Old, older, and oldest Dharmasastra: the manuscript tradition of the Manu Sastra, the original text of the Manu Sastra, and the first dharmasutras.

Fitzgerald, James L.

I. INTRODUCTORY COMMENTS

No scholar today has done more to put on the record what the old Brahmin tradition considered to be dharma than Patrick Olivelle. (1) Beginning with his early works on yatidharma, the asramadharma, and the samnyasa-upanipds, continuing with the critically important edited volume on dharma in 2004 (2)--which contained a major piece of research and reflection by him on the word's later Vedic and post-Vedic history (3)--and the two big volumes under discussion here, and culminating in a critical edition of the Vaisnava Dharmakistra in 2009 and numerous recent articles and chapters, Olivelle has furnished those interested in mapping the word and idea "dharma" an abundance of materials, insights, and arguments. Those materials assist us fundamentally on a daily basis, and the insights and arguments will hold our attention for a long time to come. (And as all know, this tidal wave of publications on dharma is hardly all that Olivelle has been doing in the past few decades!) All of Indology stands deeply in Olivelle's debt for so much important work so well done.

The two books under discussion here are contributions of fundamental scholarship: new presentations in Sanskrit and English of the five earliest texts of the tradition of Brahmin dharmagastra--the four earliest dharmasutras and the Manavadharmasastra, which came into existence four to five centuries after the first dharmasutra. (4) The men and institutions of this tradition had, perhaps, more direct influence on the shape of the social and political realities of India than any other single stream of voices in South Asian history. Though the learned men of the dharmagastra and their texts are profoundly out of temper with a number of the norms that have come to prevail in the past one hundred years--they were the scholarly members of a cultural elite claiming a special status, with attendant benefits, privileges, and exemptions--it is important that this tradition be charted with as much historicistic detail and nuance as can be mustered. Precisely because the misalignment between dharmagastra and modernity is not simply a thing of the past, fundamental scholarship such as Olivelle's here is much more than an antiquarian completion of the record. There are even deeper misalignments of intellectual and moral presuppositions between the Brahmin elite of India and the elites of the modern Western worldview--misalignments that contribute to the contemporary tensions between some of the pious of both traditions--and on their margins these two works may occasionally help illuminate these matters, as I hope some of my comments below will do, particularly with regard to the implicit norms of oral and written texts in connection with the Manu gastra (see III.3 below).

As will become clear, the editions of the Sanskrit texts presented in these two volumes are fundamentally different in nature: Olivelle's sutra texts are carefully rearticulated presentations of refurbished and improved versions of older editions of those texts, while his text of the Manavadhamasastra is an entirely new critical edition based upon fifty-three manuscripts, the great majority of which have never been used before.

These two volumes provide the best available presentations of these texts to date--they will be the scholarly standard for these texts for Sanskritists and non-Sanskritists both for many years to come. These books, and some of Olivelle's attendant scholarship, raise some critically important issues that require much more space to address than is available here and now. So OliveIle's theories about the dharma of the dharmagastra being the result of a Buddhist revival of the word (DS, 14) and his conviction that the Manavadharmasastra was composed in a single authorial burst (MCL, 5-7) will be treated in detail elsewhere. Here I attend to matters of the basic scholarship that grounds these books, coming to focus upon translational issues--principally of the word dharma--for the sutra volume and issues of textual criticism for Manu 's Code of Law. Particularly important with his new edition of the Manu gastra is the work Olivelle has done there to focus our attention upon what he calls the "after-life" of the text, a matter that should become a principal rather than a secondary concern of Western scholarship. His new edition is based upon a splendid marshalling of new resources that may, conceivably, yield even more results in the not-too-distant future (see my suggestion regarding cladistics analysis at the end of this review).

II. THE DHARMASOTRA VOLUME

The heart of Olivelle's Dharmasatras consists of /Hi pages presenting the texts of the four extant old sutras with Olivelle's translations of them on facing pages, in the chronological order of the core texts as he has determined it: Apastamba, Gautama, Baudhayana, and Vasistha. Each of the four sutras is preceded by a brief discussion of important features not shared with the other three, and this notice is followed by a detailed conspectus of its contents. Following the texts and translations are 234 pages of notes. The body of notes for each sutra commences with a description of the refurbished edition of the sutra Olivelle presents, and the notes (which are signaled by note-numbers in the body of the Sanskrit texts) present variant readings, discussions of the reading adopted, quotations from four commentators, and discussions of the translation and/or meaning of the text. Three appendices follow the notes: one a glossary of ritual vocabulary, a second explaining the names of gods, people, and places, and the last a four-page listing of fauna and flora. A bibliography and a relatively detailed index of groups of terms under topic headings bring the book to a close.

A general introduction preceding the body of the work briefly takes up various general matters regarding the form, content, and history of the siitras. Salient among these are their authorship and dating, their internal structures, the meaning of "dharma" in these works, and a concluding discussion of the import of the plurality of contending voices contained in these stitras (texts often wrongly presumed by modern readers to be normative in simple and straightforward ways).

Olivelle's revised editions come from his painstaking reviews of and corrections to the first and, in three cases, subsequent editions of the four sutras produced by four nineteenth-century German editors (Georg Biihler's two editions of Apastamba, Adolf Stenzler's edition of Gautama, Eugen Hultzsch's two editions of Baudhayana, and A. Anton Fuhrer's two editions of Vasistha). Olivelle corrected and revised these to one extent or another in light of various philological discussions of the first editions, especially criticisms from Otto von Bohtlingk, (5) and several subsequent editions of the sutras published in the first half of the twentieth century. In the case of two sutras (Gautama and Vasistha) Olivelle gathered two new manuscripts of each from Nepal and also brought in the quotations of them in later dharmagastra literature; in the case of Vasigha he made use of a preliminary report of the results of Harry Falles gathering of new manuscripts of this text in preparation of a critical edition of it. (6) The amount of editorial innovation on Olivelle's part seems to vary significantly from one text to the next, with Apastamba remaining more or less the same as Biihler's second edition of it, (7) but Gautama and Vasistha eliciting more numerous editorial alterations on his part. Olivelle makes it perfectly clear that the editions in this volume are based on the work of earlier editors, though only dharmasastra specialists may take cognizance of just how much has been forwarded to the present work from earlier editions and annotations. (8) Olivelle's editorial project is unusual, but given the long absence of entirely new critical editions of the sutras based on a thorough collection of new manuscripts, Olivelle's synthesis here is a welcome and valuable scholarly resource.(9)

An interesting liminal issue arises as we pass from the editor's establishing the readings of the text to questions of their interpretation and translation. I refer to the need for conscious decisions about how the individually demarcated units of the texts, the "sutras," (10) are to be connected, or not. The last, and only, scholar previously to translate all the dharmasutras, Georg Baler, rendered all discretely numbered units of these texts in a single numbered stream of separate paragraphs, one for each such "sutra." (11) Olivelle has presented the contents of the texts in a radically different way, though, unfortunately, without any discussion of his motives, grounds, or method for doing so. (12) In a dramatic departure from Buhler's presentation, Olivelle places an interpretive overlay on the texts by distinguishing three levels of content within them and using his own labels for each level. (13) He uses white space to suggest continuities and discontinuities in the matter, grouping sutras into distinct paragraphs and distinguishing verses from prose by giving each stanza its own paragraph printed in smaller type and indented on both sides. This overlay is a profound commentarial intervention in the presentation of the texts, but one that constitutes a justifiable and carefully considered interpretive translation of these texts to a radically different intellectual context from that of their origin, even if there are significant elements of subjectivity at many interpretive cruces (and thus grounds for occasional disagreement). Some of these very significant departures from Buhler's presentations of the translated texts were anticipated by Elvira Friedrich in a conceptually provocative examination and translation of Apastamba that appeared in 1993. (14) Preparatory to an effort to read Apastamba divorced from the guidance of the main traditional commentator, Haradatta, Friedrich carefully studied the structure of Apastamba and his instructional patterns. She explicitly posed questions about the grouping of sutras into sentences and paragraphs and, having made her interpretive judgments, used her own labels to guide her readers' understanding of the translated text; she did not, however, distinguish multiple hierarchical levels in the text, as Olivelle has done. (15)

II.1 Olivelle's Translation of the Dharmasfitras in General

Though Buhler's English translations of the sutras remain generally correct and accurate within the scope of what current scholarship can be confident of understanding, (16) Olivelle's translations and annotations clearly supersede them in a number of ways. Olivelle's translations are directed at a wider scholarly readership than were Buhler's, which basically were aimed at Sanskrit philologists. Olivelle has removed various scholastic speed-bumps in the translations (the parentheses and brackets often used to mark off the elements of a translation that are implied in the text from those that are explicitly present), achieved a smoother and more rapid English through justifiable rearrangements of the syntax that still remain faithful to the sense of the Sanskrit, and modernized the English diction. (17) The often technical rules of the sutras cannot possibly be turned into smooth prose for readers who are not pundits, but Olivelle's translational practices go a long way toward making them more accessible and palatable to non-Sanslcritists than they are in Baler's translations.

When we leave considerations of presentation and general style and focus on the rendering of what the texts say, we enter upon the complicated ways of words, which, semantically, are like jigsaw-puzzle-pieces in their constrictions and expansions of reference and signification. (18) It would be a pleasure to discuss the intricacies of Olivelle's translations of the dharmasutras at length, but there is not space here to do more than make a few selected observations, which will necessarily focus more upon passages where I disagree with Olivelle in small or large ways than on our areas of agreement.

I start by mentioning a few of the many places where Olivelle's translations are clearly admirable in one way or another. Olivelle's translation of the start of Baudhayana 1.10.1: "Each drop of water when sprinkled purifies an area as large as a cow's hide" is direct and elegant compared to Miller's "A drop of water which is allowed to fall ..." (19) Then Vasistha 11.20:

  If, however, a man who knows the Veda is afflicted
  with defects that defile those alongside whom he
  eats, Yama has proclaimed him faultless; he
  undoubtedly purifies those alongside whom he eats.
  (20)

This is a fine sentiment rendered by Olivelle more accurately (21) and more forcefully than by Buhler.

Before entering upon a few disagreements or suggestions for improvements, I mention the two outright mistakes I have come across: Olivelle inadvertently failed to translate the word buddhi at Baudhayana 1.1.15, and at Vas 30.5 the word prana was mistranslated as "out-breath" and apana as "in-breath." (22) Now to a few interesting differences.

The very first injunction for snatakas in Baudhayana 1.5 ordains for them antarvasa uttariyam, (23) which Olivelle renders as "a lower and an upper garment," just as Buhler did. But it seems clear we have here "an inner garment and an outer garment," that is "underwear" and "outerwear" over it. The commentator Govindasvamin glosses antarvasas with katisatra ('hip-string'), which would seem to imply an attached kaupinacchadana over the genitals. (24) And shortly after, Baudhayana enjoins the outer garment, uttariya, to be made of hide (ajina). Likewise at Vasistha 12.14 and with Apastamba's injunction for householders at 2.4.21: nityam uttaram vasah karyam. (25)

A small change to the translation of Vasistha 21.16 (26) would make it more immediately intelligible. Currently it starts out "If a Brahmin inadvertently has sex with the wife of another Brahmin ..." Of course he did not have sex inadvertently, but was unaware of who the woman's husband was. This small hitch would be ironed out were the translation to read "If a brahmin has sex with a woman he learns afterward is the wife of another brahmin ...," reversing the temporal polarity of the adverb apreksaparvam "not being aware beforehand" to "becoming aware afterward," which the sentence says pragmatically though the word does not.

At Baudhayana 1.1.5 Olivelle renders khalu merely with "Now ...," which is certainly an improvement over Buhler's "forsooth," but which fails to convey the pragmatic force of the statement following the khalu. Baudhayana has begun his work (1.1.1-4) by enumerating the sources of dharma as, first, Veda; second, smarta teaching; and third, sistagama. The text of Baudhayana continues with two statements characterizing the sistas. The first of these (sutra 5) lists various moral virtues they are said to have: sistas khalu vigatamatsara nirahamkarah and more in the same vein. Then a sloka defining sistas in terms of learning and intellectual abilities (sutra 6). Khalu signifies here something along the lines of "you know," or "wouldn't you say (or, agree)," marking the information imparted as not merely factual, but as something the speaker is representing as common--or at least shared--knowledge. Seeking such complicity from the hearer does more than argue the claim that the agama of sistas is a legitimate source of dharma: it seeks to foreclose doubt or dissent ahead of time. (27) "You would agree, wouldn't you that, sistas [are clearly a legitimate third source for knowing what is dharma, since they] are free of envy and arrogance and ..." (28)

11.2 Dhanna and Its Translation in the Sutras

The principal concern of the dharmasutras is to identify the particular forms of action and behavior that are "dharma." Olivelle's presentation of dharma in his introduction mirrors this goal. He lists dharma in terms of those things to which the satras refer when using the word:

  Dharma includes all aspects of proper individual
  and social behavior as demanded by one's role
  in society and in keeping with one's social
  identity according to age, gender, caste, marital
  status, and order of life. The term dharma may be
  translated as "Law" if we do not limit ourselves
  to its narrow modern definition as civil and
  criminal statutes, but take it to include all
  the rules of behavior, including moral and
  religious behavior, that a community recognizes
  as binding on its members. (DS, 1)

This listing is an accurate general representation of the word's pattern of reference in the sutras. It goes beyond what the sutras explicitly tell us, offering an interpretation of what dharma is by proposing the English translation "Law" enlarged by the further specification of that to include everything "that a community recognizes as binding on its members." Later in his introduction (DS, 14-17) Olivelle expands upon this interpretation importantly and provides more detail about the actions that are "dharma." Quoting Wilhelm Halbfass's account of an essential aspect of dharma,29 Olivelle writes that (some of) the earliest instances of the word

  [refer] to the rules and statutes connected with
  the "continuous maintaining of the social and
  cosmic order and norm which is achieved by the
  Aryan [Arya] through the performance of his Vedic
  rites and traditional duties." (DS, 14)

Olivelle amplifies further a bit later, adding that dharma was at first, "and in the Brahmanical tradition continued to be, associated with the ritual. ... dharmas are the rules of correct ritual procedure. ... [later] dharma came to include norms of correct behavior within both the ritual and the moral/social spheres" (DS,14). In addition to using "Law(s)" in his translations, Olivelle varies the rendering of "dharma" at various points with "righteousness, duty, rites" (Baudh 1.7.8), "ritual purpose(s)" (Ap 1.7.18), "merit" (Gaut 11.11, Baudh 1.4.1), etc., and sometimes renders it as the adjectives "righteous, lawful" (Ap 1.7.21), "legal" (Baudh 1.1.8, 13), "equitable" (Baudh 1.18.15), etc., which clearly stretch the implied "idea" of dharma from "binding norms" (that imply punishments for violation), i.e., Law, to "right" (action) more generally, "ritual" (action), merit-producing (action), and socially helpful (action). In his survey of those behaviors to which the word dharma might refer, and in the nuanced range of his translations of the word, Olivelle has given us an excellent registration of the use of the word in the dharmasutras.

Besides labeling some forms of behavior as dharma explicitly, the sutras enjoin behavior as dharma implicitly by specifying some transcendent good the behavior produces for the doer. These benefits are typically gaining, or thriving in, loka(s), svarga, svargaloka, or their clear equivalents (divi, amusmiml loke, amutra, paratra, nakaprsthe), "winning both worlds" (ubhau lokau [abhijayati], "this one and the next"), and, more generally, siddhi 'success', best understood as "getting the supramundane good or capability one seeks." In addition to these locutions, the sutras occasionally use two other words to enjoin specific forms of behavior as dharma: kyema and nikfreyasa, words that sometimes refer to paradise, and sometimes to a good beyond paradise, that is, moksa or its equivalent. Nikfreyasa, however, does not necessarily, and so does not always, refer to a good beyond this world--more on this shortly. Ksema, however, does always refers to a good beyond this world--sometimes following upon doing good deeds and sometimes following upon seeking the Self. Olivelle translates ksema with "bliss" in the seven instances in the sutras where injunctions are said to result in someone's gaining kyema--four of which (Ap 1.23.3, 2.24.14 and 16, and Baudh 2.18.9) enjoin the pursuit of the Self, but three of which promise ksema after one scrupulously lives in (any one [?] or all [?] of) the four agramas (Ap 2.21.2), or observes particular restrictions while teaching the Vedas (Ap 2.5.18), or assiduously performs Vedic rituals (Baudh 2.4.23). Though I would prefer to render ksema with "rest, peace, ease" (which is sometimes absolute--the word can and does come to refer to moksa regularly), the translation "bliss" is a not unjustifiable interpretation of the subjective experience of an absolute and permanent ksema.

But with regard to nikfreyasa as a specification of the good that may come from doing what is enjoined, Olivelle's usual "prosperity" misses sounding the necessary high note. The word nihsreyasa is a term of relation, of comparison, built upon the comparative adjective sfreyas (which is frequently used also as the superlative), and outside the dharmasutras is regularly nominalized as "the better, the good, the best thing (to do)." It is "the good that has nothing better than it," "the highest good," "definitively the best" (niscitam sreyatz). (30) But though the word comes eventually to refer regularly to transcendent goods realized after death (especially the absolute good of moksa), in these texts and in the MBh the "highest good" designated by nikfreyasa is not a one and only summum bonum. It has a wide use referring to the highest good attainable by a particular person in a particular situation, often with implicit comparison to lesser, or worse, possible developments or outcomes of a course of action. The word functions as an adjective signifying the best result, the best advice, etc., extending sometimes, as mentioned, to the referential level of "absolute value," the "absolute reality." (31) While English "prosperity" need not be construed simply in a mundane and concrete fashion--that is, it can signify an absolute prospering similar to "bliss"--I do not hear anything superlative or transcendent in OliveIle's renderings with "prosperity" (see Ap 1.9.13 32 and 2.16.2 33 and Gaut 11.26,34 where Olivelle renders it with either "prospers" or "prosperity"). I suspect that in fact he does intend a "prosperity" that approaches the generality of Friedrich's Glilckseligkeit, "blessedness, beatitude, felicity, bliss," but I think that is a matter of my inferring it rather than hearing it in his translations.

The first instance of this usage shows up on the first page of the entire sutra collection and involves me in a larger disagreement with Haradatta, buhler, and OliveIle all together. The matter is Apastamba's characterization of the four vanjas, which says (as I read it) that "the highest good" (nihsreyasa) attainable by each varnia is relative to its rank, in the order of brahmins, ksatriyas, vaisyas, and sudras.

  catvaoa varna brahmanaksatriyavaisadrah [4] tesam
  parvah purvah janmatah sreyan [5] asadranam
  adustakarmanam upayanam vedadhyayanam agnyadheyam
  phalavanti ca karmani [6] susrusa sudrasyetaresam
  varnanam [7] parvasmin purvasmin varne nihsreyasam
  bhuyah [8] (Apastamba 1.1.4-8)

  The four Social Orders are brahmins, ksatriyas,
  vaisyas, and sudras. [4] Each earlier one of
  these is superior to any following, because of
  birth. [5] For those who are not sudras and
   do not engage in degraded work there are
  Initiation, Veda-recitation, Fire-Installation,
  and Fruitful Rites. [6] For the sudra there is
  Doing the Bidding of the other Social Orders.
  [7] The highest possible good is greater for
  each earlier mentioned Social Order. [8]
  (jlf: Ap 1.1.4-8)

Olivelle rendered nihsreyasam in sutra 8 with "prosperity," after reading this text quite differently from the way I have done. Olivelle, following Haradatta's interpretation of the sequencing of these struas, as did Buhler, rendered the last three words of sutra 6 as a separate sentence and then saw the nihsreyasam in sutra 8 as pertaining to sudras alone. (35)

  "Those who are not sudras and are not guilty of
  evil deeds may undergo initiation, undertake
  vedic study, and set up the sacred fires; and
  their rites bear fruit. 7sudras are to serve
  the other classes; 8the higher the class they
  serve, the greater their prosperity. (olv: Ap
  1.1.6-8)

The "prosperity" here is fine as long as it is seen as being other-worldly as well as this-worldly. Friedrich, who also has read the syntax here against Haradatta's construction, glosses nihsreyasa again with Gluckseligkeit, which fits well the primarily other-worldly sense of the injunction here.

With Vasitha 1.1 (athatah purusanihsreyasartham dharmajijnasa "And after that, an inquiry into dharma for the sake of knowing how to acquire the highest good people may gain" [jlf: Vas 1.1]), Olivelle alters his translation of the word. He comes quite close here to "highest good," with "(for the sake of attaining) the highest goal (of man)" (olv: Vas 1.1; parentheses inserted by j1f).

11.3 Final Observations on the Dharmasutra Volume

There are unfortunately a number of typographical errors, most of which are merely superficial irritants. See the end of the review for a list of the more serious ones. There are also some minor annoyances. One involves substance: Olivelle has relegated to his notes on the Baudhayana sutra the listings of each chapter's contents in the form of pratikas, giving, in a reversed order, each sub-section's first "sfitra."36 These content listings occur in most mss, and in both of Hultzsch's editions of the sutra, and they are a basic feature of the text's oral construction and transmission. They should be immediately visible on the surface to all who consult the Sanskrit text, no matter how casually; and then their (in my judgment permissible) omission in the translation should be signaled with notes to the translation. A couple of non-substantive annoyances: first Olivelle uses asterisks in this book in two different ways: one--announced in an introductory note on page xvii, but of infrequent occurrence--signals that Olivelle is quite uncertain of the reading at a given point. The other--not announced, but occurring frequently in the translated text--signals the presence of an endnote annotation that discusses the content of the text. Secondly, I want to object to a practice of OliveIle's that I find problematic. There are times when OliveIle's translation is knowingly based on a reading different from the Sanskrit text on the facing page (e.g., at Gaut 1.27). In these instances Olivelle is translating a reading that is provided only in his notes. 37 And while these discrepancies are signaled, the signaling method used is not unique to these departures: the "discarded reading" is flagged in the Sanskrit text, but only with the regular marker for notes. When initially reading the translation, one has no reason to expect that a note marker will lead to notice of a deviation of the translation from that text, until and unless one is baffled at the discrepancy of the text and the translation. Some special marker should have been used to alert readers to such changes.

Notwithstanding these minor issues and any differences of translational judgment. Olivelle has admirably performed a truly great service with this volume by bringing together in one place a meticulous synthesis of almost one hundred fifty years of scholarship on the early dharmasastra tradition together with his own industry, acumen, and erudition. Olivelle is highly versed in the technical details of this tradition and has combed through the commentaries and ancillary literature alertly, with the result that the volume's many notes are highly informative and bristle with a host of instructive judgments. The ancillary materials are extensive and helpful. This volume is a major contribution and I salute Olivelle for it with gratitude.

III. THE MANAVADHARMASASTRA VOLUME

Olivelle's Manu 's Code of Law begins with a substantial introduction of sixty-seven pages outlining the text and, at some length, OliveIle's understanding of how and when the Manavadharmasastra came to be. As to when: Olivelle dates the Manu to the "2nd to 3rd centuries CE" (MCL, 25), ultimately on the bases of its occasionally levying fines of gold coins, which would make it later than the Kushana ruler Vima Kadphises (ca. 100 C.E., the first known issuer of significant numbers of gold coins), and its representing social and political conditions that seem to reflect the interval "between the empires" of the Mauryans and the Guptas, rather than those of the Gupta empire. This broad dating of the text of Manu is as good as our current knowledge allows. With regard to the compositional process of the text, Olivelle has given us an interesting theory. He sees the text as the product of a single compositional effort by one author working alone or perhaps directing a committee of assistants (MCL, 7). Olivelle traces a series of meta-textual notices of content-transitions that occur at various junctures across the text of Manu38 and concludes from them that the original of the Manu sastra could only have been the creation of a single visionary author who had organized the whole work methodically around an "exquisite structure" (MCL, 19). This learned dharmasastrin synthesized in Slokas the teachings of the now ancient dharmasutras with ideas from the teachings of the arthasastra. In making this argument, Olivelle not only rejects Buhler's conjectural postulation that the Manu sastra was a reworking of an ancient Manavadhannasfitra (as did most Sanskrit scholars of the twentieth century); he also rejected 131ihler's demonstration that a further source of the teachings found in the Manu sastra was an oral sloka tradition of dharma teaching that served as a common source for a number of dharma teachings found in parallel in the MBh and the Manu.39 Olivelle's tracing of the full thread of the meta-textual transitional announcements in Manu is a major advance in our understanding of the text, but it is hardly necessary to attribute them to the same person who composed or arranged the text in the form those statements describe. Nor is his abrupt dismissal of evidence of an oral sloka tradition of dharma teaching common to both the MBh and Manu plausible--he attributes this parallelism instead to a third or fourth century epic committee's confused cribbing from the Manu sastra (MCL, 23). In my considered judgment OliveIle's account of the original composition of Manu is not warranted by the evidence. I shall discuss his arguments for his theory and rebut his conclusions in proper detail elsewhere; for the moment, I will say only that I think that, apart from his belief in a Manavadharmasastra as the first stage of the Manavadharmasutra's existence, Buhler's understanding of the genesis of Manu is much closer to what likely happened.

This general introduction is followed by Olivelle's excellent new English translation of the text, his copious annotations to the translation (114 pages), a separate introduction of twenty-six pages to the critical edition of the text (somewhat too brief, I think--see below), the Sanskrit text with critical apparatus at the foot of the pages, and then an abundance of critical notes to the text and its edition (sixty-nine pages). These major components of the work are followed by thirteen pages of appendices (fauna and flora; names of gods, people, and places; ritual vocabulary; and weights, measures, and currency), ten pages of bibliography, more than twenty-five pages of dharmagastric parallels, a pada index in more than seventy-five pages, and, finally, a twenty-page index to the translation. Preceding and interspersed among the major components are various shorter notices: a preface, a list of abbreviations, an editor's outline of the translated text, notes on the translation, notes on the critical edition, and an extra appendix of fifty additional verses found in several manuscripts of chapter 7 (immediately following chapter 7 in the edition). It is also worth noting that OliveIle has frequently translated (in the notes to the translation) material from the textual apparatus of his edition, that is, extra verses that were widely attested and significant variant readings--commendable supererogation. All of this is presented on the page very attractively, especially in terms of the large Devanagari font used to set the Sanskrit text.40

111.1 The Written Tradition of the Manavadharmasastra

Olivelle rightly emphasizes, more than once (e.g., MCL, 378, at some length), that one of the most important tasks involved in the critical edition of an old text is the registration of the written tradition from which the text is recovered: that is, documenting a text's life after the original manuscript leaves the hand of the last authorial or editorial agent--the historically critical sub-discipline of "bibliography."41 I strongly second this point, for such tradition is the living tissue of the intellectual history we need to know in better detail: it is the basic form of the life of the text itself. One of the greatest advances of Olivelle's edition of the Manu sastra is his gathering of a far wider and deeper representation of that tradition than has been done before, and his making it possible to focus our attention upon it with greater resolving power than ever.

There have been numerous editions of the Manavadharmasastra previously, (42) but only that of Julius Jolly in 1887 was in any way a critical edition: it was based on twenty manuscripts, seven commentaries, and five previous editions. (43) Olivlle's new edition is based on a survey of ninety-one manuscripts (44) and the full collation of fifty-three of those for use in this new edition. Olivelle made use of nine commentaries on the text (including those of Bharuci, Ramacandra, and Manirama, which were unavailable to Jolly). He also consulted the citations of the Manavadharmasastra in twelve other works or authors that frequently cited it and kept a referential eye on the manuscript testimony of five previous editions of the text. Olivelle correctly eschewed including as testimonia shared verses found in the Mahabharata and other dharmasastra texts, as their exact status in relation to the written tradition of Manu cannot be ascertained.

Of the fifty-three manuscripts ultimately used for the edition, nine were written in southern scripts: four Malayali, three Grantha, two Telugu. Among the forty-four northern-script manuscripts used, Devanagari dominates overwhelmingly, though not so thoroughly as it did with Jolly's edition: thirty-one Devanagari, three Bengali, three Newari, three Old Nagari, and two each in Oriya and Sarada. (45) Of the fifty-three only seventeen have clear dates. In addition, the colophon of one ms., 8Kt5, has a date of "Sam 388," which must be a scribal or typographical error for "Sam 1388" (1331 c.E.). Since Olivelle neither comments on this anomalous number nor indicates "no date" in his description of the ms., it would seem to be a typo; and thus we would seem to have one more old, dated manuscript as wel1. (46) Thus ten of the dated mss, are between 130 and 300 years old and eight are more than 300 years old. Of these eight, three are more than 550 years old: a Devanagari manuscript from Lahore (La1) dated 1450 C.E., (47) the Bengali manuscript from Kathmandu (BKt5) seemingly dated 1331 C.E., and a manuscript written in Old Nagari located in Kathmandu (NKt4) dated 1182 C.E. Of the dated manuscripts Jolly used (all in Devanagari), one went back to 1497 and another to 1591with the others being basically seventeenth and eighteenth century. (48)

Thirty-two of the fifty-three manuscripts used are complete. Unfortunately eleven of the fifty-three are incomplete in substantial ways, although only two of the dated manuscripts (Lo4 and mT[r.sup.3]) are so. Seven of the remaining forty-two (including two of the three oldest and three other dated ones) are virtually complete, with occasional lacunae (though, as the sixteenth-century BCa's chapter I was so torn it could not be collated, it might be reckoned as functionally incomplete). Problematic is the fact that three manuscripts have no indication of completeness or incompleteness (see n. 46 above). Olivelle regularly gives notice in the edition's apparatus for each stanza of any manuscripts in which some or all of the sloka is missing, (49) but these notices apply only to those manuscripts that contain that portion of the text in the first place. Notice of the absence of given manuscripts for whole chapters or significant portions of chapters is not furnished in the apparatus, only in the description of manuscripts in the introduction--making full critical use of the apparatus very inconvenient. A comprehensive tabular presentation of this information would make the use of the apparatus far easier, and I offer a substitute for that in the note below. (50)

111.2 The Complex Apparatus: Observations and Suggestions

Such a large and complex population of witnesses poses further challenges to users, some of which Olivelle has nicely anticipated and ameliorated with two remarkable innovations. Commendably "[d]eparting from normal editorial practice, [Olivelle has] articulated [his] reasons for adopting a particular reading in the endnotes" (MCL, 375). To his great credit he has done so often. In addition, departing commendably from the norm again, Olivelle often offers his users, in the endnotes to the edition, the convenience of a positive listing of the manuscripts that contain the reading he has chosen--this in addition to the usual negative listing of the apparatus, which is often quite long and difficult to construe meaningfully. (51)

I would, however, suggest that in future editors with similarly extensive witness populations devise a more user-friendly form than Olivelle has used here for listing such a complex apparatus. His system of manuscript sigla (52) makes it more difficult to track the collocations and filiations of the witnesses in the apparatus than that need be. At the very least the manuscripts of the variants should be reported in terms of the families the editor has distinguished (NT [NT-x or NT-y] and ST, see below), using a regular order within those families for the reporting and, perhaps, using classificatory markers pre-pended to the lists (N, Nx, Ny, and S respectively). More radically, perhaps the sigla of the manuscripts themselves would be better if they reflected their classification rather than the details of their locations and their scripts. Thus, instead of GM[d.sup.1] [.sub.T]M[d.sup.3] [.sub.T]M[d.sup.4] [.sub.G]M[d.sup.5] [.sub.G]My T[r.sup.1] [.sub.M]T[r.sup.4] [.sub.M]T[r.sup.5] [.sub.M]T[r.sup.6] for the nine ST manuscripts, their indications might read simply as [S.sub.1] [S.sub.2] [S.sub.3], etc., to [S.sub.9]; and instead of Hy jm J[o.sup.1] K[t.sup.2] [.sub.o]m[d.sup.2] T[j.sup.2] [.sub.M]T[r.sup.3] for the NT-y manuscripts, N[y.sub.l] N[y.sub.2] N[y.sub.3], etc., to N[y.sub.7] ; and likewise for those of N[T.sub.-x], as N[x.sub.1] to N[x.sub.37]. (If the manuscripts within the respective families can be ranked for their relative quality, then labeling or ordering witnesses accordingly would be even more helpful--of course, establishing a set of ranking claims is a matter of a different sort altogether.) Obviously a manuscript's location and its script are critically important matters of its existence, but this information can be laid out once and for all in the description of manuscripts. Some might think the indications of scripts in the sigla of the apparatus too important to discard; if this were the consensus, the above suggestion could be modified by adding script designators to the sequence numbers (thus, for ST, following the current order of those manuscripts: [S.sub.IG] [S.sub.2T] [S.sub.3T] [S.sub.40] [S.sub.50] [S.sub.6D] [S.sub.7M] [S.sub.8M] [S.sub.9M] or N[y.sub.ID] N[y.sub.2D] N[y.sub.3D] N[y.sub.4D] N[y.sub.5D] N[y.sub.7M]), but it seems to me that a manuscript's current location (obviously not the same thing as provenance), which is here the controlling factor of the alphabetic ordering of the witnesses in the apparatus (though it is often in the second place in the siglum), can be dispensed with in the apparatus altogether. The apparatus should be about witnesses, not manuscripts, (53) and with an apparatus as large as the one here it is more difficult than necessary to process the lists in a meaningful way.

Above and beyond the small problems noted, the "bibliographic" element alone of Olivelle's work--its reaching farther back in time with older manuscripts than ever used before, and its putting the southern tradition of Manu on the record in a serious way--has made an extremely valuable contribution that documents the Manu tradition far better than it had been previously. Most significant, perhaps, has been Olivelle's use of manuscripts currently resident in Nepal, which include one Newari manuscript over 830 years old and a Bengali manuscript almost 700 years old. (It is highly lamentable that Olivelle's efforts to secure various Kashmiri manuscripts came to naught, as are the lamentable forces that obstructed him [MCL, 367].) Olivelle has recorded the tradition far more completely and soundly than did earlier practice, and he is to be saluted for performing a highly important service to the study of the intellectual history of pre-modern Indian civilization.

111.3 Establishing the Text of the Manavadharmasastra

Like Sukthankar's edition of the MBh, (54) Olivelle's edition of the Manu is an eclectic edition, for it was determined that it was not possible to arrive at a rigorously established stemma codicum. (55) But, as with the MBh, the edition is no less "critical" on this account, the essence of the description "critical" being "the criticism," the exercise of informed judgment on the basis of the best ordering possible of as extensive a pool of witnesses as possible. Most of Olivelle's description of the witnesses and his resulting editorial process is based upon his seeing all of them as more or less equally contaminated and so their readings as thoroughly hybridized. He reported his survey of the fifty-three witnesses in these words:

  [MCL, 370] It is not possible to constitute a
  text such as the MDh purely through the classical
  (56) editorial methods dependent on isolating
  independent manuscript traditions. The MDh was too
  widely known to permit the existence of totally
  isolated and independent traditions ... In the
  process of collation it became apparent that all
  extant mss, of the MDh are at some level hybrid;
  over time readers and scribes appear to have
  changed the readings of one tradition according
  to those of another. It is, therefore, impossible
  to isolate different [371] recensions of the
  text, if we mean by "recension" versions that
  had a more or less independent life.

But he does not demonstrate the warrants for this judgment with any details from his study of the witnesses.

Olivelle's survey of the witnesses did allow him to discern three broad families among the manuscripts, no one of which could be uniquely authoritative. He found a Southern Transmission, "ST,"57 and a Northern Transmission. The latter subdivides into a "Northern Traditional" one, "NT-x," "the older and the main branch . ... followed by most major commentators: Medhatithi, Govindaraja, and Nardyana" (MCL, 372), and the "Northern Vulgate," "NT-y," that Olivelle believes probably "arose in the late medieval period, possibly under the influence of Kulluka ... or those under the influence of his well-known commentary," and which Olivelle judges to be an inferior version of the text (MCL, 373).58 This classification of witnesses is a critically important advance in our knowledge of the tradition of the Manu, one arising directly from Olivelle's diligent assembly of his extensive witness-pool and his careful scrutiny of it. But while Olivelle demonstrates that ST and NT are distinct (MCL, 371-72), he does not make clear in his introduction the general nature and extent of the differences between the NT-x and NT-y witnesses. Nor does he explain how "the vulgate" transmission, NT-y, is "the least faithful and contains the greatest number of corruptions and deviations from what must be presumed to be the old readings based on the evidence of 'NT-x' and 'ST."If one then asks what value the NT-y witnesses might have for constituting the edition, an answer is found when Olivelle writes that "In numerous places the NT-y has unique readings different from either the NT-x or the ST" (MCL: 373), and shortly after making this point he give us another reason, telling us (MCL, 374) that there are cruces where NT-y agrees with ST against NT-x. Unfortunately Olivelle does not list or discuss either of these two sets of readings that demonstrate NT-y's distinctness or value.

Olivelle describes the general decision tree he employed in his editorial elections as follows.

  The constitution of the text ... boils down to
  the editorial judgment as to which of two or
  three competing readings has the greater claim
  to be accepted as the original. ... I. At the
  most general level, a reading supported by the
  ST and the NT is adopted. ... Any variants
  restricted to a limited range of mss. must have
  been introduced by a scribe or reader. Clearly
  the vast majority of readings in the critical
  edition fall under this rule. 2. Readings
  supported by the ST and NT-x are adopted in
  preference to those supported only by NT-y ...
  [for] [t]he superiority of the former transmissions
  is clearly established by a close scrutiny of
  the critical apparatus. ... [And, moreover, these]
  are the ones established in the medieval literature
  used as external testimonia. ... 3. When ST and NT
  differ, or when ST and NT-y disagree with NT-x,
  "it is not possible to formulate a general and
  rigid principle. ... I have analyzed each case
  separately, taking into account the evidence
  provided by the commentaries and testimonia, the
  general sense of the text, and parallel passages
  within the MDh. 4. The most significant editorial
  principle is that of lectio difficilior. ... 5.
  In only a handful of instances have I rejected
  the nearly unanimous readings of the mss. in favor
  of the reading attested by the commentators."
  (MCL, 374-75)

Regarding the first of these principles, I would have liked a fuller discussion somewhere of what "supported by the ST and the NT" mean in practice--"support" in a population of witnesses can mean anything from half a dozen witnesses to near unanimity in the whole population. And with regard to the second and third principles, again one wishes Olivelle had demonstrated the superiority claimed for the NT-x and ST transmissions and that he had told us more about the agreements between ST and NT-y and their disagreements with NT-x. One wishes he had either developed a more ramified stemma, grounding some ranking of manuscripts within the different families, or demonstrated the impossibility of doing so. I would also suggest, in connection with the third principle, that parallel passages within a text do not furnish any certain standard of a reading; there is often a complexity and multivocality at the base of Brahminic intellectual traditions that make it just as likely that the same or similar ideas were recorded in multiple wordings, rather than that one wording has been distorted through errors in the written transmission. Though the principle of lectio difficilior has come under some suspicion of late, (59) I still fully agree with the general use of it for the editing of old Sanskrit texts.60 Finally, I would like to have seen further general discussion in the introduction of the cases mentioned under the fifth item above, for what Olivelle describes there is quite a radical measure.

Though I wish I could know better all the grounds for Olivelle's conclusions and procedures vis-a-vis the witnesses, when it comes to constituting the text on the foundations he has erected, the overall result is a very gratifying improvement over the Jolly edition. The differences between Olivelle's constituted text and Jolly's are not drastic (though many of the points of difference imply interesting and important points of intellectual history), but there are a number of real improvements (many of them based upon Olivelle's marshalling evidence from his newly discovered ST and his newly utilized oldest witness, NK), and they all rest upon a much better and more accessible representation of the written tradition of the Manavadharmasastra. Lastly, his newly constituted text is accompanied by Olivelle's discussing and documenting in detail and with extensive knowledge of the dharmasastra tradition many relevant considerations for the choices of reading he made.

On the basis of their absence from all ST mss., a number of stanzas that were provisionally included in Jolly's edition have been eliminated in Olivelle's: stanzas 5.6lab and 62ab of the Jolly edition are now gone and Jolly's 5.61cd and 5.62cd form OliveIle's 5.61;61 an extra unnumbered stanza between 7.85 and 7.86 in Jolly's edition is gone, and also 11.6, 11.52, and 11.248 of Jolly's edition. Among a number of readings that have been improved over Jolly are the reading at 3.95b of agoh ("a poor man") in place of guror ("one's teacher") on the basis of several ST witnesses and commentarial notices;62 at 3.114c Olivelle reads atithibhyo 'nvag eva ("right after the guests") instead of Jolly's atithibhyo 'gra eva ("even before the guests"), referring to the feeding of guests and any "newly married women, young girls, the sick, and pregnant women" who may be present at the time; at 7.164c Olivelle reads mitrena caivapakrte ("when an ally has initiated the offensive," olv) for mitrasya caivapakrte ("when an injury has been done to an ally"), a characterization of an occasion for a king's going to war; at 9.19d Olivelle reads akrtim ("appearance; representation, specimen"; Olivelle's "sample" is good) for niskrtih ("expiations") as the direct object of srnuta "listen to," in svalaksanyapariksartham tasam ca s'srnuta, an intricately difficult passage purporting to provide "scriptural" proof-texts demonstrating the presumed evil nature of women. Numerous other improvements over Jolly's edition could be noticed.63 But Olivelle has made some changes where I think Jolly was better (e.g., at 2.73a Olivelle rejects Jolly's adhyesyamanamtu gurur, in which a teacher commands his pupil to recite and later to stop reciting, in favor of adhyesyamanas tu gurum, in which the pupil will command the teacher to teach and then the teacher will command him to stop), or has retained Jolly's reading and rejected something better (e.g., in quite a complex case Olivelle keeps Jolly's krtyesu ca cikirsitam at 7.67d, though I think the krtyesu ca cikirsitam read in all ST mss. along with NK t4 and a few others from NT-x is likely better).64 Finally, though there is of-ten some room for debate at many of these cruces--given that the witness pool is so relatively undifferentiated--clearly there is no one better than Olivelle to have been in the seat making these editorial decisions for this text: he has significantly improved our text of Manu and greatly advanced our knowledge of it.

III.4 The Life and After-Life of the Text

I will close this discussion of Olivelle's Manu with further consideration of the "afterlife" or tradition of the text. Olivelle believes his constituted text carries within itself much that is actually part of the "after-life" of the author's original autograph manuscript. But before describing this theory of his, let us take stock of the basic facts of the text.

The Manavadharmasastra clearly was a deliberate creation: the final result is quite well organized overall and bespeaks an ambitious intellectual project significantly larger in scope than anything seen in the sutras. It clearly was some kind of remaniement of old traditions within some kind of institution dedicated to preserving and perhaps propagating them both orally and with letters. The question about authorship is whether this remaniement resulted from the efforts of some number of scholarly composers-editors-redactors working in relays over a relatively long period of time, now copying and rewriting the work of earlier scholars, as well as holding much old and new in their memories at the same time, or whether it happened all at once and in a short period of time, through the efforts of one man rearticulating, that is recomposing, more or less the whole of the old oral tradition and fusing it with much that was new, in continuous slokas. The former alternative is more likely in my judgment, in part because I am convinced that there existed oral sloka traditions of dharma instruction, which imply some kind of persisting scholarly institution. Some members of this tradition incorporated writing into their work at some point in time, with all its new possibilities and limitations. At the same or a different moment some authorial or redactorial agency within that tradition did transform the teachings of the tradition into some version of an ancestor of the text Olivelle has reconstituted from today's available copies.

Clearly that written text was relatively fixed as a premise of its creation and use, and also because the great majority of it was copied in writing more or less faithfully across many centuries. But was this implicitly fixed text also a closed text, that is, one that was considered by the tradition to be perfect and complete, one for which the process of transmission was made to be strongly resistant to additions and improvements, like the Veda? It seems such closure was not a normal view of a non-Vedic text. Insofar as words and texts are essentially matters of oral production and aural experience, written symbols on a physical medium are a reduction of the essence of the words, are fundamentally incomplete. And insofar as the face-to-face production of words and teachings normally implies that many voices present contribute to the conversation, that the same point will often be repeated in other words, that further points and qualifications of what is said may be added out of order in a discussion or lecture--to that extent, written versions or records of "samvadas" (literally a "speaking together"; often "conversation, dialogue") are obviously even more limited, more incomplete. We should, I suggest, imagine the adaptation, creation, use, and modification of most Brahmin intellectual discourses to and in writing as occurring within such samvadabased institutions. At times the consternation some earlier Western Indologists expressed with Brahminic textual practices65 is the result, I think, of a lack of alignment of the tacit premises operating in ancient Brahmin and modern Western intellectual discourses. 66 Within the context of a tradition that prizes the face-to-face oral articulation of primordial truths, it is these samvadas that are sacrosanct and not written copies of them; written records of such teachings will always be regarded as inherently incomplete and in need of corrections and additions.

Apart from the issue of how the text of the Manu first came to be. Olivelle argues that, for at least the first several centuries of its existence, the author's text existed within a tradition that, according to my terms above, either did not regard it as complete and closed (likely) or had no effective measures of transmission to enforce that idea. Having edited the text by principles of lower textual criticism, Olivelle turned to higher criticism and argued that the text of the Manavadharmasastra was subject to many additions, growing by over 16% of its original size, until it was "fixed" through the agency of its first commentator. 67 In that process the author's "exquisitely structured" composition was obscured by these interpolations and a clumsy segmentation of it into its current twelve adhyayas (MCL, 7, 52). By this higher critical reading of the text, Olivelle determined that his constituted text contained twenty-six passages, which he labeled "excursuses," that were extraneous to the organization and flow of instruction in the putative original Manu sastra, passages such as a somewhat redundant "Second Account of Creation" at 1.32-41, the account of "Cosmic Cycles" at 1.51-57, the treatment of the "Property of Minors and Women" at 8.27-29, and pronouncements regarding "Lost and Stolen Property" at 8.30-40.68 These twenty-six excursuses comprise altogether, by my count, 374 slokas of the constituted text's 2680, 14% of the text.69 Olivelle believes these excursuses were composed at various times and places subsequent to the completion and dissemination of the putative autograph-archetype of the Manavadharmasastra and were then absorbed over time into all copies of some post-archetypal authoritative text of the sastra by way of what must be imagined as a very extensive and homogenizing process of manuscript circulation. This process would consist of any and all corrections and additions that were made to authoritative copies of the text as isolated "improvements" or "completions" becoming more or less knowingly sanctioned or favored by the elite members of the expert dharmasastra tradition. These improvements would then have made their way into all new normative copies of the text, eventually replacing, in the queues of manuscripts to be copied, all earlier copies that lacked them. Such a process continued, Olivelle believes, until a normative text of the sastra was "fixed," in conjunction with its becoming the object of its first written commentary, a development supposedly making such improvement of the text more difficult. It is conceivable that in the course of this general process some important member of the tradition, perhaps the first commentator himself, may have deliberately composed (or collected, edited, and inserted into his copy) a number of these excursuses as a designed supplement or updating of the text--a new edition that was then absorbed throughout the rest of the tradition by way of the same homogenizing process. But, Olivelle goes on to note, even after the claimed fixing of the text "[such changes or additions to the text] did not cease completely. Changes after that period, however, were limited to the addition of individual verses and minor changes in the wording of verses detectable through 'lower criticism'" (MCL, 51).70

Most of what Olivelle argues about the excursuses is quite plausible, if we envision dharmasastra scholars in regular communication with each other and engaged in my conjectured scholarly samvada across space and time that made critical use of, and renewed, written versions of its texts.7I Such a homogenizing process of circulation is the only way, in general, to view some fragment of a written text that is found in all extant witnesses to be, in fact, a post-archetypal addition to the text.(72) But why should the fact that one member of such a tradition puts his interpretive comments on that text into written form occasion the closing of his mula text to further sanctioned additions? And what evidence is there that such a closure of the text was ever effected? All we really know is that all eventually pandemic, post-archetypal additions to the text came into existence in the 900-to-1000 year period ending with our oldest dated manuscript, [.sub.N]K[t.sup.4] of 1182 C.E. (though of course we cannot know whether they had reached all copies of the text by then, and we will never be able to say with certainty that there do not exist any copies without them). Olivelle's belief that this development made it more difficult to substantially alter the text of Manu seems based upon the notion that the circulation and copying of the text-cum-commentary operated either physically or psychologically to stabilize the mala text. It is conceivable that the creation of such a new edition under the aegis of a great scholar's name (perhaps Bharuci's) was symptomatic of an institutional change in the tradition. What might have been formerly a written text of the Manavadharmarmasatra used by scholars trained in a fundamentally oral tradition of face-to-face discourse was now the written mrila being interpreted by a great scholar whose "disclosure" (vivarana) of the text is now distributed in writing among and discussed by interested scholars. I am suggesting that perhaps the development of written commentaries either bespeaks or occasioned a shift of focus that pushed the text of the Manu into the background of a tradition that was becoming, or had become, a thoroughly writing-based tradition.

Of course the alternative to Olivelle's argument about the excursuses is that these twenty-six passages obscuring the "exquisite structure" of the text were part of the original archetype of the Manu gastra promulgated by whatever dharmagastra sanwilda (symposium) may have created the text in the first place. As I think it likely that the production of the original text occurred in such a communal, oral-based setting, I think this alternative is more likely. I agree with OliveIle about the existence of a clear structure in the text and I agree that subsequent overlays have obscured it, but I do not think it was ever as exquisite as OliveIle says. Digressions, additions, qualifications, and tangents that obscure the linear continuity of arguments are part of the normal record of texts created in contexts of oral discussion. It is of course possible that the text emerged gradually as a written text in the context of a complex dharmasastra tradition without any dramatic intervention by a single individual and continued to develop according to the mechanism sketched above.

As a final consideration of these matters, I question whether the written tradition of the Manu gastra is as thoroughly unified and as relatively homogenous as Olivelle and others have represented it to be--whether that homogeneity might arise from a highly conservative transmission of its first archetypal manuscript or, rather, as a consequence of some sanctioned text within a more dynamic tradition (see above) becoming relatively homogeneous and closed. I am disposed to be skeptical of such claims of homogeneity, but the general impression the text makes in Jolly's and Olivelle's editions does seem consistent with these claims. There is one way to learn exactly the extent of its homogeneity and, if it is less uniform than appears, also to discover the filiations among the newly discovered branches within the pool of witnesses: analysis of the variants of the witnesses with cladistics software. Such an analysis would yield a cladogram,"branch-diagram," showing where, how, and the degree to which different witnesses are close or far from each other in their variants. If the tradition is as thoroughly homogenized as it appears to "the naked eye," that appearance would be confirmed as fact; of course the hope is that the tradition is not as homogeneous as it may look on the surface and that principles for ranking some manuscripts above others systematically might appear. At the very least, such an analysis would highlight the passages where there is the greatest range of variation and the rosters of witnesses that align with those variants and, we would hope, patterns in those rosters. Of particular interest would be patterns of "additional" passages and "omitted" passages. (73) Even if no set of manuscripts were to be revealed as most proximate to a putative archetype, the editor forced to choose among variants might have more patterns at his or her fingertips to inform the frequently difficult elections that must be made on intrinsic grounds. Of special importance in Olivelle's elections was the pattern of complete agreement among available ST manuscripts, and another was the testimony of the oldest manuscript in the pool, [.sub.N]K[t.sup.4] from 1182 CE, and its agreements with ST manuscripts. It would be an advance if the apparent soundness of these patterns were to be strengthened by cladistic analysis or, even better, if other reliable patterns could also be discerned. There is no way to know at this time whether a more detailed knowledge of the witnesses might make any substantial difference in the constituted text, but more detailed knowledge is the next step to take in the charting of this tradition. (74)

This is a review article of Dharmasiitras: The Law Codes of Apastamba, Gautama, Baudhayana, and Vasistha. Annotated Text and Translation. By Patrick Olivelle. Sources of Ancient Indian Law. Delhi: Motilal Banarsidass, 2000. Pp. xviii + 767. Rs. 1295. And Manu 's Code of Law: A Critical Edition and Translation of the Manava-Dharmasastra. By Patrick Olivelle, with the editorial assistance of Suman Olivelle. South Asian Research. New York: Oxford University Press, 2005. Pp. x + 1131. $125.

(1.) Abbreviations used: Gaut, Ap, Baudh, Vas refer to the Gautama, Apastanzba, Baudhayana, ana vasistha Dharmasatras respectively; DS = Olivelie's Dharmasatras under review; JIP = Journal of Indian Philosophy: jlf = James L. Fitzgerald; MBh = Mahabheirata; MCL = Olivelle's Manu's Code of Law under review; Manu, Mama sastra, and MDh = Manavadharmadastra; olv = Olivelle; ZDMG = Zeitschrift der deutschen morgenlandischen Gesellschaft.

(2.) Patrick Olivelle, Dharma: Studies in Its Semantic, Cultural, and Religious History, vol. 32 (2004) of JIP.

(3.) Patrick Olivelle, "The Semantic History of Dharma: The Middle and Late Vedic Periods," JIP 32 (2004): 491-511.

(4.) Olivelle reckons Apastamba to have been composed about 300 B.C.E. (DS, 10) and Manu about 200 C.E. ("2nd to 3rd centuries C.E.," MCI.. 25).

(5.) Among other criticisms of Fuhrer's editorial work on Vasistha, Olivelle strongly suspects that the "very old manuscript K" that Fuhrer said he used for his second edition was fictitious, since, Olivelle tells us, its readings correspond closely to the readings von Biihtlingk suggested in an 1885 review of Fuhrer's edition in ZDMG (39: 481-88); see DS, 631.

(6.) In his notes to Vasistha Olivelle several times references "Falk, 1999," a lecture handout of Falk's that presents new readings from thirty manuscripts. See Harry Falk, "Review of Patrick Olivelle, The Law Codes of Apastamba, Gamma, Baudhayana, and Vasigha," Orientalische Literaturzeitung 98 (2003): 568. By inadvertence this resource was not formally noted in Olivelle's introductory review of the edition of Vasistha nor in the bibliography. See too DS, 631-33.

(7.) The text of Apastamba presents a number of interesting issues of language that have been discussed for more than a century. See DS, 471, Georg Baler, "Einige Bemerkungen zu Bohtlingk's Artikel fiber Apastamba," ZDMG 40 (1886): 527-48, and also two important articles of Olivelle's that discuss Buhler's and his own practice of relying upon the commentator Haradatta: "Unfaithful Transmitters," JIP 26 (1998): 173-87, and "Sanskrit Commentators and the Transmission of Texts: Haradatta on Apastamba Dharmasutra," J1P 27 (1999): 551-74.

(8.) Harry Falk's suggestion that the standard editors of each sutra deserve a mention on the title page of the book is not without some merit. See Falk, op. cit., 568.

(9.) There are times--such as at DS, 467-70, when reproducing Buhler's (and others') descriptions of manuscripts and Buhler's description of their genealogy--that quotation marks would be useful to show exactly where Buhler's (and others') own words begin and end, but in spite of this occasional fuzziness, their words are attributed to them explicitly.

(10.) I write "sutras" since these texts, as we have them today, often mix genuine sutras, non-sutra prose, and verses together in the flow of their statements.

(11.) Many of Buhler's "sutras" are in fact verses that have been included in the text with or without explicit announcements of the fact.

(12.) Olivelle has kept the same overall numbering of the higher-level divisions of the texts as established in the nineteenth-century editions and Buhler's translations, but he has avoided the past practice of including running reference to the parallel segmentations of the patalms in Apastamba and the adhydyas in Baudhuyana. He observes that while "[e]xamining manuscripts of all the Dharmasutras, I have found that division and numbering of sutras are very inconsistent" (p. 531). and in fact those inconsistencies are often observed in the printed editions as well. Olivelle also preserves, for the most part, the same lowest-level numbering of the discrete units of the texts (the "siitras") as Buhler's translations, which is helpful. But I wish Olivelle had maintained the Baler translations' full set of textual segmentations, the patalas in Apastamba and the adhyayas in Baudhayana, for one misses them when consulting various printed versions of the dharmasutras, which is where most of us find the commentaries.

(13.) Sections distinguished with centered headings all in capitals (e.g., "KING" before Apastamba 2.25.1), with, one step down from there, centered and capitalized headings (e.g., "Royal Fort" before Apastamba 2.25.2, "Protection of Subjects" between 2.25.

(14.) and 2.25.15, etc.), with, sometimes, a further step down, left-justified, bold-font, capitalized headings at the beginning of paragraphs consisting of either several numbered "sutras" (e.g., "Appointment of Security Officers" before Apastamba 26.4) or only one such unit (e.g., "Fall of Women" before Gautama 21.9). 14. Elvira Friedrich, Das A. pastamba-Dharmasatra: Aufbau und Aussage (Frankfurt am Main: P. Lang, 1993).

(15.) See Friedrich, pp. 8-9. Though based on an interesting idea that was executed with care and precision as far as she went, Friedrich's approach to Apastamba was not as radical as she suggested she wanted it to be. She failed to acknowledge formally the extent to which our basic understandings of these texts, and the lexica we employ to construe them, depend upon the learned scholarly traditions of Sanskrit in general, and upon Haradatta and dharmasastra commentators like him in particular. These traditional understandings fundamentally condition our sense of dharmasastra through the scholarship of Buhler, Kane, and the other pioneers of the Western-inspired study of it--scholarship that she knows well and follows often enough, even as she also relies on lexica that also have a significant grounding in traditional scholarship.

(16.) And, for scholars who may wish to grapple with the original texts, Buhler's translations will remain important resources in the long-term scholarly effort to understand the ambiguities and other obscurities embedded in them, given all the specialized scholarly apparatus these translations contain.

(17.) The result is the "accessible English" Olivelle promised in his Foreword (DS, v), which has nothing to do with its not being German (see Falk, op. cit., 567), for their standard earlier translations are in the English of Georg Buhler (The Sacred Laws of the Aryas: As Taught in the Schools of Apastamba, Gautama, Vasishtha, and Baudhayana, 2 vols. [Oxford: Clarendon Press, 1879 and 1882]). One implication of Falk's remark on this, that the fund of German Indological scholarship is often inaccessible to many in the Anglophone scholarly world today, is, lamentably, all too true, though not for want of effort on the part of some like Patrick Olivelle.

(18.) And at the moment of "utterance" they are volatilely dynamic components of the pragmatic stream of signification, which is radically truncated for readers of written texts far removed from that moment.

(19.) Baudh 1.10.1: gocarmamatram abbindur bhumeh sudhyati patitah. Olivelle's extrapolation of abbindur to "each" (and every) of many water drops is clearly implied in the Sanskrit and presents the generality of the point with appropriate immediacy and clarity. And in the context his rendering of the causative ppl. patita as "sprinkled" is not only permissible but called for; Baler's "allowed to fall" is not accurate.

(20.) atha cen mantravid yuktah sariraih panktidasahaih / adusyam tam yamah praha panktipavana eva sah II Vas 11.20.

(21.) The dusaoas are attributes that one may have that will defile, "spoil," others whom the afflicted person may approach, as Olivelle points out in a note to Ap 2.17.21. According to Vas 11.19 these are such conditions as blindness, leprosy, impotence, etc. They are not, as Buhler renders, merely attributes that "exclude" the afflicted from the commensality of the graddha.

(22.) See Kenneth G. Zysk, "The Science of Respiration and the Doctrine of the Bodily Winds in Ancient India," JAOS 113 (1993): 198.

(23.) Baudh 1.5.2.

(24.) See Gaut 3.18 and Baudh 2.11.19, which enjoin the breechclout for the possession-less bhiksu and the parivajaka respectively.

(25.) The scandal repeatedly emphasized in the MBh of Draupadi's being brought into the sabha while merely ekavastra was that she was clad only in her 'underwear,' that is, in something like a petticoat covering her from the waist down (ekavastra adhonivi, 2.60.15) or chemise that was sufficient clothing for women in the women's apartments of the palace. Her wearing only one piece of clothing was not specially related to her menstruating, as van Buitenen claimed (n. to 2.60.15). At MBh 11.9 many of the women who were suddenly turned out of the women's quarters to go through the city and out to the battlefield the morning after the Bharata war had ended were ekavastra.

(26.) brahmanas ced apreksapurvam brahmahadaran abhigacchen nivrttadhartnakarmahah krcchro 'nivrttadharmakarmano' tikrcchrah II Vas 21.16.

(27.) See Murray B. Emeneau, "Sanskrit Syntactic Particles--kilo, khalu, nunam," Indo-Iranian Journal 11 (1969): 251-62, in which he showed that the standard Western lexicographical glosses were insufficient.

(28.) For strong doubt regarding the legitimacy of sistacara or sadacara as a source of dharma see MBh 12.254.20ff.

(29.) Wilhelm Halbfass, "Dharma in Traditional Hinduism," in India and Europe (Albany: State University of New York Press, 1988), 315-16.

(30.) Ketfika ad Papini 5.4.77.

(31.) Thus, while largely true, Hacker's connecting dharma to "salvation" (Heil) in an essential way is too absolute and needs to be modified (see Paul Hacker, "Dharma im Hinduismus," Zeitschrift fur Missionswissenschaft und Religionswissenschaft 49 [1965]: 96-97, 104). The good produced by actions of dharma are "transcendent" in that they depend in an essential way upon the absolutely good energy of such actions, which energy operates in the future. beyond the scope of contemporary observation. But the benefits produced are not necessarily other-worldly.

(32.) Ap 1.9.13 illustrates dharma giving rise to a good that is mundane but of the highest importance. It enjoins that a brahmin who plans to recite the Veda, but who has some wish to speak first with a woman who is menstruating, speak with another brahmin both before and after speaking with that woman. The result promised: evam tasyah prajanihsreyasam. OliveIle's English rendering "In this way the child she bears will prosper" is too definite. The apotropaic measure enjoined here does not guarantee the woman healthy offspring, but only that the woman's future reproductive soundness shall not have been harmed by her having talked, while menstruating, with a man who has the Veda on the tip of his tongue. Nihsreyasa here refers simply to her (uncertain) reproductive prospects being no worse for this encounter: I would render it "Thus will she have the very best she can with regard to her babies [even after such a dangerous encounter]" (lf). By qualifying the benefit with an adverb, Friedrich's rendering heads in the right direction without making the full point explicitly (p. 83): "So bringt das fur ihre Nachkommen am meisten Gluckseligkeit."

(33.) At Ap 2.16.2 we have a result promised to come from the performance of sraddha rites, which could be this-worldly or other-worldly or both. It involves virtually the same compound, prajanihareyas, as Ap 1.9.13, and here specifies that Manu's having instituted among humans the sraddha rites honoring ancestors was "accompanied by the greatest possible good for their progeny" ( jit.). I agree with Buhler that we should take prajanihsreyasa as an archaic instrumental, specifying the consequence of his doing so as an accompaniment. Haradatta, who reads prajanihsreyasaya here, a dative of purpose, acknowledges the difficult reading Buhler and Olivelle elected (prajanihsreyasa ca), but explains it with the somewhat fantastic "Vedic" suggestion that the "letter 'c'" has been substituted for the "letter 'y'" in the otherwise correct dative form. That is, for some reason prajanihsreyasaya ended up being written prajanihsreyasaca. "nihsreyasaceti pathe chandaso yakarasya cakarah."Haradatta has clearly gone too far in this instance. Here again Olivelle renders nihsreyas(a) with "prosperity" (and appears to follow Haradatta's reading): "he did that also for the prosperity of the people," which, again, I think, appears to define the promised good too much.

(34.) At Gaut 11.26 a king judging a legal dispute in which there is conflicting evidence is enjoined to arrive at his final view (nistha) after consulting men steeped in the Vedas. and "thus he will have the best possible outcome" ( jlf: tatha hy asya nihsreyasam bhavati). Again, no particular result is guaranteed, and, as with Ap 2.16.2, the result might be either or both this-worldly and other-worldly. Should the king try to decide such a matter not so aided, something less than the best may well result; he may decide wrongly and thus endanger the future prospects of both himself and his subjects--an implicit possibility not sounded in Olivelle's "He will attain prosperity."The relative and indefinite nature of the nihsreyasa good here and in Ap 1.9.13 makes clear that the behavior enjoined in them is naimittika and that the positive good it produces occurs in place of a likely negative consequence should the measure not be taken.

(35.) I believe Haradatta is wrong here to see sutra 8 as referring to sudras alone; that interpretation is possible, but it seems neither compelling nor likely. The grammar of sutra 6 makes the noun phrase phalavanti ca karmani more likely to be a fourth item in the series given in that sutra than to be a separate sentence. But Haradatta took it as a separate sentence, specifying the rewards that come to members of the three upper varnas from living a life of Veda recitation and fire-offerings. Taking phalavanti ca kartmani in that way introduces and then closes the topic of the rewards available to non-sudras for doing those three earlier dharmas. Then, in this reading, when sutra 7 specifies the general form of action prescribed for sudras alone, namely susrusa, it is logical that there would be some parallel specification of the good they acquire for doing this dharma. It seems more likely to me that the phrase phalavanti ca karmani should be seen as a fourth injunction of dharma for which non-sudras have an adhikara. namely the fruitful rites they are to perform after installing the fires. That phrase is, thus, not a separate sentence, does not close off the topic of the good that non-sudras acquire from these pursuits, and so allows us to read sutra 8 as a specification of a "best possible good" (nihsreyasa) for each vara in descending order from the first. So too Friedrich, p. 53.

(36.) According to Olivelle's text, the four top-level units of Baudhayana are prasnas, and adhyayas are subunits of prasnas, as are the khandas. Curiously, in his notes to Baudhayana, Olivelle designates the four prasnas as adhyayas.

(37.) This puzzling practice was employed occasionally in his earlier edition and translation of the Upanisads.

(38.) Georg Buhler had pointed these out loosely in the introduction to his translation of the Manu sastra (The Laws of Manu [Oxford: Clarendon Press, 1886], xiii, lxvi n. I), but he saw them as headings of the subdivisions of the ancient Manavadharmasutra that he thought was the sutra precursor of the Manavadharmasastra (see below).

(39.) In the introduction to his translation of Manu (pp. lxxxii-xc) Bilhler analyzed two passages of parallel verses found in Manu and the MBh and showed they cannot be explained satisfactorily as a one-way stream of bor-rowing--third-century borrowings by the MBh from the Manu sastra according to OliveIle (MCL, 23). I have myself compared the overlapping stanzas in a third parallel passage pointed out but not analyzed by Buhler (Manu 11.1-40 and MBh 12.159.1-23), and I find Buhler's descriptions and conclusions apt for that passage as well.

(40.) But regrettably the binding of this big book (glued, not sewn) does not stand up to the rigors required to use the book in a sustained way. The text, the notes to the text, the translation, and the notes to the translation are all in separate parts of the book, and one must switch back and forth between them constantly to see all that is relevant to any given stanza or passage. It did not take long for pages and groups of pages to loosen and then separate from the backing.

(41.) A term I prefer over the more precise words paleography and manuscriptology, because bibliography is a well-developed discipline in the Western humanities from which Indologists can learn much of value.

(42.) See MCL, 369 and Robert Lingat, Les sources du droit dans le systeme traditionnel de l'Inde (La Haye: Mouton & Co., 1967), 93-94.

(43.) Julius Jolly, Manava Dharma-Sastra: The Code of Manu. Original Sanskrit Text Critically Edited according to the Standard Sanskrit Commentaries, with Critical Notes (London: Trubner & Co, 1887). Seventeen of Jolly's mss. were in Devanagari script, and he had one each in sarada, Grantha, and Telugu characters (Jolly, pp. vi-xiv). Jolly's edition deliberately relied upon readings found in manuscripts accompanied by commentaries and in the commentaries themselves, failing to appreciate several of the pitfalls of working with commentaries whose own texts and readings of the base text had not been critically verified. See MCL, 376-77.

(44.) Olivelle tried unsuccessfully to obtain copies of other mss., including four Sarada mss. from Shrinagar (MCL, 367).

(45.) Sixteen of Jolly's seventeen north Indian script mss. were Devanagari, the Sarada manuscript of the Deccan College (Olivelle's sPu 6Jolly's "K") being the only exception. In addition Jolly made use of two Grantha manuscripts and a Telugu one. Three of Jolly's manuscripts were among Olivelle's fifty-three: two in Devanagari characters and the other in Sarada.

(46.) Unfortunately the manuscript descriptions are not consistently uniform and complete. Physical descriptions of the manuscripts are sometimes given, other times not. At times the reader is left to infer that we have "no date" because of missing end-pages or illegible colophons, but other times "no date" is explicitly indicated even when there is no colophon. More seriously, while the completeness, or incompleteness, of most mss. is noted explicitly with indications of missing folios or portions, three mss. (Jo', Jo2, and wKt1) have no such indication, an omission that impairs the reliable use of the critical apparatus (see nn. 51 and 52 below).

(47.) This date is no longer available on the manuscript itself, as the final folio is missing; the date is taken from the pertinent catalog at Lahore published in 1932.

(48.) Jolly, pp. vi-xiv.

(49.) Though at times only for the first sloka of such an absence, which can be a problem when a witness is missing several consecutive slokas, as. e.g., with 5Pu6 between 12.5c and 12.12b, the same ms. between 12.23c and 12.28d, and Jm between 12.66 and 12.126.

(50.) Based on the information available in the description of the manuscripts: The first chap. is completely absent in four mss. (Be3, BCa, wKt6, 4Tr5) and the first 63 slokas of chap. 1 are missing in Pu9. The first 69 slokas of chap. 2 are missing in Be3 and the first 178 slokas of chap. 2 are missing in wKt6. All of chaps. 2, 3, 4, and 5 are missing in mT15. Chaps. 5 and 6 are missing in 0Md2. Chap. 7 is missing in two mss. (0Md2, Pun and the last 170 slokas of chap. 7 are missing in Pu9. Chap. 8 is mostly missing in five mss. (wKt6, La2, 0Md2, Pu9, Pun and the last 74 slokas of chap. 8 are missing in Be3; the bulk of the slokas in chaps. 8 and 9 are missing or mixed-up in Ox3. Chap. 9 is missing in Pun chaps. 9, 10, and 11 are missing in six mss. (Be3, wKt6, La2, Lo5, 0Md2. Pu9); chap. 11 is missing in one other (0x3), and the last 105 slokas of chap. 11 are missing in mTr3. Chap. 12 is completely missing in nine mss. (Be3, Kt6, La2, Lo4, Lo5, 0Md2, Ox3, Pu9, mTr3), and the last dozen or two Slokas of chap. 12 are missing in two others (wKt3, Tr1). This summary ignores a few very small discrepancies (e.g., while wKt6 and La2 are basically absent from chap. 8 on, the first includes the first eight slokas of chap. 8 and the second the first five slokas of chap. 8) and the lacunae reported for the virtually complete mss. Bo, Ho, NKt4, and La1.

(51.) See, for example, the endnotes to 1.7d, 13a. 44b, 46a, 61d, 2.40d, 44d, 47d. For such a complicated apparatus this measure is very helpful to users. but it makes the generous editor transparently accountable in a way editors usually are not. For interests of my own I studied the text and the apparatus of chap. 12 closely and was led to compare the positive and negative listings of readings in a number of instances. The first couple of comparisons turned up small problems and these led me to check all such reciprocal listings in chap. 12, which uncovered further small discrepancies with the listings of the witnesses. These discrepancies are not very significant overall, given the number of witnesses in each branch of the tradition and especially given the relatively small number of readings being decided on the intrinsic merits of the variants. The discrepancies I refer to are typically manuscripts not being listed either in the apparatus or in the positive listing of the relevant endnote: e.g., at 12.6c four mss. are unaccounted for in the listings (BCa, La1. NNg and sOxl; La1 was listed among the varia for 6b); at 12.11d the ms. Kt2 is listed among the varia and also in the positive list, and again BCa is not mentioned anywhere (in fact it seems BCa is actually absent in most or all of chap. 12 without any notice of the fact): at 12.16d Olivelle states that the selected reading "is found in all the ST mss.," but actually the ST witness mTr5 is given as a variant reading and mTr6 is not accounted for (as BCa too is not); at 12.22d BBe2, BCa, and Pu3 are unaccounted for. At 12.42b Lo1 is recorded as reading both ca kacchapah and sakacchapah And so on in chap. 12 and elsewhere.

Two other problems that I encountered in the apparatus of chap. 12 are systematic and could have been ameliorated by Olivelle's taking measures to notify all users prominently of, first, any general and, then, any local absences of witnesses. Nine of the fifty-three mss. are absent for all of chap. 12 (Be3, wKt6, La2, Lo4, Lo5, 0Md2. Ox3, Pu9, and mTr3), with Tr' absent from 12.98 onward and wKt3 from 12.118 onward. As I suggested earlier, a "Table Summarizing the Coverage of All Manuscripts" prominently featured in the front-matter of the edition should prevent duly attentive users from looking vainly for absent mss. in the apparatus. Likewise, in three instances (at 12.5, 12.23, and 12.66), merely local [but extended] absences of particular mss. are noted by Olivelle, but only where the absence begins, with no repetitions of the absence at the top of each affected sloka, where sloka-specific absences are normally recorded. I would suggest any absences not advertised in a table such as just suggested be posted for each pertinent sloka; or, if that would be too space-consuming, in a listing at the beginning of each chapter of any extensive manuscript lacunae in that chapter. To fail to take such measures diminishes the utility of the apparatus.

(52.) Olivelle's system refers to manuscripts with a subscript capital letter indicating the script ot a manuscript (e.g., "G" for Grantha, etc., or no letter to indicate Devanagari), a capitalized string of two letters indicating the current location of the manuscript (e.g., "La" for Lahore, "Ng" for Nagpur, etc.), and a superscript number to distinguish among multiple manuscripts from the same location. Thus BCa signifies the one manuscript from Calcutta, which is in the Bengali script; mTr4 represents the fourth manuscript from Trivandrum, which is in the Malayali script.

(53.) Of course a "witness"--a logical construct--is an interpretation of what is found in a manuscript--a material entity.

(54.) See Vishnu S. Sukthankar. "Prolegomena."in The Adiparvan (Poona: Bhandarkar Oriental Research Institute, 1933), 1-CX.

(55.) "After all that has been said above, it is needless to add that the constituted text is based on all versions of both recensions and prepared on eclectic principles," Sukthankar, CII.

(56.) OliveIle refers here to the stemmatic method, which was touted not as "classical" but "revolutionary, modern, scientific" and "objective" when first proposed by Karl Lachmann in the early decades of the nineteenth century. It is now widely regarded as discredited (see p. 58 of Lee Patterson, "The Logic of Textual Criticism and the Way of Genius: The Kane-Donaldson Piers Plowman in Historical Perspective," in Textual Criticism and Literary Interpretation, ed. Jerome J. McGann [Chicago: Univ. of Chicago Press, 1985], 55-91 and 212-19). While 1 agree that the stemmatic method suffers many of the shortcomings its accusers, most notably Joseph Bedier, have directed against it (see David HuIt, "Reading It Right," in The New Medievalism, by Marina S. Brownlee, Kevin Brownlee, and Stephen G. Nichols [Baltimore: Johns Hopkins Univ. Press, 19911. 117-19, and Patterson, "The Logic of Textual Criticism," 57-60; a more radical set of reflections upon the fundamental ideas of text, author, writing, philology, etc., is found in Bernard Cerquiglini, In Praise of the Variant: A Critical History of Philology, tr. Betsy Wing [Baltimore: Johns Hopkins Univ, Press, 1999]), I believe that the stemmatic method is a valuable regulative ideal that should play at least a heuristic role in the interrogation of' a set of witnesses, preferably with the aid of cladistics software. Such software can afford great help in analyzing collation data even with highly contaminated witnesses. See Philipp A. Maas, "Computer Aided Stemmatics--The Case of Fifty-Two Text Versions of Carakasamhita Vimanasthana 8.67-157," Wiener Zeitschrifi fur die Kunde Sud- und Ostasiens und Archiv fur indische Philosophic 52-53 (2009-10): 63-119.

(57.) A southern version of the text had interested Jolly. Buhler, and others but was impossible to establish earlier with the resources then available. It is important to note that Olivelle's ST is not limited to manuscripts written in southern scripts (the Devanagari T[r.sup.1] is such an exception), nor does it occur in all southern-script manuscripts (the Malayali [.sub.M]T[r.sup.3] is such an exception).

(58.) Olivelle calls the NT-y version of Manu the "vulgate" "because this is the version represented by most printed editions of the MDh and the one accompanying the commentary of Kulluka in most mss" (MCL, 372).

(59.) See Hult, "Reading It Right."120-21.

(60.) My own close experience with the Santiparvan as edited by S. K. Belvalkar makes me willing to defend the principle as of great general value in practice.

(61.) The stanza number 5.62 has consequently been skipped in Olivelle's edition to keep the subsequent stanza numbers aligned with earlier editions and secondary sources. He does the same with the other stanzas dropped.

(62.) The word in question specifies the recipient enjoined for the gift of a cow; agu. Literally 'someone without a cow', is a word that appears to be completely unknown to both Sanskrit epic traditions.

(63.) Among many possible examples: atindriyo 'grahyah for atindriyagrahyab at 1.7a; samyag vardhante for tasya vardhanta at 2.121c; amaithuni (with "virgin" the most immediately meaningful sense) for maithune at 3.5; niyata replaces nihita at 12.10c; niyacchati replaces nigacchati at 12.1 Id; and sarisrpah replaces sakacchapah at 12.42b.

(64.) The stanza 7.67 is concerned with the king's appointment of an envoy to deal with other kings. Olivelle's text and Jolly 's read here sa vidyad asya krtyesu nigudhengitacestitaih, I akaram ingitm cestem bhrtyesu ca cikirsitam, while all ST manuscripts and NKt4 (plus NT-x's Be'. NPU', and Pu3) read krtyesu again in d, in place of bhrtyesu. OliveIle gives three good reasons for preferring bhrtyesu in d: "broad manuscript support"; "krtyesu here may have been influenced by the same word in pada-a"; and "the ca in this pada calls for something different from krtyesu."But the two lines do not fit well together, even when reading bhrtyesu in d. The repetition of the same ideas in c already mentioned in b suggests that two originally separated half-slokas have been fitted together. There is no question that the sloka can be construed grammatically, as Olivelle's translation shows (alertly rendering the specialized sense of krtya nicely), and clearly the reading bhrtyesu in d helps make that possible: "By means of concealed hints and gestures, he should decode the bearing, hints, and gestures of the rival king with the help of seducible men in his service and uncover his plans with the help of his servants" (olv., Manu 7.67). I would prefer to do it differently: "He [the envoy] should learn of his [the enemy king's] bearing, his tell-tale signs, his activities, and his intentions by way of secret signs and actions (directed) toward (those) servants of (that king) who can be set to act against him."I would suggest that 7.67cd originally read krtyesu in d, had the sense of the enemy king's "projects, or affairs," and was dependent upon cikirsitam "his intentions for his undertakings."I think the krtyesu in pada a did, from the start, have the sense Olivelle sees in it, and I think that the person who fused these two half slokas was attracted by the pleasure of using the same word in two different senses, one obscure or technical and the other common. The eventual change of the second krtyesu to bhrtyeyu by some scribe or editor effected greater unity in the still awkward pair and eliminated the recherche pleasure of the two different krtyesu-s.

(65.) For a mild example see Buhler's comments on Medhatithi's "chief weakness" (introduction to his translation of Manu, pp. cxix-xx).

(66.) See Olivelle's apt comments on the sutras' divergent voices at DS, 17-18.

(67.) Bharuci (6th c. [dating of his editor Derrett; see MCL, 367]) is the earliest known commentator. See MCL, 51-52. Olivelle cites here a model of the tradition sketched originally by Richard Lariviere in his introduction to his edition and translation of the Naradasmrti (The Naradasmrti: Critical Edition and Translation [Delhi: Motilal Banarsidass, 2003],4--referred to accidentally with an outdated reference to its first edition, "Lariviere's (1989, xii).

(68.) OliveIle methodically accounts for his decisions regarding these twenty-six passages in a chapter-by-chapter survey of the text (MCL. 52-62). See Miller's introduction to his Manu translation, pp. lxvi-lxxiii for a parallel survey of the contents of Manu that discusses some of the same passages, though for Buhler these potentially later additions are being judged against his inferences about the contents of the hypothetical. metrically refashioned Manavadharmasutra.

(69.) If the putative autograph were 2306 slokas, 374 amounts to more than a 16% expansion of the original. Olivelle gives the total as 329 slokas. 12% of the current size (MCL, 61). Interestingly, eight of these excursuses occur in the first chapter of the Manu, making up 85 of its 119 slokas. Two more (with 17 slokas out of 249) occur in Chapter 2 and the other sixteen occur in chapters 8-9 and 11-12.

(70.) Very interestingly, an incidental survey of eighty-five instances of extra slokas found in chapters 2, 3, 5, and 10 (slokas thus relegated to Olivelle's apparatus as "additional" material) turned up only two instances in which such additions were found in NT-y mss. (virtually all mss., including those of NT-y, have the extra 5.61ab and 5.*62ab [see above] and in addition after 3.259 the NT-y witness 0Md2 contains an additional verse found in ten other mss.). Clearly various NT-x and ST mss. were much more energetic in "completing" the text than was the NT-y branch after some particular point in time.

(71.) I would suggest that the historical situation of the Manu sastra was likely very different from that of the MB/i, because there seems to have been a relatively coherent scholarly tradition of dharmasastra using and maintaining the text of Manu for a very large part of its existence, perhaps into modern times. The base of the MB/i, on the other hand, arose. I believe, as largely "formulaic-improvisational" narrative poetry performed before live audiences (see pp. 104-5 of my "No Contest between Memory and Invention: The Invention of the pandava Heroes of the Mahabharata," in Epic and History, ed. Kurt Raaflaub and David Konstan [Oxford: Blackwells, 2010], 103-21). There seems not to have been a similarly tight-knit institution surrounding even the written Sanskrit text of the MBh once it was created, which, as a written Sanskrit textual phenomenon, existed in the midst of various, predominantly vernacular, performance traditions (see p. 72 of my "Mahabharata," in Brill's Encyclopedia of Hinduism [Leiden: E. J. Brill, 2010], 72-94). Though manuscripts of the written Sanskrit MBh were similarly "imperfect" or "incomplete" copies of Vyasa's text and were frequently "improved" with corrections and extra passages when copied, the lack of a tightly coherent scholarly tradition sanctioning and circulating these changes means that the pace of textual homogenization was much slower and likely occurred completely only infrequently or never at all--that is. it seems unlikely that "improvements" in the text of the MBh ever displaced completely the original readings everywhere in the tradition. Hence some manuscript copies of the text were considerably shorter than others, a fact that became an important criterion for a relative ranking of MBh manuscripts in Sukthankar's editorial procedure.

(72.) I took a quick look at the status of the two excursuses in chapter 2 (2.2-5 and 88-100) and found only three half-slokas missing from the first of them in one ms., T[r.sup.2]. I expect that a check of them all will turn up only occasional and insignificant instances of some portion or all of an "excursus" not being present in all witnesses.

(73. Of course, purely logically, one witness's addition is another's omission and vice versa.

(74.) Significant typographical errors (the first number refers to the page). In the sutra volume: 191, last line, change "522" to "569"; 472.1. 22, change "(ango va" to "ario va"; 523, 1. 10 from bott., change "A. 1.23.3-" to "A. 2.23.3-"; 570, I. 13, change "the Hultzsch" to "Hultzsch" (in a number of places "Hultzsch's" is misspelled with "Hutzsch's"); 605, 1. 6, two times, change "1.11.29" to "2.11.29" and "1.11.1-25" to "2.11.1-25"); 735, in the bibliography, the paper "On the Term antahsamjna-" has been accidently assigned to Michael Witzel's authorship, rather than to the correct author, Albrecht Wezler, who is listed immediately above Witzel. In the Manu volume: 9, 1. 16, change "9325" to "9.325"; 24, n. 46, 1. 3, change "vyavahapada" to "vyavaharapada"; at 354. II, I, the lead paragraph of ms. description omits 0 = Oriya; 357, re ms. Kt6, line 5, change "[.sub.w]K[t.sup.5]" to "[.sub.B]K[t.sup.5]"; 359, I. 20, change "at 1224" to "at 12.24"; 369, I. 7 from bott., change "Calcutta edition" to "Calcutta editions"; 370, 1. 10, change 1497 C.E. to 1487 C.E.; 372, 1. 9, change "7.67" to "7.67d"; 372, 1. 10, change "3.114b" to "3.114c"; 578, apparatus to #102, 1. 2, there is no ms. [.sub.G]M[d.sup.4]; is [.sub.G]M[d.sup.3] meant? (esp. since 1Md4 is listed in 1.3); 631, 1. 5, "to the Fiihrer's mss." should read "to Fiihrer's mss."; 883, apparatus following #247, the point regarding the omission of the number 248 from the published critical edition is made twice in the discussion of this additional verse; 939, n. to 5.64a, 1. 3, change "rajani for day" to "rajani for night"; 965, 6 from boa.. change "NT one" to "NT ones"; 981, 10 from bott., 1 infer "prayer must be a typo for "preta"; 1023, 1. 7 (rt. col.). change "MBh 12.188.3" to "MBh 12.88.3."to"MBh 12.88.3."

JAMES L. FITZGERALD BROWN UNIVERSITY