The Original Language of the Karpura-manjan
By Richard Salomon, Seattle, WA
I. Introduction: The Problem of the Language
1. Räjasekhara' s Karpura-manjan (henceforth KM), written in the
late 9th or early 10th century A.D., is one ofthe most important works
of Prakrit literature. It is the first drama known to have been composed
entirely in Prakrit, and the only such one of any consequence.
Moreover, its essential importance is enhanced by the edition and
translation with glossary and extensive notes by Sten Konow and C.
R. Lanman. ' This edition has served as an introductory Prakrit text for
generations of American Indologists (and, I believe, for others as well),
and it is therefore all the more regrettable that, as will be shown below,
it presents a highly distorted view of the Prakrit dialects. Subsequent
editions, most notably that of Manomohan Ghosh, ^ have been impro¬
vements, but none is entirely satisfactory.
2. Specifically, the long-standing question of which of the several
Prakrit dialects the KM was composed in has never been conclusively
answered. Despite the acknowledged literary and linguistic importance
of the text, one may read, in various studies of Indian or Middle Indo-
Aryan literature, that the KM was written in Sauraseni,' in Sauraseni
and Mahärästri," or in Avanti Prakrit.' Other writers decline to take a
' Räja-9ekhara's Karpüra-manjari. Ed. Sten Konow and tr. Charles
Rockwell Lanman. Cambridge: Harvard Univ. 1901. (Harvard Oriental
Series. 4.); Repr. Delhi: Motilal Banarsidass 1963.
^ RäjaÄekhara's Karpüramanjari. A Prakrit Play. Calcutta: The World Pr.
Private Ltd. 1972. Other recent editions are those of R. P. Poddar: An Intro¬
duction to Karpüramanjari. Vaishali: Research Inst, of Prakrit, Jainology, and
Ahimsa 1974. (Prakrit Jain Institute Research Publication Series. 2.) with
Ghosh's text; and Rämkumär Ächärya's Karpüramanjari. Varanasi: Chowk¬
hamba Vidyäbhavan, 1963. (Vidyäbhavan Sanskrit Granthamala. 12.)
' ViSvANÄTH Bandyopädhyäy: Poll o Präkrt Sähityer Itihäs. Kalikätä:
Särasvat Laibreri [B.S.] 1379, p. 179.
" Syämä Varmä: Acärya Räjasekhara [Hindi] . Bhopal: Madhya Pradeä Hindi
Granth Akademi 1971, p. 139.
^ Surendranath Majumdar Sastri: Avanti Präkrit of the Karpüramanjari.
In: Indian Antiquary 50 (1921), pp. 80-2.
stand on the issue,* or simply ignore it, stating that the text is "in
Prakrit."' The problem, moreover, is of consequence not only for the
study of the KM itself but also for Prakrit literature in general; and
even more broadly for Middle Indo-Aryan dialectology. Last but not
least, a close re-examination ofthe problem ofthe KM's language(s)
tums out to have considerable bearing on the difficult problem of estab¬
lishing reliable standards for the editing of Prakrit manuscripts.
3. The editions ofthe KM (especially Konow's) have been more or
less marred by preconceived notions on the language question, which
have led the editors to make arbitrary and unjustified emendations to
the text, or to overlook important data contradicting their preconceived
notions. The result (again, particularly in the case of the Konow
edition) has been to produce a text which is no more than the outcome of
a self-fulfilling prophecy. The editor who sets out with the notion that the
text is in Öaurasem prose and Mahärästri verse does indeed produce
such a text; but this text is his ovra, and not the manuscripts' . Thus such
preconceived ideas on the language question have obscured the real
data. Only a fresh and impartial look at the evidence of the manuscripts
themselves can be expected to produce a better answer to the problem
ofthe original language ofthe KM. The results of such an attempt will
be presented in this paper.
II. Previous Views on the Language of the KM, and the
Editions Based Thereon
A. Konow
4. As observed above, Konow and Lanman produced their edition
(footnote 1) without ever questioning that the standard formula for the
dramatic uses of the Prakrits in the Sanskrit drama—i.e. Sauraseni for
prose and Mähärästri for verse—applied to the KM as well. (See for
instance pp. xxii and 200.) Konow confesses in his editor's Preface
(p. xxii) that in order to fit this formula he has "in some places, intro¬
duced the peculiar forms of the two dialects, even against the reading of
all manuscripts." The necessity of such manipulations, however, does
' Sures Chandra Baner.ii: A Companion to Middle. Indo-Aryan Literature.
Calcutta: K. L. Mukhopadhyay 1977, p. 81.
' M. Winternitz: History of Indian Literature. Engl. tr. by Subhadra Jha.
Delhi: Motilal Banarsidass 1963, Vol. Ill, part 1, p. 268; M. Krishna-
machariar; History of Classical Sanskrit Literature. Repr. Delhi: Motilal Banar¬
sidass 1970, p. 630. Achärya, in the introduction to his edition (op.cit., foot¬
note 2) says that the text is in S ( Yah nätak Sauraseni Präkrt me Ukhä gayä hai, p. 5).
The Original Language of the Karpura-manjan 121
not seem to have shaken his faith in the traditional doetrine; but, as we
will see shortly, it should have.
5. The problem is that Konow's "some places" turn out to be many
indeed. One need not search far into the text to see this; I will begin with
the very first verse (1. 1). The Konow edition reads as follows:
Bhaddarn hou sarassala kaino nandantu väsäino
annänani pi pararn paattaü vara väni chaillappiä /
Vacchomi taha mäaht phurau no sä kim ca pancäliä
r%to vilihantu Icavvakusalä jonharji caorä via //
6. As it stands, this is good Mähärästri, with the characteristic elision
of all Sanskrit intervocalic t-s; as, in Konow's view, all the verses must
be. But a look at the variant readings yields some surprising results.
For hou, (Sanskrit bhavatu), we see that 8 mss. read bhodu and 2 hodu,
with Skt. intervocalic t becoming d, the hallmark of Sauraseni Prakrit.
What is remarkable here is that these 10 readings comprise all of
Konow's mss.; in other words, Konow himself emended the Sauraseni
(b)hodu to Mähärästri hou without any manuscript justification. His
"some places" begin with the second word of the text.
7. This is only the begirming: for paattaü (Skt. pravartatäm) in
Konow's text, we find in the variants 5 mss. with paattadu, 3 with
pavattadu, and one each peyattadu and yaattadu. Again, all the mss.
have the Sauraseni ending -du, for which Konow substitutes the
Mähärästri -u. Further: for Konow's phurau. (sphuratu), 7 of 10 mss.
have forms in -du; and for mo, 8 of 10 have ndmu or rtdio.
8. Thus in four cases of Skt. intervocalic t in the first verse, the 10
mss. have altogether 35 times d, the typical S development, and only
5 times the M elision. (The case of sarassai = sarasvati in the first line
will be treated separately below, footnote 12). How can this verse be
considered M, if one gives any weight at all to the testimony ofthe mss.
themselves? (Even the exceptional M readings can be shown to be non-
original; see § 39-42.)
9. Nor is this an isolated case; indeed it is more often than not the
pattem in the majority of verses throughout Konow's text. To take only
one more of the many examples: in verse 1. 11, Konow's text has kiim,
paunjaium, mm, and icchai for Skt. krtim, prayojayitum, imam, and
icchati; all proper M forms. But again, the variant readings tell a very
different story. For A;Mm of the text, we find "Mss. kidi-." For paunjaium,
we see paurnjaidum and various similar forms in -idum (except for one
obvious scribal error) in a total of 10 mss., or all of them; for mm ofthe
text, edam (once etam) in all 10 mss. again; and for icchai, forms in -diov
-ti in 6 of 10.
10. The conclusion is quite simple, and unavoidable. The two verses
cited, and most (but not all; see § 28-29) ofthe verses ofthe text are in
Ö Pkt.; the M is simply the product of Konow's wholesale arbitrary
emendation. To give an idea of the amount of distortion in Konow's
edition, a count ofthe first act only shows 16 cases where the notes
give a variant labelled "mss.," i.e. where Konow went against the
unanimous readings of all the mss.; and no less than 32 more places
where the number of variants for a given reading total all the mss., so
that again the text reading is an editor's emendation. In all, at least 48
emendations in the first act alone; Konow's "some places" thus tums
out to be quite an understatement.
11. This is still not the whole story. In many other cases, Konow has
chosen verse readings from a minority of the mss., which will be seen to
fall into a significant pattem. Here verse 11.46 may be taken as an
example. Konow reads:
Tikkhänam taralänam kajjalakaläsanivattiänarn pi se
päse pancasaram silimuhadhararfi niccam kurmntänam a /
nettänam tilaaddume nivadia dhädi maacchia jam
tam so manjaripunjadanturasiro romancio wa tthio //
As usual, a perfect M verse on the surface. But once again it is inter¬
esting to see where Konow gets the M forms samvattiänam {samvarti-
tänäm), nivatjtiä {nipatitä) , romafwio {romäncitah) and tthio (sthitah). The
first is pure emendation, such as we have seen above; all the mss. have
samvattidänam or various other readings with the Ö ending -idänam.
For the other three words, however, the pattem is different. For each of
them, 6 of 9 mss.* have a S form (nivadidä, etc.; romancido; and tthido or
vattadi) ; while the M form for each of the words is found in the same 3
mss., S, T, and U, which constitute the southern recension ofthe text
(see Konow's Critical Account of the Manuscripts, pp. xxv-xxvi).
12. This pattem of M readings in the three southem mss. versus S in
the others recurs constantly throughout the text; see for instance 1.29,
II. 4, and III. 3 for particularly striking examples. As will be shown in
more detail below (§ 40), this southern recension shows a very decided
tendency toward M forms not only in the verses but even in the prose,
where, as all would agree, they cannot be correct. For this and other
reasons (below, § 39-42) the southem recension represented by mss.
STU is the less reliable one, and those verses whose readings are Ö in all
mss. except STU may be safely considered to have been originally
composed in
' The total number of mss. varies from place to place in the text, as several of Konow's mss. were incomplete.
The Original Language of the Karpura-manjan 123
13. Konow, however, has heavily relied on STU for his readings of
the verses, as they fit his preconeeived notion that all the verses must
be in M. On the other hand, in the prose dialogues he normally takes the
readings of the northern recension, which have the correct S forms.
Konow's edition, in short, is a hybrid. The prose portions represent
essentially the northern or § recension, while the verses are from the
southem or M recension; and both, as we have seen, are heavily loaded
with Konow's own emendations where none ofthe mss. provided the
readings he wanted.
14. There are problems in the prose portions of the text as well.
Konow notes (p. 202) that certain S words are consistently written in
all the mss. in what he considers to be incorrect forms; thus raana
( ratna) , sarassai (sarasvati) , mihuna (mithuna) , and so on. These he does
not hesitate to emend to the supposedly correct S forms radana, saras-
sadi, and midhuna. Referring to these and other supposed violations of
the mles for § given by the (very late!) Prakrit grammarian Märkan¬
deya (pp. 202-3), Konow concludes that "Räjagekhara's linguistic skill
was not so remarkable as he likes to tell us" (p. 203). In other words,
according to Konow, the (to him) disturbing violations of the gramma¬
rian's mles in the text ofthe Km are simply due to Räjasekhara's ignor¬
ance of Prakrit—a remarkable charge' against one ofthe major figures
of traditional Indian literature,'" and above all the one who raised
Prakrit to a new level of importance as an independent vehicle for the
drama. And even if Räjasekhara's Prakrit was "wrong" (which, as will
be shown shortly, § 19ff., it was not), on what grounds can an editor
correct the author himselfl An editor's task is to reconstmct the original
text as the author wrote it—not as the editor thinks the author should
have written it."
15. But emend the prose passages Konow does, against all the mss.
and, almost certainly, against what Räjasekhara himself wrote. Thus
we find in the prose text of Konow's edition 7 times maragada- (Skt.
' A charge which is nevertheless uncritically echoed by A. Bbrribdale
Keith: The Sanskrit Drama. Oxford: Univ. 1924, p. 236: "He cannot distinguish accurately Qauraseni and Mähärästri in his drama."
'" It is true, however, that at the time of Konow's edition Räjasekhara's great
critical work, the Kävya-mimämsä, had not yet been discovered, so that his
stature in the field of literature was not as clear as it is now.
'' Even Kälidäsa, as the commentators like to point out, violates the rules of
Sanskrit grammar from time to time. But would any editor ever suggest, for
instance, that lohavat in Kumärasambhava 11.59 should be emended because it
violates Panini V.l. 115, tena tulyam kriyä ced vatih (Mallinätha, vatipratyayo mrgyah)^
marakata) , a spelling which occurs only twice (in the same ms.) , while the
word appears 50 times in the mss. as maragaa-. Other of Konow's
"Sauraseni" spellings never occur at all: cadutthi (11.6.23, 29.5) is
always cautthl in the mss.; midhuna (111.9.2) is an emendation for
mihuna; idha is for iha; and many more.'^
16. It is thus clear that Konow feels the rule that Skt. -t- becomes -d-
in S is sacred and inviolable, and that words like maragaa or cautthl
simply show Räjasekhara's ignorance. Likewise, he thinks that the
grammarians' rule -th —► -dh- is invariable, so that mihuna cannot be
right. The fact, however, is that these phonetic rules are general only,
and are subject to numerous exceptions; the grammarians need not be
taken literally when they prescribe these as fixed rules (and in fact they
do not always agree among themselves on these points).
17. This practice of editing (and emending) Prakrit texts strictly
according to the rules of the grammarians was much in vogue among
earlier westem Indologists, notably Richard Pischel, who took as his
axiom that "the grammarians are not to be corrected with the help ofthe
MSS ; but the MSS are to be improved upon with the help of the gramma¬
rians."" More recently, however, it has been shown by Johannes
Hertel'" and Ghosh" that such a "blinde Vertrauen auf die heimische
Grammatik für die Auffassung der Prakrits und für ihre Gestaltung im
Drama"'* is no longer valid. For since Pischel's time, new forms of
early dramatic Pkt.— most notably those ofthe Central Asian fragments
of Asvaghosa's plays"—have been discovered which show that the
'^ Sarassai, as noted above (§ 14), is one of these S words which elides Skt. -t-.
It occurs 3 times in prose dialogue, spelled sarassai in all mss. in each case. Thus the form sarassai in verse 1.1. (discussed above, § 8) does not argue against the
position that this and other verses were composed in as this is clearly a
permissible S form.
Comparative Grammar of the Präkrit Languages. English tr. by Subhadra
Jha. Delhi: Motilal Banarsidass 1965, § 42. (Hereafter referred to as
'Pischel.')
'"" Mundaka Upanisad: Kritische Ausgabe . .. Leipzig: H. Haessel 1924. (Indo-
Iranische Quellen und Forschungen. 3.), pp. 7-8.
'^ Op.cit. (footnote 2), p. 29.
" Hebtel, op.cit. (footnote 14) p. 7. It is ironic, however, that this same
editor has wrongly emended such non-Päninian, but original forms as kämabhih
and samdhayita in the Upanisad. See my Unguistic Analysis of the Mundaka
Upanisad. In: Wiener Zeitschrift für die Kunde Südasiens 25 (1981), pp. 94-6.
" See Heinrich Li'ders: Bruchstücke buddhistischer Dramen. Berlin: Georg
Reimer 1911. (Königlich Preussische Turfan-Expeditionen. Kleinere Sanskrit-
Texte. Heft 1.)
m
The Original Language of the Karpura-manjan 125
grammarians' picture of the dialects was incomplete and therefore
unauthoritative.
18. For the present context, however, there is an even more relevant
and decisive source, in the inscribed Skt. dramas of the medieval
period, most notably the Pärijäta-manjari-nätikd}^ and the Lalita-
vigraharäja-nätaka. '' These texts were inscribed on stone tablets at the
time of their composition by the court poets, to be preserved in the royal
palaces. Like most Skt. dramas, they contain large portions of S and M
Pkt. Their special value for Pkt. studies (which up to now seems not to
have been fully realized) lies not in their literary qualities, but in the fact
that we have in them original Pkt. texts, exactly as they were composed
in the medieval period (12-13th centuries), without any suspicion of
scribal changes in the process of transmission. They are, in effect, auto¬
graph mss. of the texts, and show exactly what literary S and M were in
these times, regardless of what the grammarians, contemporary or
later, might say.^°
19. A study of the language of these inscriptional dramas confirms the
conclusions to be drawn from an impartial analysis of the mss. of the
KM. Specifically, Skt. -t- is sporadically elided in certain S words in
these dramas, and in many cases these are the same words where this
phenomenon was observed in the KM. Thus for instance in the Pärijäta-
manjari we find the following prose (S) forms^':
cautthie (line 22): as in KM (above, § 15).
Ed. E. Hultzsch in: Epigraphia Indica 8 (1905-6), pp. 96-122.
" Ed. F. Kielhorn in: Nachrichten von der Königlichen GeseUschaft der
Wissenschaften, Göttingen 1893, pp. 552-70.
Pischel (§11) belittles the value of the Lalita-vigraharäja-näfaka: "We cannot assess a higher value for this inscription than for any unique manuscript.
The inscription is full of gross errors of the dialects concerned, just like any manuscript of a drama. " [The Pärijäta-manjari had not yet been published when
Pischel wrote his grammar.] Here Pischel conspicuously ignores the aU-
important fact that the inscribed drama, permanently recorded at the time of its
composition, is very significantly diiferent from an ordinary drama ms. Any
dialect "errors" in it (other than those of the engravers, which are usually
obvious) must be those of the author himself Thus Pischel would have us
believe that not only Räjasekhara (see § 22), but also Somadeva, author ofthe
Lalita-vigraharäja-näiaka, did not know the languages in which they wrote. To
this list of literary ignoramuses would have to be added Madana, author ofthe Pärijäta-manjari, in which similar "errors" are common (see below, § 19).
^ ' For a list (not complete) of such "noteworthy" forms in the Pärijäta-manjari,
see A. Hillebrandt's review (Göttingische Gelehrte Anzeigen 170 [1908],
pp. 99-100) of Hultzsch's Pärijäta-manjari.
maragaya (lines 60, 71): as in KM (maragaa; the presence or absence
of the glide semi-vowel or ya-sruti is sporadic and in¬
significant) .
niyamva (line 42): the forms niamba and niambini appear in KM
verses 11.1,24, 26. These verses have otherwise S readings,
and thus the same form in the prose of the Pärijäta-
manjari confirms that these are S verses. (Cf. the case of
sarassai in 1.1, note 12.)
piayamo (line 38): cf gä(jhaaro in Km IV. 1, a S verse. Apparently
the t of the Skt. superlative suffix -tama was subject to
elision in S.
20. Similarly, in the S prose ofthe Lalita-vigraharäja-nätaka, we find
raana or rayana (1.8, 34, 38) for Skt. ratna, exactly as in the KM mss.^^
21. The point of all this is that such S forms as raana, maragaa, caut-
thia, etc. were not wrong, as Pischel, Konow, and others thought.
They were consistently used by the various authors, and could not be
the result of their individual mistakes; they were, quite simply, the
correct forms in S, or at least in late dramatic S. The change of -t- to -d-
was not, as at least some ofthe grammarians would have it, universal; it
was a general tendency, not a fixed rule. For modem editors to cling to
it so inflexibly, in the face of such consistently attested exceptions, is
self-deceiving. We must accept that the phonetic mles of the Pkt.
dialects were subject to numerous exceptions, and render the texts
accordingly.^'
22. The treatment of Skt. intervocalic -th- was similarly uneven, and
the inscriptional dramas are helpful here in the same way. Pischel,
Konow, et. al. took at their word grammarians such as Märkandeya,
who prescribed that -th- becomes -dh- in S, even though the mss.
frequently show -h- in many such words. Thus Konow emends mihuna
(III. 9.2) of all the mss. to midhuna in his text (as does Pischel in his
Konow discusses the irregular Prakrit forms in the Ldtita-vigraharaja- nätaka (which he calls "die absolut beste 'Handschrift' eines Drama"; p. 478) in
Göttingische Gelehrte Anzeigen 156 (1894), pp. 478-83.
" Phonetic developments do not always take place simultaneously in all
words in a given Middle Indo-Aryan dialect. Such changes often first make their apparance in a few words of a dialect, and become general only at a later stage.
Compare, for instance, the sporadic voicing of intervocalic t in ASokan Pkt., e.g.
hida = hita. A comprehensive investigation ofthe nature of these sporadic early
developments, to determine what phonetic or other factors govern them, could
be highly rewarding for the study ofthe history ofthe Indo-Aryan languages.)
The Original Language of the Karpura-manjan 127
edition of Sakuntala^'*). The same is the case with manoraha (III. 8.1),
emended to manoradha. Here again the Pärijäta-manjari consistently
has mihuna (lines 5, 46, 48) in S, and the Lalita-vigraharäja-nätaka has
maiwraha (II. 8), proving that these are original and accepted S forms.
23. Indeed, even the grammarians on whom Pischel relied so heavily
are themselves not in agreement on these crucial points. Thus the rule
for th -> dh in ^ is bahulam or 'general' (i.e. not universal) according to
Purusottama 9.9^', while others (pseudo-Vararuci, Märkandeya, Hema¬
candra, etc.) make it mandatory, t-^ dis likewise bahulam in both Puru¬
sottama (9.9) and Märkarujeya^* (9.20, citing S atulia for atulita), but
mandatory according to the others." All of this leaves Pischel's
approach very much in doubt. As we have seen (footnote 20), he thought
"poets" (§ 22, p. 23) did not themselves know the difference between Ö
and M. But now it appears that the grammarians were not much better,
as they themselves could not agree as to whether such forms as atulia or
jaha were proper S. Who, according to this logic, is left to really know
S?
24. The point here is that the Pkt. dialects were never as fixed or as
regular as both the traditional grammarians and the modern Indologists
wanted them to be. The phonetic rules for the various dialects were
subject to variation and sporadic application, as is incontrovertably
shown by the inscriptional data. To attempt to artificially regularize the
dialects, whether on the grounds of ms. corruption (now no longer
supportable), or of the authors' ignorance (never a very creditable
notion) , is simply to distort the data and disguise the true character of
Kälidäsa's SahmtaLä. 2nd ed. Cambridge: Harvard Univ. 1922. (Harvard
Oriental Series. 16.), 1.20.3 (p. 8); variants, p. 157, has "odd . . . mihu-."
" Appendix I of Rämasarman's Präkrta-Kalpataru. Ed. Manomohan Ghosh.
Calcutta: The Asiatic Soc. 1954. (Bibliotheca India. Vork no. 278), p. 162.
Purusottama's sütra 9.15, dadhäv aspastam uccaryau, "d and dh are to be
pronounced unclearly [in Sauraseni] ," may be a clue to the whole problem ofthe
treatment of intervocalic dentals in S. It seems to suggest that they were
pronounced with affrication, the degree of which may have varied under diffe¬
rent circumstances, tnus causing the uncertainty as to how they should be
rendered (the Indian scripts having no provision for writing such affricates). But
note also the contrasting sütra (II. 1.7) in Rämasarman's grammar, dhabhau
davat spxstam uccäraniyau; even this seems to have been a point of controversy.
^' Märkandeya's Präkrta-sarvasva. Ed. Krishna Chandra Acharya.
Ahmedabad: Prakrit Text Soc. 1968. (Prakrit Grammar Series. No. 1.) (Prakrit Text Society Series. No. 11.), p. 108.
" Compare the remarks of Ghosh (op.cit., footnote 2), pp. 29-30, note 2, on this point.
the dialects.'^ The KM, and all other Pkt. texts, must be edited by the
primary data, not by reference to such external sources as the gramma¬
rians, who are themselves inconsistent. The unfortunate results of the
latter procedure, as in the case of Konow's KM, have already been
demonstrated.
B. Ghosh
25. Ghosh, in the Introduction to his edition of the KM (footnote 2)
criticizes Konow on the grounds that he "rejected the testimony of his
best ms. on the plea that it went against grammatical rules" (p. 10) and
"in opposition to the testimony ofthe mss., including the best one, he
made violent changes in the orthography of the metrical portions ofthe
play to give it a Mahärästri feature" (p. 11). The justness of these
charges we have seen above in detail.
26. Ghosh's method was to choose a "best ms.," namely Konow's
ms. W, and to take it as the basis of his text, adopting the readings of
other mss. only in specical cases. His edition is thus basically a
corrected transcript of ms. W, which he considered to be the oldest and
most accurate. While this method of textual editing is certainly subject
to criticism,^' it cannot be denied that Ghosh's edition is far superior to
Konow's, since it at least avoids the extensive arbitrary emendations
of the latter.
27. Ghosh points out (pp. 11 and 48ff.) that Konow's assumption
that the KM was divided into S prose and M verse was wholly unjusti¬
fied, as we have already seen to be the case. Observing as we have that
^* While examples from S have been the subject of discussion here, similar
conclusions concerning the variability of M forms can be drawn from the same
data. Thus the Pärijäta-manjan shows such irregular M forms as -sahido (line
18), and vahadi (line 35). (Hillebrandt's list is again incomplete here.) The
Lalita-vigraharäja-nätaka likewise has moladi (II. 3), agahida- (11.23), and
amunida- (11.23).
" This practice of basing an edition on a "best" (or ädarsa) ms. is the standard method in editing texts in Bengali and other modern vernaculars, where it seems to be necessary due to the informal and fluid process of transmission to which vernacular texts are subject. Prakrit texts, in this respect, seem to stand some¬
where between the relatively faithful transmission process of Skt. literary texts and the free renderings ofthe vernaculars. Thus in reconstructing Pkt. texts we should at least attempt to apply the preferred methods of classical criticism,
weighing and comparing the testimonia of all the mss. according to their
evaluated position in a stemma codicum.
Ghosh used in addition to Konow's mss. 8other8(pp. 15-20), mostof which,
according to him, closely resembled Konow's mss. and did not add a great deal
to the text. Many other mss. of the KM are extant; see New Catalogus Catalo¬
gorum. Univ. of Madras 1967, Vol. Ill, pp. 189-91.
The Original Language of the Karpüra-manjari 129
Konow frequently changed Ö verses of the mss. into M, Ghosh conc¬
ludes that the KM was originally composed entirely in S (p. 11).
28. But here Ghosh has gone too far in reaction to Konow's manipu¬
lations. For while it is beyond question that much of Konow's M is arti-
fical, we need not assume that all of it is. Thus for instance verse 1.1,
discussed above (§ 5ff.), appears in Ghosh's edition in its true S form.
But the next verse (1.2) is a different matter. It appears in Ghosh
almost exactly as in Konow (except that the latter has -thana- instead
of -ghana-):
akaliaparirambhavibbhamäim ajaniacumbanädambaräim dürarn /
agha4iaghanatädanäim niccarn namaha anangaraina mohanäim //'"
What sort of S is this? The verse has 4 typically M forms with elided Skt.
-t-, each one unanimously attested by the mss.: akalia- (Skt. akalita-)
ajania- (ajnäta-), aghcujiia- (aghatita-), and rai-na {-ratinäm, or rather
-ratyoh) . This, it must be emphasized, is just the opposite of the cases of
the previously discussed verses, in which all or most of the mss. have S
readings.
29. This counter-example is far from unique. Many other verses in
Ghosh's edition have consistently M readings. For example we have in
verses 11.13-16 the following forms:
11.13 lambhiä (= lambhitä), bhamia- (bhramita-)
11.14 -panollia- (* -pranolita-)
11.15 nivesiä (nivesita), kario (käritah) II . 16 rehai ( rebhati) .
Ghosh largely glosses over the problem of these verses. He does make
note ofthe occasional elision of Skt. -t- in S verb forms (pp. 52, 143).
But he does not approach the problem ofthe occurrence of M. forms in a
consistent pattem; that is to say, he fails to explain why several of the
verses in his own text have all M forms. In view of this data, his theory
that the KM is entirely in S caimot be accepted. Some of the verses are
clearly in M, by the unanimous or near-unanimous testimony ofthe mss.
'" Both Lanman and Ghosh seem to have missed the point of this verse.
Lanman translates (p, 223), without comment.
Ever cherish ye deep reverence for the loves of Cupid and [his wife] Rati, in
which no flurried embraces are noticed, no noisy kissing is going on, nor
amorous beating of the breasts.
Ghosh notes on this (p. 180 note 6) "Räj[aäekhara]'s view of the union of
Anaiiga and Rati shows that he believed in the sanctity of love between a male and a female'." This is quite wrong. The point of the verse is a poetic pun on
Kama being Anaiiga, 'The Bodiless One.' Since his body was burnt to ashes by
the angry Siva, he cannot engage in embraces, kissing, or beating of breasts.
Surely there is no moralistic message intended here.
9 ZDMG 132/1
Thus while Ghosh succeeded in presenting a better text than Konow, his answer to the question of its original language(s) is not satisfactory,
though it is closer to the truth than Konow's.
C. MaJUMDAR-S ASTRI
30. In his brief article on "The Ävanti Prälq-it ofthe Karpüra-manjari"
(see footnote 5), Majumdar-Sastri proposed that the prose portions
of the drama were written in Ävanti, and not S, Pkt. His argument is
based on two features of the Ävanti dialect mentioned by the gramma¬
rians in their brief descriptions of it: that it contains many colloquial or
provincial words (lokokti), and that it arises from a mixture ofSauraseni and Mähärästri.
31. While it is true that the KM has many colloquial or desi words, this
is hardly definitive as to dialect, but rather is characteristic of many
types of Prakrit. Äs for the supposed mixture of S and M forms in the
prose, 1 have already tried to show that this is due more to a misunder¬
standing ofthe nature of S than to any peculiarity specific to the KM's
dialect. The example of Ävanti cited by Markandeya (11.1), moreover,
does not agree vrith the dialect of the KM : eso kirädo maam amisaranto
vedasalaägahanam paittho (= esa kiräto mrgam anusaran vetasalatäga-
hanam pravistah) . Here the grammarian points out the mixture of Ö
forms (e.g. kirädo, vedma-) with M words such as laa-. But in the S of
the KM the Skt. word lata always appears in the form ladä; it is not one
of the several, but limited words in which -t- is elided.
32. There is thus little reason to accept Majumdar-Sastri's idea
that the prose of the KM is to be identified vrith the poorly attested
Ävanti Prakrit. Moreover, he does not discuss the much more important
question of the language of the verses.
III. Determination of the Original Language
33. Up to now, we have seen that the majority ofthe verses ofthe KM
have the characteristic S treatment of Skt. -1-, while several others have
the M forms. So far, the verses have been discussed vrith reference to
phonetic features orüy. Thus before proceeding to a more detailed
analysis of the languages of the individual verses, it is necessary to
discuss the supposed morphological distinctions between S and M, and
to see how they apply to the main question.
34. Two features of the verbal system which are often thought to
differentiate the two dialects are the gerund and the passive. The
The Original Language of the Karpura-manjan 131
gerund, according to the gramijiarians, should have the suffix -una in M,
and -ia in S. Generally, the prose sections of the KM do have -ia
gerunds (e.g. daia I.20A0, riai'a 11.11.3, uppä4ia 1.20.31; exceptions,
e.g. gliettünal. 12.3, lcaduall.8.3), and the verses -üna {e.g. käünall.8,
datthüna 111.6; but also vinijjial. 13). On the surface this would seem to
support Konow's division of S prose andM verses. But Pischel (§ 584)
notes that "in S . . . we find in many dramas the forms in -tüna, -üna," so
that the gerunds in -üna in the KM do not rule out !§ verses. (It would
seem, in fact, that the distinction between the gerund types is more a
matter of prose vs. verse rather than of S vs. M.)
35. In the passive forms, § verbs are supposed to have stems in -la-,
and M in -ijja-. Looking at Konow's edition, one again gets the impres¬
sion that the prose and verse are neatly divided according to the rule;
but this is due in large part to his manipulation of the text. Thus his
kadhiadu (III. 3.6) is an emendation; all the mss. have kahijja(d)u or
kahijjai. Pucchianti is likewise emended for pucchijjanti etc. of the mss.
in 1. 18.13; muniadi (1.28.4) for munijjadi or other forms in -ijja- in 8 of 9
mss.; jäniadi for jänijjadi etc. in 9 of 10 mss. (1.28.2); and so on. S
passives in -ijja- are also attested in the Pärijäta-manjari {sevijjantl,
line 42) and the Lalita-vigraharäjan-ätaka {pekkhijjanti, viloijjanti, 1.5);
cf also Pischel § 535.
36. In nominal forms, we find the same pattem; the supposed strict
morphological division of the two dialects concemed is umeliable. The
locative singular masculine ending -mmi is considered by Pischel to be
appropriate in M, but not in S. Its frequent appearance in prose in the
KM and elsewhere he attributes (§ 366a) to Räjasekhara and other
"authors, who did not know to write S correctly."
37. More examples could be cited (for instance, of pronominal
forms); but the results are the same in each case. The morphological
distinctions between S and M are variable and unreliable, and many of
them seem to be a matter of prose vs. verse usages, rather than of
dialect differences. It remains, therefore, that the most reliable test of
dialect is the treatment of Skt. dentals, which gives a generally con¬
sistent (though not, as we have seen, infallible) means of differentiation.
38. Taking, therefore, the phonetic criterion as the most decisive, I
have attempted to determine which of the verses were originally in S ,
and which in M. Generally, the ms. testimonia for each verse fall into
one of three pattems. In some, as for example 1.1 discussed above, the
mss. agree in having all or nearly all S readings. In others, they have all
or predominantly M forms. In these two cases, the solution is clear. In
many verses, however, the mss. are divided, mostly along the lines of
9'
the southem and northem recensions, which give M and S reachngs respectively.
39. In such cases, I have usually decided in favor of the S forms ofthe
northem recension as genuine, on the following grounds:
40. a. The southem recension is so heavily Mähärästri-ized that M
forms occur frequently even in the prose portions: e.g. tagaam
{=tadgatam, II. 11.2), citthai {tisthati, II. 6.10), etc. Since there is no
question that the prose portions of the KM were originally in Ö (M is
never used in prose), these forms can only be cormptions, which cast
doubt on the validity ofthe M forms in the verses ofthe southem recen¬
sion as well."
41. b. An alteration of S forms to M follows the trend ofthe historical
development of the languages in question. It is therefore much more
reasonable to expect that original S would tend to be altered into M,
than vice versa.'■^
42. c. The conversion of the verses to M makes the KM fit more
readily into the standard literary convention of S prose and M verse.
The direction of change would be expected to be toward the standard,
rather than away from it; for this reason too, it is hardly possible that
the verses were altered from an original M into S, while the reverse is
quite likely.
43. Applying these principles, I have obtained the following results:
ofthe 144 verses in Konow's KM, 83 were originally in S and 57 in M;
the original dialect of the remaining 4 verses was indeterminable. The
figures for the separate acts are given in the following chart:
" Pischel similarly found that the southem recensions of Skt. dramas regu¬
larly Mähärästri-ized the ^ passages. See Zur Kenntniss der Qauraseni. In:
Beiträge zur Vergleichenden Sprachforschung 8 (1876), p. 138.
" I make this general assertion without entering into the controversy as to the status of the Prakrits as spoken or living languages at the period concemed. I would point out, however, that there is a peculiar logic behind Konow's charge that Räjasekhara's "faulty" knowledge ofthe Prakrits was due to the fact that
"the living knowledge of these dialects was, at that time [10th century A.D.], considerably diminished" (p. 204), when he chooses for his strict authority as to
these dialects (pp. 202-3) the very late grammarian Märkandeya, who lived in
the 16th or 17th century A.D. Compare on this point the comments of Amulya
Chandra Sen on Konow's edition in TTie Karpüramanjari. In: IHQ 4 (1928),
pp. 567-70: "The standard with which Konow has measured Räjasekhara's
Prakrit is Märkandeya's grammar and he has not given that importance to
earlier grammarians that they deserved" (p. 568). Cf. also the remarks by S. N.
Ghoshal in the Preface, pp. fxx-lxxi, of The Indian Drama [TTie Samkrit Drama]
(English tr. of Sten Konow: Das Indische Drama. Calkutta: General Printers
and Publishers 1969).
The Original Language ofthe Karpura-iiKinjun 133
Act I
Total II III IV
s 25 24 17 11 83
M 9 26 15 J7 57
Indeterminate 2
2
4
(For a detailed, verse by verse determination, see Appendix A. It must
be noted here that this compilation includes several verses whose
identification is not certain.)
44. As clear as the evidence ofthe mss. is, the pattem of occurrence
of the two dialects is not so regular as might have been wished. The
dialects are not divided by speaker, as most of the characters speak
verses in both dialects, often with equal frequency; thus the RäjäheiS 27
verses in S and 30 in M. In terms of meter, there is a partial regularity
insofar as the äryä is strongly preferred for M (28 of 32 äryä verses are in
M); the other meters, however, are used for both dialects.
45. There is, however, a more or less regular pattern in the placement
of the verses in the text, in that the verses of one dialect tend to be
grouped together, especially when they concem the same subject. Thus
long strings of M verses are found in 11.13-23 and 33-43. Examples of
similar sequences of S verses are 1.26-36 and 11.24-32. The pattern,
however, is not invariable; in some parts of the text S and M verses
seem to alternate freely.
46. Since this mixture of S and M verses is admittedly unusual, and
since no single determinant for their altemation has been found, it may
be pmdent to offer some justification for this conclusion above and
beyond the testimony of the mss. themselves. It should be pointed out,
first of all, that Räjasekhara was an innovator in writing the first all-
Pkt. drama. If he was willing to go against convention to this extent, it is
not altogether surprising that he would also introduce an unusual
mixture of dialects in his verses. And since he was, by his own claim, 'a
master of all languages' {savvabhäsäcadura, 1.7.1), he could certainly be
expected to use moie than one dialect.
47. The use ofthe two dialects by the same character should not raise
any serious objection, as this is the normal situation in Skt. dramas as
well, where the Pkt.-speaking personages use S in their prose and M in
verse. Moreover, as has been shovm, the difference between the two
dialects is actually quite small.
48. An apparent parallel to the use of Ö verse along with M in drama
is found in Pärijäta-manjari 1.29 (lines 30-1):
Devenahavanijjidänam sahasä vamdikadamteuram
dimtenam samidä vioyaviyanä jänam vasar(ituggame /
Ede te namiüna päyajuvalam dam4appanämena de
patticchattabharena bhümivaino vaccamti väsälayam //
This verse has 4 S forms: -nijjidänam (= nirjitänüm) , vamdikada-
{vandlkrtd), samidä [samitä) and ede {ete). Only -vaino (= patay ah)
appears to be an M form; but elsewhere in the same text (line 40)
vanassalnarri (= vanaspatinäm) appears in prose, indicating that the
stem vai- for pati- was one of the accepted S forms with elided -t- (cf
above, § 21). This verse, then, is in S, while the rest ofthe Pkt. verses
are in M, providing at least a partial parallel to the altemation of
dialects in the KM verses. (One wonders how many more such cases
would appear in literary Pkt. were it not for the levelling influence ofthe
processes of transmission and editing to which those texts, unlike the
Pärijäta-marijan, were subjected.)
49. Finally, and perhaps most importantly, there is the testimony of
Märljandeya. The grammarian refers 6 times to Räjasekhara, including
5 quotations, all from the KM. It is clear from the form of the quotations
that he was familiar with the northern recension of the drama," which,
as we have seen, preserves the original S ofthe verses. Moreover, one of
the verse passages which he quotes (KM II. 8, in his commentary on
sütra VI. 48) reads tarn, manne maha''* dukkidam parinadarri dukkhäna
sikkhävanam, with typical S forms. Yet in his sixth reference {sütra
VI. 4) to Räjasekhara (and presumably to the KM) Märkandeya criti¬
cizes his use of the -di/-de verb endings in M verses: Räjasekharasya
mahärä^tfyäsraye^u sloke^v api drsyata iti cet, tasyaivävimrsyakäritvam,
Tiäparädho'nusäsaTiänäm ("If it be maintained that [the verbs endings
-di and -de] are observed in verses of Räjasekhara which are composed in
Mahärästri-- that is his carelessness, and not the fault ofthe [gramma¬
tical] texts"). The point here is not to argue the differences between
grammarians' and poets' ideas of Pkt."; it is that Märkancleya evident-
E.g. in sütra VII. 61, he quotes KM 1.17 as vaamti cettanilä, which is the
reading of the northern recension; the southern mss. (STU) have äarnti. The
same pattern holds true for Märkandeya's other citations of Räjasekhara (i.e. of the KM) in 1.50 (KM I.ll), III.77 (1.24), V.118 (1.32), and VI.48 (II.8).
The published editions have e(d)o7?t tarji maha instead of tam manne maha.
Märkandeya is evidently referring to verbs in -di which do occur in M verses in some of the mss. of the northmi recension; e.g. anuharadi (1.6, mss. WO) or jaladi, haradi, etc. (III. 20, mss.PN). It is difficult to determine whether such
readings are original, but they are possible in view of such definite M readings as vahadi in the Pärijäta-manjari (cf footnote 28).
Märkandeya's criticism of R is also cited by Pischel (§ 22, p. 23) in the form
The Original Language of the Karpura-manjan 135
ly was aware that Räjasekhara's verses (in the KM) were in both S and
M. If all the verses were in M, he would not have had to specify them as
maharastryasrayesw"^; if all the verses were in the comment would be
meaningless. We have, therefore, in Märkandeya's grammar a clear, if
indirect testimony within the tradition of Räjasekhara's use of both S
and M in his verses.
50. There remains to be discussed only the matter of the traditional
prescriptions of the language ofthe sattaka, the dramatic type of which
the KM is the outstanding example. Ghosh" cites the definition given
in Hemacandra's Kävyänusäsana (VIII. 3)'* in support of his theory that
the KM was written entirely in S:
Viskambhakapravesakarahito vastv ekabhäsayä bhavati /
Apräkrtasainskrtayä sa sattako nätikäpratimah //
This definition does clearly say that a sattaka is composed in "one
language, neither Sanskrit or Prakrit." Ghosh takes "Prakrit" here to
mean the Prakrit par excellence, namely M; thus according to him the
definition perfectly describes his all-S version of the KM. Krishna
Chandra Acharya, who evidently supports Konow's ideas on the
language ofthe KM," disagrees, sajing that a sattaka "is written in one
language only and not in both Pkt. and Skt. as in the case ofthe Nätikä.
Here one language means Pkt. in general."'"' It does not, however, seem possible that the phrase vastv . . . ekabhäsayä . . . apräkrtasarnskrtayä could mean "a piece in one language, and not in (both) Skt. and Pkt." A
more plausible suggestion is that of Chakravarty (op.cit., footnote
38, p. 171), that the original sattakas (which are not extant) may have
rajasekharasya mähärästryäh prayoge slokesvapi dxsyata iti kecit, as evidence of his "dialectal errors." This only goes to show that grammarians and poets, be
they ancient or modern, tend to disagree on matters of usage.
Also in sitira III. 77, Märkandeya reproaches Räjasekhara for his 'apapäfha'
kadu- (= kratu) in an M verse (KM 1.24; 8 of 9 mss.).
'* The exact phrase varies from ms. to ms. (mahärästryäsraye sloke'pi, mähä-
rästnprayoge, etc.), but all the readings make same point (see Acharya's
edition, p. 67, note 7). " Op.cit., note 2, p. 11.
'* Actually, the definition is originally Bhoja's; see V. Raghavan: Bhoja's
Srngara Prakäsa. Madras: Punarvasu 1963, pp. 540-1. The same definition is
used in the Nätyadarpana as well; cf Chintaharan Chakravarty: Characte¬
ristic Features of the Sattaka Form of Drama. In: IHQ 7 (1931), p. 171.
"Acharya {Präkrta-sarvasva, introduction, p. 82) cites such forms as Aiai'and pisunei as examples of M in the KM; but he fails to note that the former occurs frequently in the Ö prose portions as well (1.4.19, 5.5, 6.2 etc.). The latter form does not occur at all, except as a variant (1.20.6) in the umeliable southern
recension. ''° Ibid. p. 81.
been "in spoken dialects which were different alike from Sanskrit and literary Präkrt, i.e. in one ofthe vibhäsä' s or apabhramsds (sub-literary vernaculars).'"
51. But the matter does not end here; for there are other, quite diffe¬
rent traditional opinions on the appropriate language(s) for a sattalca.
Thus the Sähitya-darpana of Visv&neLtha!*^ prescribes sattakam präkrtä-
sesapathyam (VI. 276), i.e. a sattaka should be recited in Pkt. entirely
(without specifying the type(s) of Pkt.). More specific is the definition in
the jBÄäwa-prafcäs'amof Saradatanaya"': sattakam . . . sürasenamahäräs-
traväcyabhäsädikalpitam, "a sattaka is composed in the languages
spoken in Sürasena and Mahärästra, etc." (i.e. in S and M). Öäradäta-
naya also notes other opinions on the proper languages, citing the views
that the king should speak Mägadbi or Sauraseni, or should not speak
Pkt. at all (i.e. should speak Skt.)."'' (The latter opinion is supported by
the evidence of another sattaka, the Rambhä-maüjan, in which the king
does speak Skt.'*') But the final view which he cites, presumably as the
correct conclusion, is that the sattaka should indeed be in Pkt.: Tena
tasyähuh bhäsäm täni präkrtim pare / Räjasekharaklptarn tadyathä
Karpüramanjari (p. 269, lines 10-11).
52. The conclusion to be drawn from all this is that there was little
agreement among the various authorities as to the correct language(s)
to be used in a sattaka. (In Räjasekhara's own definition of a sattaka in
KM 1.6 he does not address the question of language.) But it is certainly
worthy of note that the most specific determination on the point, that of
Säradätanaya, describes exactly the pattem of the KM as it has been
formulated in this paper, and moreover specifically refers to it as the
prototjrpe of the sattaka. (Unfortunately neither he nor any of the other
writers go into the specifics ofthe distribution ofthe dialects within the
sattaka.) Equally important is the lesson that one must avoid drawing
conclusions (as did Ghosh) from only one of several differing authori¬
ties. The case is much the same as we saw in discussing the contradicto¬
ry opinions ofthe grammarians (above, § 23), and the point is the same:
'" Cf also Raghavan, op.cit. (footnote 38), p. 541 on this point, noting a reference to Apabhrainsa sattakas.
" Ed. ÄCÄRYA Krsnamohana Sästri. Varanasi: Chowkhamba Sankrit
Series Office 1967. (Kashi Sanskrit Series. No. 145.), p. 452.
'*' Ed. Yadugiri Yatiraja Swami and K. S. Ramaswami Sastri. Baroda:
Oriental Inst. 1930 -I- 1968. (Gaekwad's Oriental Series. No. 45a.), p. 269, line 4.
'''' Ibid., hnes 7-8: Na vadet präkrtim. bhäsäm räjeti katicij jaguh / Mägadhyä saurasenyä vä vaded räjeti kecana //.
Chakravarty, op.cit. (footnote 38), p. 172.
The Original Language of the Karpura-manjan 137
the internal evidence of a given text outweighs extemal testimonia,
especially when, as is so often the case, the latter are mutually contra¬
dictory.
IV. Conclusions
53. a. A detailed analysis of the available ms. data on the text ofthe
KM indicates that the drama was originally composed in both S and M
Pkt., with the prose portions and the majority of the verses (roughly
60%) in S, and the remaining verses (about 40%) in M. Thus both
Konow et. al., who thought all the verses were in M, and Ghosh, who
believed the whole text was in S, were partly right and partly wrong; the
tmth lies somewhere in between.
54. b. The Ö (and to a lesser extent the M) ofthe KM differs from that
prescribed by most of the grammarians, especially in the irregular
application of phonetic mles given by them as mandatory. These irregu¬
larities are definitely authentic (i.e. not due to ms. cormption), as they
are corroborated by inscriptional and other sources for later dramatic
Prakrits. The ascription by Pischel, Konow and others of such forms
to the ignorance of the authors is entirely unacceptable, as most of the
grammarians on whom they rely are later than the poets, and them¬
selves disagree on many ofthe cmcial points. The logical conclusion is
rather that the Prakrits, even in the late medieval period, were not
nearly as fixed and standardized as the grammarians, traditional and
modem, would have us believe. Specifically, S and M were in practice
quite similar, being distinguished chiefly by their dilTering treatment of
dental consonants. The grammatical and lexical distinctions between
them prescribed by the grammarians were at best sporadically
observed, and cannot be relied on vrithout corroboration.
55. c. In determining the original readings of Pkt. texts, the only
accurate and productive method is to give the greatest weight to the
mss. themselves. To "correct" the mss. according to preconceived
notions of what the dialects of the text were, and what forms those
dialects should take, is circular logic, and disguises the tme nature and
usages ofthe languages. If the mss. lead us to unexpected and unfami¬
liar results, we must try to understand these results, rather than to bend
the data to fit preconceived molds, or to write them off as due to the
leamed authors' ignorance of their ovra languages.
Appendix A: Verses of the KM by Language
56. In this appendix the verses are identified, as far as possible, as
originally S or M. For each verse one or more characteristic forms ofthe
appropriate dialect is given from Konow's text or variants. Many of
the original S verses are consistently represented in the southem
recension (mss. STU) with M forms; such verses are indicated by the
abbreviation 's-M.'
57. For several verses the identification of the original language is
uncertain, due either to the absence ofthe best criteria (words with Skt.
-t-, etc.), or to conflicting or inconsistent testimony in the mss. Such
verses are noted with a question mark after the number.
58. In the case of 4 verses (part C), it was impossible from the
available data to make a judgment as to the original language.
A. Sauraseni verses (83)
59. 1.1 {{b)hodu, paattadu, etc.; see above, § 6-9); 4 (-^ppanadisu,
nada-, etc.); 5 (kahijjadu, edam; bhanna't^); 11 (kidim, paümjaidurfi,
edam, icchadi; cf above, § 9); 12 {-sudam, parinedi; s-M); 13? (vinijjia, a
S gemnd form); 14 (gade, -pumjida-; s-M); 16? (jädam, -ävattida-; lek-
khijjae*^; but aU mss. have papphülliä, etc. — a permissible S form?) ; 17
(-andolia-, cumbida-; s-M); 18 (mumca(d)ha; s-M); 20 (khalidd, jädä;
s-M); 21 (-pamkhidmu, jädi, khamcida-; aidiha- = atidirgha*^); 22?
(-ppasädä = prasädät in 8 of 10 mss.); 23? (dikkhidä, bhädi; pijjae,
khajjae*^; s-M); 26 (-alambida-, nivasidam, -tthidä, änidä, abbhude-; but
dhoa-'*^; s-M); 27 (lihijjadi, calidam, camkamanado); 28 (-kkhada-,
-ladäe); 29 (-kedaa-, dhavalido, nhävido, churido, jädo; s-M); 30
(vetthidum, jädi, vilihidum, jädi; -pasai- = prasrti^" , -valiam = valitam^^;
s-M); 31 (samummiladi); 32 (khalidani, sajjida-, -ppahudino;
rakkhijjae*^); 33 (-kidarfi, bhädi, -gadä, valaia-^); 34? {-ladä, 6 of 10
Bhannaiis evidently a permissible § passive, as it occurs in prose in 1.20.4 (7 of 10 mss.). Cf also Ghosh's note on the form, p. 145; and footnote 47, below.
Passives in -(i}jjae or -ae occur so consistently in the mss. in verses with otherwise S forms (1.23,11.29,30, III. 9,33) that they must have been accepted S forms, at least in verses in the later dialect. Cf § 35, § 73, and footnote 46.
"* Ai- for Skt. ati- is apparently permissible in dramatic S; cf ainiuna = atini- punä (11.29.10), in 8 of 9 mss.
Dhoa = dhauta may have been acceptable in S ; cf 1.30.1, where 4 mss. have dhoa in prose.
'° Pasai'occurs in prose in 1. 16.3 (7 mss.; the rest have a different word), and thus is correct in S.
" Valia (1.30, 11.24) and valaia (1.33, IV.20) occur regularly in otherwise S verses, and thus seem to be correct in Ö. (But note also valaidain 3 of 10 mss. in 1.33.)
The Original Language of the Karpüra-manjari 139
mss.; (hädi, 2 mss.; but tiiei or na ei in 8 of lO'^); 35 {edarn, jänädi,
sarnpadi, gadam, jädä, etc.; s-M); 36 (vilasadi).
60. 11.1 (calidä, tthido, jädä, ladä; niamba-^^, caüvvihä^'*; s-M); 3 (jädä,
pmmkhido; s-M); 4 (vahuttadi, khuttadi, lottadi, visaftadi, etc.; s-M); 5
(nivadidä; s-M); 6 (kadhida-, saradi, -nirikkhidesum, etc.; s-M); 7
(kedai-, pesidam, lamchidam; s-M); 8 (vanicido, edaixi, dukkidam pari-
nadam; cf above, § 49); 9 (-ladäe, jividäsä); 10 (kade; s-M); 11? (khada-,
-ladä; hut jaladi 'm 2 mss. only, others jaioi) ; 12? (uvvadidam, etc., 6 of 8
mss.; rosäniarfi^^); 24 (pisunadi; -niamba-^^, -valia-^^); 25 (samum¬
miladi, laggadi); 26? (mahurijjadi; niambirilo^^; hut pakidi- = prakrti va.
southern mss. only, contrary to the usual pattem—cf pasai, footnote
50; s-M); 27 (kijjadi, -du, kiradi, etc. in all northem mss.; -sariä- =
sarif^) ; 28? (indeterminate from the text; but presumably S as it occurs
within a sequence of Ö verses, 24-32— cf above, § 45); 29 (nihido,
-ucchamgado, -udddmarido, kaddhijjadi or -jjae*^, etc.; but all mss. have
bolev'^); 30 (-nihidani; disae in 8 of 9 mss."'); 31 (kunadi; s-M); 32
(-kkanida-, -janida-); 44 (lambhido; s-M); 46 (-sarnvaggidanam, niva¬
didä, romaOcido, tthido or vattadi; see above, § 11); 47 (ranida-, kadam,
thidain); 50 (jädo).
61. lll.l (kijjadu, navoggaderndu-); 3 (-gadarn, tädidumanä, sarnthidä,
dharidä, etc.); 4 (todo, tthido, jida-; caüssatthisu, cf footnote 54); 5
(ekkekkadäe, sainthidä); 7 (-vahalida-, -sarnmilidänam, pidido; s-M);
8? (the verse as it stands is indeterminate, but the altemate reading
offered in ms. W is clearly Ö (bhodi, aditthe) , so that the main verse was
probably S also); 9 (samghadadi, mahijjadi; dijjae*^; s-M); 10?
(-ghadanädi- in 6 of 8 mss., -vaddhida- in 6 of 8; but eiin all mss.'^); 15?
(kijjadi in 3 northem mss., jadi in 2; bhedo in 2); 18? (jaadi; maragaa-
correct in S, see above, § 15,19; s-M); 21 (pasidadu); 22? (haradi; rosä-
niarrv'^); 23 (-thida,-, bhodu); 25 (tthide, samjädä, gadd); 28? (jä(tein4 of 8
mss.; s-M); 31? (bhädi in 3 of 6 northem mss., others have different
word; s-M); 32? (kado in 6 of 8 mss.; but most have hoi/hou).
The sporadic S dropping of -t- in the verb ending -ti seems to occur especial¬
ly frequently in the environment of e (preceding or following): niei or na ei (1.34), bolei (11.29) ei (III. 10); and the frequent passives in -ae (cf. § 70, footnote 47).
IViamba = Skt. nitamba is correct in §; cf § 19 above.
'" Caiivviha = Skt. caturvidha is correct in S; cf caiitthi in § 19 above.
Rosania seems to be correct in S; cf III. 22, § 61.
Sariä is correct S; it occurs in prose in III. 3.7 (all mss.; Konow emends sarida) .
62. IV. 1 (sahidavvo, kadham; gädhaaro^^); 2 [idam, -karälido); 3
{lahadi, vipphuradi, tado, khamdijjadi or -jjae*^; s-M); 4? {edam, ede,
etc.; padosa- = pradosa in 2 mss.; but sialä = sUalc^^); 5 {-churida-,
-ladä-, -pariväsida-, {b)hodu); 8 {nivesida-, -ukkhidänam, kunadi); 10?
(no clear evidence from text", but this and verses 12-15 can
be presumed to be S, as they are part ofa monologue, verses 10-18, by
the vidüsaka, in which the identifiable verses are all S; cf. § 45); 11 {ido,
-anugada-, disadu); 12-15? (cf IV. 10); 16 (väld)ida-); 17
{-kada-/-kida-; -gidi-); 18? (A:o(iM/!.aMa-in 3 northern mss., -wädom-in4;
-häsida- in 2, others -häsia-); 19 (jaadi; s-M); 20 {jädi, pumkhidehim;
valaia-^ ^).
B. Mähärästri verses (57)
63. 1.2 {akalia-, ajania-, aghadia, -raina; cf.above, § 28); 3? {-suänam,
deu—hut 2 mss. have dedu); 6? {anuharai, but 3 mss. have -di; bhannai
can be either Ö or M""); 7 {hoi; according to Ghosh this verse is
spurious); 8 {hoi, hou, except STU hodu, contrary to the usual pattern);
15 {rai- = rati in 8 of 10 mss.; vää, 8 of 10; sialä inconclusive'*); 19?
(gäliassa, 7 of 10 mss.); 24? {-daiena, suraa-; kadu- = kratu, cf footnote
35); 25? {-avainnarn) .
64. II. 2 {parahua- = parabhrta, jampie, rai-); 13 {lambhiä, bhamia-);
14 {-panollia-); 15 {nivesiä, kärio); lß{rehai); 17 {thavio, sevai); 18 {nive- siarn, carnkamio); 19 {-jania-, kaäirn); 20 {raiä, vattai); 21 {darnsiarn); 22
{pasähiä); 23 (rehae); 33 {-tthia-, -campiam, hakkärai); 34 (kilakilai); 35
{vikkharai, -vaino); 36 {-panollia-, -däviäirn); 37 {dei}; 38 {pasai-,
kouhallena); 39 {hou/hoi, etc.; padai); 40 {eäi, -cariäim, lihai); 41?
{viharai; vihäiin 5 mss., -diin 4:; jie, joa etc. in 5, jädam in 4); 42 {täveij;
43? {-haä in northem mss., -hadä in southem—opposite of the usual
pattem); 45 {samuggirai) ; 48 {dharai, -devaa= devatä, vilasai); 49 (A;om-
halena, -lasia-, nivasai).
65. III. 2 {-kavalianta, mälal, -valia-); 6 {daiä,i); 11 {-luntia-, paasai);
12 {paadei); 14 {mahialarnmi); 16 {unnaehim); 17? (-i;airiö, ghadei,
paadei; but 6 of 8 mss. have rfwrfto or -yo = dvitiyah— a permissible form
in M?); 19 {-valaiarn, rai-); 20 {kirai, -niggaam, jalai, harai; mss. PN
have S readings); 24? (text in Konow indeterminate, but variants
GädJiaaro is correct^; it occurs in prose in 1.20.43 (all mss.; Konow emends gädhadaro; cf. also § 19).
Siala may be correct in S; cf IV. 6. The word does not occur in prose in the KM, however, so its status is uncertain.
"in these verses the vidüsaka is describing the carcari dancers. The verbs are therefore all in the 3rd person plural, which is the same in S and M, thereby ruling out the convenient dialect distinction of the singular verb (i.e. -di vs. -i).
The Original Language of the Karpüra-manjari 141
include nijjia, -sarngaä; but also -niggadä in ms.P); 26 {nahaalammi =
nahhastale, purnjijjai); 27? (-pärävaä, -dülsaa; but 5 mss. have -ladä
instead of -lad); 29 (vattai); 30 {-raia-, jaai); 33? {lahejja, optative,
presumably a M form; rajjae and dijjae indeterminate'").
66. IV. 6? {-sialä, -larji, 4 times—but may be correct in S as well'*); 7
{-lad); 9 {rairn); 21 {Iceai-); 22 {keai-); 23 {-manasavai, hoi, samjää,
bhariam); 24 (viramaii). [The last 3 verses occur in the southem mss.
only and, as noted by Ghosh, are probably not original.]
C. Indeterminate verses (4):
67. 1.9-10; III. 13 {niambininarn is indeterminate"); III. 34.
Appendix B: Special Features of Sauraseni in the KM
68. Below are summarized the unusual phonetic and grammatical
features of the S portions ofthe KM. The forms listed are generally well-
attested in the mss., though they are frequently regularized in the
editions (especially Konow's).
69. Skt. intervocalic -t- is elided in these words and particles: ai-,
gädhaara, caütthi/caüvviha; dhoa{'i.), niamba/niambinl, pasai, maragaa,
raana, rosänia, valia/valaia, sarassai, sariä.
70. [Note that elision most often occurs in the environment of i or e
(footnote 52), and less frequently ofwand o. Mostofthe cases where tia
elided between two a vowels involve special circumstances; thus
gädhaara with the addition ofthe superlative suffix (ef § 19), and raana,
(= ratna) as the result of svarabhakti.^'^
71. Tha is always in this form, never idha as prescribed by the gram¬
marians and found in some other S texts.
72. Intervocalic -th- is generally reduced to h, as in mihuna and
manoraha.
73. In the verses, the 3rd person singular passive is regularly formed
in -ae (cf footnote 47), a form not otherwise noted in the grammars. The
passive bhannai occurs in prose as well as verse. In the prose, passives
are generally formed with either of the stems -ta- or -{i)jja- plus the
ending -di.
Abbreviations used in the text
KM = Karpüra-manjari S = Sauraseni
M = Mähärästri Skt. = Sanskrit
Pkt. = Prakrit
It might also be noted that maragaa = marakata is a loan word (cf Greek
onapayöog, Latin smaragdus, etc.), though it is questionable whether this would have any bearing on its Prakrit developments.
By Arvind Sharma, Sydney
I
Ancient Indian literature abounds in pejorative references to woman¬
kind. ' This is not to say that positive sentiments towards women are not
expressed at all;^ rather that the negative references abound to such a
degree as to suggest the deduction that there was a significant misogy-
nistic trend in the intellectual circles of ancient India.'
A strong protest against this trend, however, can be heard in the
writings of the polymath Varähamihira." Varähamihira is traditionally
counted, along with Kälidäsa and others, as one ofthe nine gems which
adorned the court of king Vikramäditya.' Historians are inclined to
place Varähamihira around the sixth century A.D.* and he is regarded
as one of the leading lights of the Gupta period of ancient Indian
' Pandurang Vaman Kane: History of Dharmasästra. Vol. II, Part I.
Poona: Bhandarkar Oriental Research Inst. 1941, pp. 576-577.
^ Ibid., pp. 578-581.
' See P. Thomas: Indian Women through the Ages. London: Asia Publ. House
1964, pp. 217-220 etc.; but also see Pandurang Vaman Kane, op. cit.,
p. 581.
"* "Uncharitable and unmerited remarks against womanhood were not
wanting. Varähamihira raises his solitary voice against such allegations. He
openly says that women are the veritable goddesses of fortune and should
always be honoured and given wealth, that all the faults which women are
accused of are also committed by men but the latter in their audacity treat
women with scorn though they are superior to men in virtues, that whether she
be wife or mother, the origin of men depends on women, and that the lustful
craving of man does not subside even when he is centenarian and he keeps away
from it only due to incapacity, while women do so by courage and patience
(LXXIII. 4, 6, 11, 14)". (Ajay Mitra Shastri: India as Seen in the Brhatsam¬
hitä of Varähamihira. Delhi: Motilal Banarsidass 1969, p. 208).
^ R. C. Majumdar, H. C. Raychaudhuri and Kalikinkar Datta: An
Advanced History of India. London: Macmillan 1948, p. 149.
* Ibid.