotations are treated more according to their interest for restoring Vedic
passages. - (Another selflessand very useful work is his bibliographic collection regarding word
256 M. WITZEL

indexes to Vedic texts which supplement the Vedic Word Concordance of Vishva Bandhu,
see: Kratylos, Jahrg. 29 (1983), Wiesbaden 1984, pp. 1-25.)
2 This is not so strange as it may appear. Even a Vedic text, the Paipp. Sarith., goes back to
one archetype MS; I suspect the same for the Saunaka version; the Maitr. transmission equally
is very narrow and the Ka.tha Safiahit~ has survived in 2 - 3 closely related MSS only. It is also
well known that medieval scholars sometimes took great pains to collect good MSS of a text
and that the fame of a particular MS was such that it was copied time and again, thus becoming
the ancestor of a new, later perhaps the only surviving line of transmission.
3 This shows that - at least this part of Patafijali's text - is not in such an "excellent condition"
as Scharfe thinks, see Gramm. Lit., Wiesbaden1977, p. 154; similarly, Kielhorn IA 5, p. 242 =
K1. Schr. p. 170, but see W. Rau, Die red. Zitate . . . . p. 102.
4 Cf. already M. Miiller in the introduction to his edition to the RV (Preface to the 5th vol.
of the 1st ed.), London 1892, p. C sqq.; cf. also Jayanta Bha.t.ta's Veda quotations, in: author,
The Veda in Kashmir (forthcoming). - D. V. Garge, Citations in Sdbara.Bh(tsya, Poona, 1952.
s For a similar exercise, see Suppl. Bd., ZDMG, 22. Deutscher Orient. Tag Tfibingen, March
1983, Wiesbaden 1985, pp. 256 sqq.
6 I add here all references from the Paippal~da Safiahit~ of the Atharvaveda (PS) as known to
me from my photocopy of MS P (=Or).
7 #066, #346 a > d; #624 gd > ka; - #535 -d > d m - #695 sa saveclayuh. > sam s ~ cf. MSS 44,
p. 1 6 6 ; - #368 p.rthivy~Pat, and AVP(K)I prthivydm A V(S) 7.62.1, AVP(Or) 2 0 . 8 . 6 ; - #505
-r~m > -ra; - #086 read with MSS K, D: asyai tram > td. Cf. also: #057 *ty#.h > nah; but cf.
Vas DhS, Rau p. 78: na.h in a quotation from the "K~thaka"; #360 ka > ku; #395 pa/pu, cf.
MSS 44, 165; #251 ca > cu; of more interest are: #321 dadbhih> adbhih.; #505 pra minU
]anO(sic) > praminrmi]an ~ (because of a confusion between]/n), see below.
8 -m./-n (with the usual confusion of .rn./-n when written by Anusv~ra); #045 anegan menaketi >
ane~am ~ (note: before Kayya.ta, 1 lth cent.; Haradatta has: a n e g a n n . . . ; K~ik~ ed. Ary. Sharma
has anegam/menakd /. #541 yad~ hi so 'matto bhavaty atha tam tat tapat~" > y a d d hy asau
matto bhavati atha y a t tapatf; this requires a more complicated interpretation, cf. below: -
#220 ~ te > ~ se; #562 y e anti y e ca darake > antiye ca dftrake (accent!); - #372 nirv[ratam
vai purus, o y a m o ]dta d ~ s t e apakutdm gau.h > nirv[ryatam vai ya]amdna ~ ; - #636 ~fryate >
krfr yago; #414 ~yeti > ~ #207 kim > kam; #916 vandni > t.rndni; #508 m~dbhis tvd
candro vrtrahd > mddbhir istvdindro v~ ; #525 MSS corr. ya, ed. u!
9 Compare, for the time being, the tables of Bfihler's and other paleographies as well as the
MSS mentioned in my Tiibingen lecture (see ann. 5). There, and here, - I frequently will refer
to Nepalese scripts and MSS as it is, apart from Sinkiang, only here that old Sanskrit MSS have
survived, alphabets of which I collected for the purpose of investigating various archetypes of
texts. (Among the oldest dated MSS from Nepal so far is the Skandapur~.na of M~nadeva
Samvat 252 = 810 A.D.). Note: the collection of such materials is preliminary, and can be
judged only when all the major peculiarities, through the centuries, of various areas of India
have been collected. I know more of the Northern alphabets and may have overlooked some
typically Southern developments. Therefore, a countercheck by a specialist of early Southern
alphabets would be welcome!
lo The last occurence seems to be that in a Rajasthan inscription of 679/80, the Dhulev plate
of Bha.t.fi, see P. G. Pal, Early Sculpture o f Kashmir, Leiden diss. 1986, p. 186, ann. 69.
11 Cf. also K. C. Chatterji in V~ik 4, p. 87 sq. C tvd, tvtfl).
12 Also, ks > k t is possible around ca. 1000 A.D. (24).
ta So far I do not find examples of mistakes based on typically Eastern N~gari.
14 Note that no typical Kashmir mistakes have been found yet (t/bh, e/L o/u etc.,) in the
archetype of Mbh.
15 Probably because of the preceding lini which had presented problems in the ligature lihyap ~ .
Note that AVP(Or) reads correctly apeciran while AVP(K) has upecaram. This text was
re-introduced into Kashmir only in 1419 ; at Kaiyya.ta's time the Kashmir PS still had a different
ON THE A R C H E T Y P E OF P A T A N J A L I ' S M A H A B H A S Y A 257

reading (before the text was lost during the Muslim persecutions around the end of the 14th
cent.).
16 Haradatta is dated as being not later than the first half of the tenth century, see: P. S.
Sharma, Aligarh Journ. of Or. Stud. II (1985), 75, according to whom he lived in Coladew
on the Kgveri. He especially quotes TaRt. texts but also knows the SwidhySya-Br. of the lost
Ka.thB (where a typ. Taitt. phenomenon, the change of gn > w does not take place).
17 It would be useful to countercheck this by referring to other mistakes in Haradatta (and
other MSS, e.g. early Candra MSS in Nepal.).
18 Otherwise one could think of the Caulukyas of Gujarat or Of a Maharastra author.
19 See Scharfe, JAOS 96, p. 274/6.
20 It is striking that W. Rau apparently was aware of the possibility to use paleographic
reasoning for attaining a better text, for in MSS 41, p. 177 he uses this argument for correcting
Vedic passages; in the present case, however, he concentrated on the quotations themselves.
21 I have been eoUecting such materials, but under present conditions in Holland, with ever
changing pressures (of fashion) on Sanskrit (as well as many other subjects) by the Ministry
of Education, the Board of the University and even the Faculty, there is no hope for a speedy
conclusion of this or other larger works. Time allowing, I will present a preliminary survey
shortly. - Cooperation in this field would be appreciated, as a single person can hardly
overlook or handle all the materials even when the computer is used.
22 This applies to other disciplines like Dharma texts, Tantra, Medicine, etc. as well.
23 See author, 'On the localisation of Vedic texts and schools', Materials on Vedic S~ikh~is,
No. 7, forthcoming (Fel. Vol. Eggermont, ed. by G. Poller, Louvain).
24 The hapax Atharvaveda quotations mentioned by W. Rau are AVP 12: AV 8: GB 2. This
has to be slightly modified, as indicated in the diagram: Of the 8 AV (Saun.) hapax quotations,
#140 AV 15.14.4 is found with the same text in PS 18.41.4, a book not available in AVP(K);
the same applies to #058 AV 18.4.17 = PS 18.77.7; - similarly #624 AV 19.36.5 = PS 2.27.5
(where only the line in question is missing in K); - #360 AV 13.1.56, 57 corresponds to PS
(Or) 20.28.3, in the K version, however, to PS(K) 20.27.5, as P~das d of verses 3 and 5 are
exchanged here. Only 4 quotations remain for the time being: #283 AV 8.10.21 similar to
PS(K) 16.134.8; #064 AV 5.31.4 sim. to PS(K) 16.36.8; #749 AV 5.4.5. sire. to PS 19.8.13;
#373 AV 3.26.5 (nilimpd) sim. to PS 3.11.5 (vilimpa-). - Note that GB probably is nothing
but an Anubra-hmana of the Paipp. school, attached to the Vulgate/Saunaka school only in
medieval Gujarat.
25 See 'On the l o c a t i o n . . . ' .
26 The Ka.tha quotations can be expanded as well, as TB 3.10-12, TA 1 - 2 and parts of TA 10
correspond to Ka.tha chapters which are partly found in the .Rcakas, and partly are not yet
available.
27 On the development of sibilants in this dialect, see however, T. Damsteegt, Epigraphical
Hybrid Sanskrit, Leiden 1978, p. 44 sqq.

APPENDIX

(1) : ~ ' , l l , I ~

(3) ~ l p , . ~ . ~ ~ > t{74-~.~i ~

(5) ~ ~ ~.1
(6) I - ~ , a" - ~r
258 M. WITZEL

(7) -t
(8) l- t.tt
(9) ?.~~~
(lO) 1"-/I - ~1
(11) g : ~ ' , ~" :~1",
Benares Tirhut Nepal
1182 A.D. I080 A.D. 810 A.D.
(12) :~ - "El , l i
(13) ~ ~ ~g[
(14) ~ -
(15) :~ ~~kl, ~ ~ ~ Nep. 810
~[" ~ ~ , ~1" ~ 7 Ben. 1026, 1180

9 ~ nil ~ ~, - "~r~l
(17) q : ~l" Nep. 810
"~l "~[ Tir. 1080
~1 : ~1" Ben. 1180
(18) ~I :~' Nep.
: ~l TiE
I t ."~1" Ben.
~I,W, ~" Dev.
(19) Ii~ , "ttl
: ~-I Dev.
tl" : ~ Dev.
(20) ~ : ~ (Post) Gupta
: ~ Nep.

Tir. Ben. Ben.


1080 1144 1182
(21)),l ~II
ON THE ARCHETYPE OF PATANJALI'S MAHNBI-I/~.SYA 259

(22) *j* > ~ : ~1" Post. Gupta


~ :~ Ben. l142,1182
Guj. 1121
(23) t ~ ~ Nep.
~" ," ~ Tir.
: ~ Ben.
(24) ~ ~. ~ :

Você também pode gostar