The recent resurgence in Vedic studies has been driven, in large part, by innovations in two key areas of textual analysis: a systematized relative chronology of the texts, and an understanding of the stylistic elements at work within them, ranging from meter to lexical choice, morphology, and accent. Both frontiers bear closely on each other and, in turn, form the foundation upon which the current study of the self is constructed. This recent work brings into focus the discussions which have been ongoing for decades and simply overlooked by scholars for whom Vedic literature was sufficiently explained by Max Müller over 100 years ago. It is not easy to say why this has remained the accepted understanding among most scholars.52 It is possible, however, that the recent availability of additional manuscripts of Black Yajur Veda texts and other previously unknown Suutras has added additional correlation to the research done in the late 19th century by Arnold (1897), Avery (1880), Lanman (1880), and Oldenberg (1888; cf. Witzel, 1995a: 85; 1998: 289).

    This wealth of data about the Vedic materials which has gone largely unnoticed. Like a lost heritage, it has to be reclaimed--not just by Vedic scholars, but by Indologists as a whole. Without careful reconstruction and awareness of the actual developing ideas and texts of ancient South Asia, even the most conscientious scholarship concerning later eras has a weak foundation which is susceptible to the winds and whims of the current political climate. The material for a more precise reading of the early development of Vedic history is plentiful and long-lived, going back as far as 120 years. This "wealth of data" is scattered throughout numerous articles and introductions to translations or editions of primary texts. There is also the statistical work of E. Vernon Arnold (1897, 1905), the grammatical analysis of Mylius (1970), Hoffman (1975), and J. Narten (1968).

    It is essential for the success of this study to construct the most pre

cise available framework for the diachronic analysis of the terminology related to the self. An additional objective in the detailed study of the RV which follows immediately below is a precise analysis of the internal arrangement of each MaNDala such that the changes in vocabulary for the self can be correlated with the respective families and--where the data allows--geographical regions of Vedic India. In many cases this study will offer only a beginning for understanding geographic and ethnographic shifts as the field of such inquiry is quite new and available data is limited. However, it will be impossible to analyze with any precision whether there were competing notions of the self in Vedic India without the results of this examination as a point of comparison from one family of poets to the next. The texts chosen for this study represent the periods of Early and Middle Vedic: the Rg Veda representing the former "Early Vedic" period, and the Rg Veda Khilas, the KRSNa and Shukla Yajur Vedas, and the Atharva Veda as well as related BraahmaNa's representing the later, or Middle Vedic period.

Relative Chronology of the Texts

    An excellent summary of the early explorations of the relative date of the Vedic materials is found in the recent publication of Lars Martin Fosse's dissertation "The Crux of Chronology in Sanskrit Literature: Statistics and Indology, a Study of Method (1997). Inattention to this stratification engenders the problem in which scholars frequently persist in generalizing between decidedly earlier portions of the Rg Veda and later portions (Witzel, 1995a: 96).53 Perhaps the attention of modern scholars has been unintentionally directed elsewhere as a result of the popular movements in methodology. For instance, as recently as 1968, J. Narton's study of verb usage in early Vedic presented layers of text stratification. "Dei Marke RV hat also a priori mehr Aussicht auf sprachhistorische Relevanze als etwa TB, die Marke Vedisch54 mehr als Episch, Episch wiederum mehr als Klassiches Sanskrit, da dies ja weitgehend nur noch Lernsprache war" (1968: 115-6).

    More recently, Mylius worked with a general partition of stylistic periods in Vedic literature which were more precise than Max Müller's generic sequence of SaMhitaas, BraahmaNas, AAraNyaka's and UpaniSads, initiating a categorization of the literature after the SaMhitaas as "Middle Vedic" (1970: 423). Mylius strove to correct the theories of date based either upon the evolution of the Brahmii script or astronomical data (1970: 427). Specific problems have been perpetuated by the premature comparisons across

cultures--e.g., with the Avesta--when internal dates of the Vedic corpus remain unclear: "Bevor diese und andere Ansichten diskutiert und witere Anhaltspunkte gesucht werden, empfiehlt es sich, einige allgemeine Bermerkungen zur Methodik der Vedadatierung zu machen" (1970: 422). He draws instead upon a philological analysis of the periods of redaction for several genres of literature--Buddhist, BraahmaNas, GRhyasuutras, Shrautasuutras, etc.--to build a matrix of composition periods and cross references for determining relative sequence. He concludes, nonetheless, that for the earlier Vedic literature, "Die Schicten der vedischen Literatur sind chronologisch nicht scharf gegeneinander abgegrenzt" (1970: 428). The Rg Veda's chronological stratification, while clearly redacted over several centuries, occurred "so muß sich die endgültige Zusammenstellung . . . wegen der starken geographischen, sozialen und religiösen Veränderungen über mehrere Jahrhunderte hingezogen haben" (1970: 429), the specific periods of which remained unspecified at the time of his writing.

    The available research in Vedic chronology was not limited to publications in German. There is the translated work of Weber (1892) or the original work of Keith (1914, 1920) which regularly described the various layers of stratification in the Yajur Veda texts. Winternitz also noted multiple voices in the redaction of the Rg Veda (1959, I: 57), a developing tradition of innovation in recitation as recorded in the Yajur Veda tradition (1959, I: 163), and the potential relationship between text history and social history (1959, I: 172).

    It is ironic that the temporal and chronological errors referred to above--e.g., Gonda's generalization about Vedic social structures, et al--also postdate studies which present more detailed stratification schemes, as do Narten and Mylius. Specifically, Gonda himself characterizes Vedic literature as an evolving, developing body of ideas, expressions, and commentary (1975: 24-25), yet he nonetheless obviates his conclusion when applying semantic field analysis to the problem of a term or concept used in several different texts. Still, the inattention to the reality of internal chronology in Vedic texts persisted. Some twenty years after Gonda's articulation of semantic field study mentioned in Chapter One, "Some notes on the study of Indian religious terminology" (1962), he published "Translating the Veda (with regard to AiAAr. 2.6.2)" In which he quotes from RV, AV, ShB and various UpaniSads with complete inattention to historical periods (1980-81: 15f.) in his efforts to clarify a semantic field of terms which includes bráhman, dhii, vaaja, and maayaa.55 The predilection to reductionism

in the quest for of central meanings rather than attention to historical development is also clearly evidenced in his otherwise comprehensive history of Vedic literature, which remains all but silent on chronology amid exhaustive detail in analysis.56

    Thus, the analysis of the developing terminology for the self, as a philological exercise, requires careful attention both to the limits of philology itself and to the stratification of the texts. The consensus of a conference on the subject of philology, held at Harvard in 1988, was that it entailed " . . . a Kulturwißenschaft based on texts" to which Witzel adds, ". . . the study of a civilization based on oral and written texts, in contradistinction to such subjects as linguistics, history, archaeology and sociology which also make use of other categories of evidence" (Witzel 1995a: 115). Thus, the summary of the best discernible chronology, presented below, will facilitate proper historical analysis of the developing terminology for the self while avoiding reductionism, oversimplification, lack of attention to genre, or chronological blurring of distinct periods. The variety of sources scattered in different volumes and articles, as well as those in the introductions to editions of the major texts have been gathered below in order to establish a working chronology of Early and Middle Vedic. This yields a picture, if not of the actual development of the conception of the self, then of the changes in how that conception was expressed. The actual conception may very well remain outside the realm of academic scholarship to determine, considering the resources which have survived time, cultural change, and ideologies of orthodoxy.

    The following chronology is divided into the major categories of Early and Middle Vedic. Within these I propose to offer a detailed chronology of text development for those sub-periods outlined and comprehensively defined by Witzel (1989: 124ff, 1995a: 96-97), based on the pilot study by Narten (1968: 115-6), as part of a five-part stratification included under Early, Middle, and Late Vedic: 1. Rgveda, 2. Mantra language (AV and the metric portions of the YVS), 3. SaMhitaa prose (those portions of expository writing, not in metric or accented form, of the YVS), 4. BraahmaNa prose (older and later), and 5. Suutra Language. For the current study, however, it will be the sections of Rgvedic, Mantra language, SaMhitaa prose, and BraahmaNa prose which will be of primary importance.

Early Vedic: The Rg Veda


    Early Vedic is comprised entirely of the text of the Rg Veda. date, the most thorough study of its contents, developments, dialectical variation

and peculiarity as to families of R'Si's is found in the work of Witzel (1995b: 308ff). As mentioned in Chapter One, the oldest portions of the RV are found in the so-called "family books," MaNDala's 2-7. These are clearly distinguished by the homogeneity of the family line to which each R'Si belongs, the order of the hymns according to the deity addressed and the meter employed. The six books are ordered according to the ascending number of hymns per "book" or MaNDala (Oldenberg, 1888; Witzel, 1995b).57

    Within each book the first principle of order is the deity addressed. Agni invariably precedes Indra, who is followed in turn by any one of several deities depending on each respective family collection (Gonda, 1975: 5f.; Weber, 1892: 29f.; Witzel, 1995b: 309; etc.). For instance, following Indra the GRtsamada's of RV 2 include BrahmaNaspati and BRhaspati; for the Vishvaamitra's of RV 3 it is the Vishve Devaas; the Vaamadeva's of RV 4 laud the Rbhus third; etc. Each deity's "collection" is, in turn, ordered in descending sequence based on the number of stanzas in each hymn with the longest coming first, the next longest second, and so on. Where the number of stanzas remains the same, the hymn with a meter containing a greater number of syllables--e.g., the jagatii with four 12-syllable paadas--takes precedence over one with fewer syllables--triSTubh with four 11-syllable paadas--in terms of relative order (1995b: 309).

    As early as 1897, E. Vernon Arnold suggested: "For the successful study of the Rigveda no preliminary is more urgently needed than a true grouping of its parts." The order of deities addressed depends upon how many hymns are devoted to each divinity or group of divinities. The deity having the most hymns comes first--almost without exception. This quantity varies with different deities from MaNDala to MaNDala. This raises an interesting point with respect to certain predilections from one clan to the next for different deities. Evidently some clans "favored"--or at least the redactors did--one deity over another and this often to the exclusion of another deity.

    As the Early Vedic material is comprised solely of the RV, and the analysis of its contents is the sole concern of Chapters 4 and 5, I will attend to Arnold's suggestion in detail and provide a careful outline of the arrangement of the hymns in each chronological portion. In this way the patterns of emphasis on one deity or another by one family can be compared with that of another. As this arrangement of hymns from one MaNnDala to the next is explained, the characteristics of the key words

under examination in this study are tested according to how much each word is used by each family and its use as it changes with different deities. I have examined the Family Books in greater detail with respect to the terminology under examination because, as we saw with previous studies in Chapter 1, if any examination of the terminology for the self has been done in the RV, it has almost invariably emphasized the later books. The present study is providing a detailed record of this largely unstudied collection of the earliest known Vedic material.

    There are exceptions to the general scheme of arrangement of hymns in the Family Books as it is outlined above. Variation in the pattern is also a key to identifying later insertions (Oldenberg, 1888), and there remains a need to determine a pattern of arrangement for the 8th MaNDala (Bergaigne, 1878-83, II: 76ff.), also noted in Witzel, 1995b: 311). In addition, Lanman foresaw many of Oldenberg's findings in his own monumental study, published in 1880, which initiated the list of late RV insertions examined further in Arnold (1897), and Oldenberg (1888). Lanman also suggests the lateness of 1.162 and 1.164 (1880: 581). It was Lanman who influenced Arnold to undertake his own study of statistics with Vedic meter in 1897 and 1905.


    Of particular interest for this study is Arnold's 1897 "Historical Vedic Grammar" in which he summarizes the work of Graßmann (1876), Lanman (1880), Zimmer (1879), Brunnhofer (1881), and Oldenberg (1888) as to the different criteria by which each hymn of the RV could be placed in its relative chronological order of composition. He defines five basic criteria for determining the relative date of composition which, in descending order of reliability, are 1) lateness of vocabulary and grammatical forms (this would include, for instance, with the infinitive in toH identified many decades later by Witzel [1989]); 2) later AnuSTubh verse; 3) position in the collection (e.g., does it violate the order of arrangement according to number of hymns to a given deity, number of verses, etc.); 4) the fourth and fifth categories are equally unreliable: subject matter and a mixture of TishTubh-Jagatii verse/TriSTubh with extra syllable (1897: 211).

    In this dissertation, I have prioritized the lateness of vocabulary and grammatical forms, followed by asynchronous position in the order of arrangement as the two most reliable categories for determining lateness as there is a greater body of research in recent years on these criteria. This more recent work relies upon the semantic and linguistic data more than upon meter (Narten, Witzel, Mylius). In addition, the work on metrics is

based in large part upon statitistical studies which have had a mixed record of success and acceptance over the last century. One additional caveat to my use of Arnold is that his distinction between two periods of later composition--"C1 and C2"--are too difficult to place in historical time and, in the absence of additional manuscripts with which to compare styles and vocabulary, such detailed resolution of chronology thus remains too uncertain to serve as a control for diachrony in this study.

    Fosse notes that Whitney and others began to recognize the data to be unearthed in statistical study in their encounters with the work of John Avery, beginning in 1872 (Fosse, 1997: 18f.). Acceptance of statistical analysis of Vedic literature remained in its infancy, however, and the early efforts of scholars like Arnold or Wüst (1927) although impressive considering the absence of computer technology, came to be viewed as flawed in their results (Van Nooten and Holland, 1994: ii; Keith [1906]; Bloomfield [1901]). Fosse adds an additional nail to Gonda's coffin by noting that the work of Arnold, Wüst, and others were not only rejected in Vedic Literature (Gonda, 1975: 27), but they were not even fully consulted (Fosse, 1997: 28).

    In spite of presenting the current state of the art for statistical technique in the analysis of Sanskrit texts, however, Fosse himself concludes that, with respect to philology, statistical studies remain largely problematic in their application.58 The fundamental factor which affects the application of statistics to a text like the RV is that one must begin with categories in order to compare the statistical analysis within them (Fosse, 1997: 2). One begs the question before asking it. To begin with MaNDala's is far too broad and, as the analysis below reveals, the MaNDala's have very limited chronological integrity. For this obvious reason, Fosse does not pursue his statistical experimentation in the RV (1997: 3). The present study takes due consideration of Fosse's conclusion and strives to articulate additional structural (e.g., more precise analysis of the internal chronology for the hymns of the Family Books--RV 3 seems later) and terminological categories (the changing vocabulary for the self) such that statistical analysis of the RV might overcome these shortcomings.

    Witzel (1989: 127) observes that with the advent of electronic texts and systematic statistical methodologies--something now provided in Fosse's model--additional delineations of date, social strata, and geographical region can be determined. What is first necessary is a chronological framework within which to begin such a project. I am suggesting that the

outlines presented by Witzel, Narten, Keith and others--when correlated as they are below and augmented with the philological data adduced in the chapters to follow--will offer the level of detailed category formation necessary for systematic statistical analysis of Early and Middle Vedic. Thus it is important to note that I am not relying on statistical data as a basis for the conclusions in the following pages but rather as a resource for structuring the results of philological inquiry. In the following discussion, I have drawn mostly upon information adduced from the edition of the RV recently prepared by Van Nooten and Holland (1994) in order to present the structure of each Family Book in detail. Since the words aatmán and púruSa are quite scarce in the Family Books, and the meaning of tanuú and tmán is one of the significant contributions this dissertation makes to the understanding of the Vedic self, careful attention will be paid to the changing use of these words throughout the summary of arrangement in the Family Books.

    In the following section, I am including an overview of the vocabulary as it is correlated with the hymns assigned to each deity which yields, in turn, several patterns of use for tanuú and tmán which vary somewhat from family to family.59 Often this variation corresponds to deviations from the standard structure of arrangement for the hymns by a given family or to the emphasis by one family upon a deity which is less important to other families. Both tanuú and tmán are used predominantly with Agni or Indra (2/3's to 3/4's of their occasions in each Family Book). The other occasions of each word are applied to deities such as VaruNa, the Ashvins, SavitR, USas, and other gods whose realm is closer to the upper atmosphere.

    Bráhman is most prominently an "Indra word" inasmuch as it is by means of these empowering utterances that Indra is strengthened, called upon for assistance in battle, and/or lauded for his primordial victory over VRtra. AAtmán and púruSa are much less common in the Family Books and so it is much harder to identify patterns with both terms. However, the two occasions of aatmán (7.87.2, 7.101.6) occur in hymns which differ from the established patterns of the other books. RV 7 has the single largest group of hymns to VaruNa alone (without its usual other half, Mitra) and 7.87 is one of these. In addition, the only three hymns in the Family Books to Parjanya includes 7.101 (the other two are 7.102 and 5.83). In addition, the only occasions of feminine puruSiíNam is also found in a Parjanya hymn (7.102).

    Thus variations in the arrangement of the hymns, emphasis on one

god over another, or unique occasions of praise to a divinity or group of divinities all prove to be ripe fields for changes and unique uses of the key words under examination. Tanuú is most frequently used with Agni and Indra of all the deities in every Family Book with the clear exception of the 7th MaNDala. Chapter 4 explains that, in the almost complete absence of aatmán and púruSa in the Family Books, tanuú and tmán serve to represent the individual--whether divine or human. Tanuú is used in reference to the divine realm in over three quarters of the occasions where it is used in the Family Books.

    Throughout the RV, both the Family Books and later portions, tanuú never denotes a frailty in the divine realm. However, since there is not a great deal of discussion of a physical body in the RV--i.e., with words like sháriira, or "flesh words" like kraví--I have taken the position that tanuú denotes the individual so far as s/he was conceived--divine or human--as a presence which entailed both physical and abstract traits. The overview of the Family Books and later RV below confirms that this is workable. As indicated in Chapter 2, tmán does not refer to a finite individual self or essence, but rather serves to underscore a particular characteristic as intrinsic to the identity of a deity. In the Family Books there is little or no variation in the uses of tmán from one MaNDala to the next (MaNDala 6 is a slight exception as there are only two occasions of tmán) and it is a predominantly "Agni" word with 12 of its 28 occasions occurring with this deity while no other single god has more than 5 (the Maruts).

    The hypotheses regarding the meaning of each term as it was introduced in the previous chapter are borne out in the analysis of the arrangement of hymns in the Family Books. As the later portions of the RV do not have significantly large groups of hymns by one poet or family, the detailed examination of terms has not been applied. In the present chapter and the discussion immediately below the intention is to identify the patterns and variations in the use of the terminology under study in each family's MaNDala and with each deity. There has been little or no study of the terminology in the earliest books of the RV, so I have included a summary of how each family treats various words related to the self. It is repeatedly the case that a variation in the arrangement of a Family Book will coincide with a unique occurrence of one or several of the key words in this study. The priority in this chapter is to establish relative sequence more than to fix dates. However, where possible, I will note what information has been determined with some degree of certainty as to the date of the different periods of literature.

Composition and Arrangement of the Family Books


    MaNDala 2 contains 10 Agni hymns, followed by 12 to Indra, then four to BrahmaNaspati (the only one of the Family Books with hymns specifically to BrahmaNaspati). The remaining 17 hymns are addressed to a variety of deities including AAdityas, VaruNa, Vishve Devaas, Rudra, Maruts, miscellaneous gods (e.g., 2 to the seasons, 2. 36-37), and the Ashvins (Van Nooten-Holland, 1994: 114-137). The various hymns to different deities are fairly evenly distributed according to the ratio of the other Family Books.

    Tanuú indicates the aspect of the deity which is especially desired for assistance or singled out for praise--e.g., Indra's strength ( 2.16.2, 2.17.2). It is worth noting that tanuú frequently refers to Indra showing his strength, and often this will coincide with an occasion where bráhman is used as in 2.16.7 and 2.17.3. The empowerment coming from bráhman--or second-most in frequency, from Soma--is what strengthens Indra. Tmán is never used to indicate that Indra's strength is intrinsically part of the god's nature.

    When tanuú refers to humans, it invariably denotes something subject to frailty and vulnerability (2.21.6; 2.23.8; 2.34.4, 6). Agni hymns use tanuú to indicate his bright presence ( 2.1.9) or with a presence/tanuú of fury ( 2.10.5). The remaining occasions of tanuú are ascribed to more abstract deities of the skies and heavens such as the Ashvins in 2.39. RV 2.39 includes both uses of tanuú as a reference to the embellished presence/tanuú of the Ashvins in 2.39.2, and frail human tanuú which is vulnerable to injury and in need of assistance (2.39.4, 2.39.5, 2.39.6).

    Where tanuú is used to emphasize one of several possible traits--for praise or to enlist it for assistance--tmán identifies an autochthonous quality of a divinity. Like tanuú it is also most common with Agni and is rare with Indra. In addition, those occasions of tmán which are not used for Agni refer, instead, to abstract deities--usually those associated with the heavens (as in 2.32.4 with Raakaa the goddess of the full moon's day), but also with gods who preside over less tangible things such as BrahmaNaspati, the father of prayer, who is wise of his own accord (2.25.2). When the occasions of tmán are used as the criteria, Agni is the most clearly defined deity as to his inherent traits. In RV 2.1.6 Agni takes on various identities of other deities such as Rudra, TvaSTar, and so forth. When Agni is PuuSan, an all-seeing, weapon-wielding (among other traits) deity, tmán is used to

underscore that Agni himself is a protector.

    The emphasis on BrahmaNaspati in RV 2 is unique among the Family Books. In this collection of BrahmaNaspati hymns, we have two of only four hymns in the RV concerned with both BRhaspati and BrahmaNaspati together. The distribution of the lexical elements under study show BRhaspati as the more physically present, active protector of what is great (N 10.11- bRhaspatir bRhataH paataa vaa); in 2.23 we find BRhaspati associated with key terms such as tanuú (2.23.8a), -dhii (2.23.10a), kárma (2.23.12b), and krátu (2.23.15a). On the other hand, BrahmaNaspati is associated more directly with prayer (N 10.12- brahmaNaspatir brahmaNaH paataa vaa) and is accordingly found in verses with mánas (2.23.12a). These associations are consistent with 5.46 wherein BrahmaNaspati is asked for help (5.46.3b), while BRhaspati is called upon to guard and shelter (5.46.5b); and with 7.97 where BRhaspati exalts (7.97.2a), brings blessings (7.97.4a), is foeless (7.97.5b), and full of strength (7.097.6a) where BrahmaNaspati is king of prayer (7.97.3b), and called upon to favor praises and awaken thought (7.97.9b). Thus when the Family Books are examined in detail, resolution of issues pertaining to the words related to the self are revealed which, as in this case, revise traditional opinions60 that the two terms--BRhaspati and BrahmaNaspati--were simply synonymous. In fact, it appears that BrahmaNaspati is the older term which is replaced by BRhaspati (see below, Chapter 4).

    In MaNDala 3 we have 29 hymns to Agni, 24 to Indra, then four to the Vishve Devaas, and one each to Mitra (3.59), Rbhus (3.60), USas (3.61), and the final hymn to a variety of deities (1994: 138-169).

The Vishvaamitra's of RV 3 use several words quite uniquely in comparison with the other Family Books. For instance, while masculine brahmán/priest is decidedly less common than neuter bráhman/empowerment or power of prayer (17 of the masculine compared with 165 of the neuter) throughout the Family Books, the Vishvaamitra's do not refer to this type of priest at all.61 Tmán occurs only twice, in 3.3.10b where Agni is all encompassing, and 3.9.5a where he wanders as of his own nature. Both attributes are distinct characteristics of fire, and of the journey Agni takes on his way to the gods as he bears the offerings. This suggests that RV 3 is somewhat later than the other Family Books, a possibility considered below and in Chapter 4.

    The occasions of tanuú are completely confined to Agni and Indra hymns with a fairly even distribution between both deities. In addition, tanuú has the most corporeal of its significations when it is used with the

Agni hymns in RV 3. When considered in comparison to the other Family Books in which tanuú is used with more abstract (BrahmaNaspati, VaruNa) deities and those associated with the atmosphere (SavitR), the confinement of tanuú to Indra and Agni accentuates the more corporeal sense of tanuú in this MaNDala. This distinction is quite pronounced in the hymns of the rival VasiSTha's in RV 7 where tanuú is much more abstract and is largely absent in the Agni and Indra hymns. While the AAprii hymn suggests Night and Day have different colored tanuú's (3.4.6), Agni's tanuú is born in 3.15.2, and decorated in 3.18.4. The tanuú of Agni is referred to in triplicate along with his three tongues in RV 3.29.2. Tanuú is somewhat less corporeal in the Indra hymns, but the line between a simple meaning of "body" and a more abstract presence as suggested in Chapter 2 is still quite thin. His tanuú is besought to become great (3.34.1), delight in the Soma (3.41.6), gain vigor from the Soma (3.48.4), sate his tanuú's hunger (3.50.1), and attend itself to the Soma (3.51.11).

    When a hymn violates the typical arrangement of the Family Books, a change in its ideas and vocabulary almost invariably coincides. In 3.53.8, tanuú is quite abstract compared to the rest of the Vishvaamitra hymns. It is also different from the balance of the RV where tanuú is used with Indra. It is an odd occasion where the tanuú of Indra is magically or illusorily changeable (3.53.8b: maayaáH kRNvaanás tanvám pári svaám). Tanuú is used elsewhere in the other Family Books to refer to the multiple aspects of a deity's tanuú ( 2.17.7, 3.4.6, 3.20.2, 5.67.5), but never to Indra's ( click here to follow a chronological series of all occasions in which tanuú is found in the RV, in a separate browser window).62 In addition, 3.53.18 contains the only case in the 3rd MaNDala where we find the familiar statement as to the frailty of the human tanuú (bálaM dhehi tanuúSu no bálam indraanaLútsu naH). This is not altogether surprising, of course, as RV 3.53 is a late hymn (Lanman, 1880: 518; Witzel, 1995b: 311). As a later addition, it could possibly have been added to make the praises of the deposed Vishvaamitras (by the VasiSThas of RV 7, see discussion below) more homogenous with the other books.

There is additional evidence that the hymn is a later insertion such as the deification of Vaac in 3.53.15-16 which is an idea from the later RV.63 In addition, Bloomfield has observed that 3.53.7, regarding benefits of proper gifts to the Vishvaamitra's by the sacrificers is also said of the frogs in the late addition to the 7th MaNDala, 103.10 (1916: 204). Part of the praise to Vaac, RV 3.53.16 is echoed in 7.80.2 (Bloomfield, 1916: 205).

    In the 4th MaNDala, we find the Rbhus receiving five hymns (4.33-37) following 15 to Agni and 17 to Indra, then a miscellany of gods including 3

to Dadhikraa (there are only 4 hymns in the entire RV devoted to this battle horse and its exploits), then the Ashvins (4.43-45), Vaayu (4.46-48), BRhaspatitogether with Indra(4.49-50), various heavenly deities (e.g., USas, SavitR), natural phenomena (KSetrasya Pati, Lord of the Soil) in which is included one hymn (4.55) to the Vishve Devaas (1994: 170-201). The concern of the Vaamadeva family with the Rbhu's remains to be determined,64 but for the present purposes it is noted that the Rbhu hymns contain no instances of tanuú and tmán, and only once a reference to bráhman (cf. RV 2.33, also to the Rbhus, is devoid of these three words).

    Tanuú is quite frequently associated with humans in RV 4, moreso than in the other Family Books. Human tanuú strives to be worthy servants of Agni (4.2.14), the human tanuú is frail and in need of Indra's assistance (4.16.17, 4.16.20), or referring to the glorious presence/tanuurúca of humans when in battle (4.24.3, 4.38.7). When used with deities, Agni has the sweetest tanuú (4.10.6), USas is radiant (4.51.9), Indra places his tanuú with the sun (4.16.14). Tmán continues the same pattern and frequency of use as it had in RV 2 and 3. Tmán indicates that Agni, bearer of oblations to the gods, is a priest of his own nature (4.6.5), Indra himself leads the horses forward (4.29.4), and the atmospheric deity SavitR provides defense and sets the heavens in motions of his own accord (4.53.1, 5). Bráhman/neuter is again the predominant form, with only one of four instances in the Agni series, 4.9.4c, is in the masculine, and the same is true for Indra, only one masculine out of ten (4.24.2c). This predominance of bráhman over brahmán remains consistent throughout the Family Books, and in all cases of 2-7, occasions of bráhman with Indra outnumber those with Agni almost four-to-one on average, and six-to-one in RV 5 and 6.

    With the 5th MaNDala the number of hymns increases substantially (MaNDala's 2-4 having 43, 62, and 58 hymns respectively), to 87 hymns. Of these hymns by the AAtreya's, Agni receives 28, Indra 12, the Vishve Devaas 11, then 9 to the Maruts (the largest concentration of the family booksRV 6 contains only one partial praise in 6.48.11-12 & 20-21 and there are 4 in RV 7), 11 to Mitra-VaruNa (the "first" of the family books to address them as a pairRV 2 addresses VaruNa in 2.28, 3.59 addresses Mitra while 3.62.16-18 has a brief mention of the pair, and 6.67 addresses the pair, 6.68 addresses Indra and VaruNa), then 6 to the Ashvins, and a variety to atmospheric deities and others (USas/5.79-80; SavitR/5.81-82; Parjanya, Earth, VaruNa, Indra-Agni, and the Maruts each receive one). The distribution of bráhman/empowerment or power through prayer for both

Indra and Agni remains consistent with the previous MaNDala's, though brahmán/priest is slightly more frequent with Indra, outnumbering references in hymns to Agni by five to four.

    Tanuú is substantially less frequent with Indra and Agni compared with the previous MaNDala's (in RV 2, 3, and 4 tanuú is used with Indra and Agni more than with any of the other deities addressed in each MaNDala). There is also a greater emphasis upon the human tanuú than in the other Family Books. Both occasions of tanuú with Agni refer to the human tanuú as needing protection (5.4.9, 5.15.3). With Indra there is only one occasion which refers to the deity's beneficence to a dutiful worshiper (5.34.3). Tanuú is used much more frequently with the abstract and atmospheric deities than with Indra and Agni. The remaining uses of tanuú include 5.41.17 in a Vishve Devaa hymn to refer to the mortal's need for food in his tanuú, a plea for assistance in battle against the Dasyus (5.70.3). The Maruts are twice mentioned with well-decorated tanuú's (5.57.6, 5.60.4), and the dawn is described as having a bright and splendored tanuú (5.80.4, 5).

    The uses of tmán show no discrepancy with the patterns found in RV 2-4. TvaSTar, who is wealthy already, comes to assist in the sacrifice because it is his own nature to do so (5.5.9). Agni wanders of his own nature (5.15.4). It is the nature of the hymn itself which awakens Indra (5.10.4). Indra's thunder pervades heaven with a roar of its own nature as Indra's voice (5.25.8). This is the only attribute which is self-evident of Indra based upon uses of tmán in the Family Books. The Maruts, commonly associated with storms, are described as powerful of their own accord through an association with lightning (which, incidentally, instantaneously heats the surrounding air to 50,000 degrees F in a standard bolt from ground to air). ViSNu himself yokes the Maruts in 5.87.4.

    There are several exceptions to these patterns of composition and terminology in the Bharadvaaja hymns of the 6th MaNDala. This collection contains 12 fewer hymns than the 5th, with 16 devoted to Agni, 30 to Indra, several hymns to a variety of deities and personalities, 4 to the Vishve Devaas, the only hymns of the Family Books devoted exclusively to PRSan (6.53-58), 2 to Indra and Agni (6.59-60), 1 to Sarasvatii (6.61), 2 to the Ashvins (6.62-63), a handful to USas and Maruts, and a variety of combination hymns beginning with 6.67 to Mitra and VaruNa (others include Indra and VaruNa in 6.68, Indra and ViSNu in 6.69, etc.) as well as one to SavitR (6.71), and 6.73 to BRhaspati (1994: 244-285).

    Agni is treated quite differently considering the terminology in its semantic fields in comparison to any of the other Family Books. There are only two occasions of bráhman in the Agni hymns, both of which are found in RV 6.16 in verses 30 and 36. The other Family Books are not so disproportionate in the distribution of bráhman between Indra and Agni (though Indra normally will have the larger share). This is also the single largest collection of hymns to Indra by one poet or family in the entire RV. Additionally, the typically even distribution of tanuú between Indra and Agni in the other Family Books (with the exception of RV 5) is not the case here where we find only one occurrence of tanuú with Agni (6.11.2c) and 6 with Indra. One of the Indra hymns contains the only occasion of sháriira in the whole of the Family Books (6.25.4) along with an occasion of tanuú in the same verse referring to mortals in combat (see Chapter 4). While one exception does not make the rule, it is clear that tanuú signifies something different than a strictly corporeal body, even when it is used to refer to humans.

    Tmán occurs only twice, once with Agni (6.12.3c) and once with the Vishve Devaas (6.49.5c). Of 28 the occurrences of tmán in the Family Books, Indra claims only three (2.19.7a, 4.29.4c, and 7.18.20c), while Agni has 12. In Bharadvaaja hymns of the 6th book, the Indra hymns substantially outnumber those addressed to Agni and still there are no occasions of tmán in an Indra hymn. This further underscores the observation that Indra's self-evident characteristics are all but non-existent compared with Agni in the Family Books. This changes somewhat in the later portions of the RV. Similarly, of the remaining 14 occasions in the Family Books, the Vishve Devaas hymns have tmán 4 times, and the Maruts 5 times. The remaining occasions are scattered without any specific pattern apart from a predilection for abstract or atmospheric deities, such as the instances with VaruNa (7.84.1c), Dadhikraa (4.41.10), SavitR (4.53.1c, 5c), BrahmaNaspati (2.25.2a), and Raakaa (personification of the full-moon day, 2.32.4a).

    The distribution of the word tmán demonstrates that among the deities, Agni is the most clearly defined in his characteristics. The roles assigned to the fire in the Vedic ritual support this suggestion: he is the highest priest to the gods (4.6.5), he wanders and spreads of his own nature (3.9.5, 5.15.4c), he knows that it is his own nature to be brought forth from plants in the kindling of fire (6.12.3, cf. the discussion of Agni in water and plants in Chapter 5, viz. RV 10. 51), and it is he himself who reaches to the gods (7.7.1).


    The 7th MaNDala contains decidedly more hymns than any of the other Family Books, as well as several unique hymns like that to the frogs (7.103), and a spell against evil spirits in 7.104. Interestingly, however, Agni and Indra receive substantially fewer invocations by ratio than in any other Family Book, with only 17 and 15 respectively. Following these 32 hymns there is the interesting anomaly of the singular hymn by VasiSTha to his sons, 7.33 (in a separate browser window), which occupies the "rightful" hierarchical place of the next deity collection of 8 hymns devoted to the Vishve Devaas. This is the first of several deviations in the arrangement of MaNDala 7 that point to possible later additions and variance in the relative chronology suggested here.

    Notoriously "obscure" (Roth, 1846; Muir, 1967: 319-321), 7.33 warrants further inspection.65 Witzel (1995b: 311) lists it with some 60 other hymns or portions of it that violate order of arrangement as noted also by Oldenberg (1988) and others (Arnold, 1897: 212, Lanman, 1880: 581) For our purposes, it is simply worth noting that a possible reason for this aberration in order might well be reflected in the rivalry with the Vishvaamitras (who also place the Vishve Devaas as third in their "hierarchy" of deity collections). The instance of this hymn, and the rivalry to which it refers, offers an additional point of analysis regarding the family books and the nature of the self. The neuter bráhman is used 3 times (7.33.3, 4, 11) attesting to the VasiSTha's ability to empower Indra against his enemies, repair the axle of his chariot, and in placing VasiSTha himself upon a sacrificial pitcher. Indra is most frequently spoken of as responsive to the VasiSTha's words (7.33.2, 4, and 5).


    The rivalry between the VasiSTha's and Vishvaamitra's affords several comparisons of how each family uses the terminology for the self. This is especially apparent when the uses of tanuú are compared between both families. The use of tanuú between both families is almost equal which is surprising considering that RV 7 has nearly twice as many total hymns. There are 16 occasions of tanuú in RV 3 as opposed to 17 in RV 7. The use of tanuú in RV 3 is applied equally--and only--in Agni and Indra hymns (8 times each). For RV 7, tanuú is used only 2 times with Agni to designate his resplendent presence/tanuú (7.3.9), that he increases (7.8.5); and twice with Indra regarding how he assists Kutsa in battle (see discussion of Kutsa below in Chapter 5, cf. also note 92 above) by means of his tanuú (7.19.20), and the heros invoke Indra to the aid of their tanuú (7.30.2). The use of tanuú elsewhere in the 7th MaNDala differs from the Vishvaamitra's in that the tanuú is decidedly

less corporeal. Even the uses here with Agni and Indra suggest a less corporeal sense of tanuú than the occasions with both deities in RV 3. In the Vishvaamitra hymns tanuú was used only with Indra and Agni whereas the VasiSTha's use tanuú predominantly with other deities.

    In the VasiSTha hymns as a whole, tanuú is more abstract by means of the tanuú humans seek to unite with VaruNa in 7.86.2 (cf. Reat's suggestion of the tanuú as a point of connection between divine and human realms [1990: 38f., 63ff.]), ViSNu's is measureless (7.99.1), the evil spirits/rákSa use the tanuú as a veil to hide their evil (7.104.10-11), and Parjanya's tanuú has a variety of manifestations according to his will (7.101.3). In RV 7 there is a greater number of "new" ideas signified, among other things, by the first occurrences of aatmán (7.101, 7.87), and most of the occasions with púruSa of the Family Books (7.4.3, 7.29.4, 7.57.4, 7.75.8, 7.102.2), the use of tanuú with deities other than Indra and Agni that is more abstract in its significations, the wry hymn to the Frogs (7.103) which, as noted above, echoes RV 3.53.7, and is itself a later addition. These statistics are of interest especially regarding the observation later in this study that the decrease in usage of tanuú corresponds to the increase of aatmán and púruSa, both of which appear for the first time in this family book. However, púruSa invariably denotes a weak and vulnerable mortal or group of mortals, who are vulnerable to sin. The only clear exception is found in 7.102.2 where Parjanya is the germ of life in all living things, including women/puruSiíNam.

    The unique nature of the VasiSTha hymns is identifiable with other terms as well. Tmán is relatively infrequent in RV 7 by ratio in comparison with other Family Books, excepting RV 6, which had only two occasions, both of which were in Agni hymns. In the 7th MaNDala there are only five hymns containing the term, with only one each to Agni (7.7.1) and Indra (7.18.20). One occasion, 7.84.1c occurs in a hymn to VaruNa, a deity conspicuously overlooked in the Vishvaamitra collection, perhaps owing to his role as the progenitor of VasiSTha. While RV 3 has four hymns to the Vishve Devaas, tmán is not found in any series but twice with Agni, and the other predominant occasion for tmán--based upon the pattern observed above--would be with the Maruts which are not addressed at all by the Vishvaamitras.

    The remainder of MaNDala 7 is marked by deviations from the basic structure of order maintained consistently in the first five Family Books. Several more aberrations appear in the deity/number of hymns criterion

for sequencing the collection. The eight Vishve Devaas hymns that follow the self-congratulatory focus of 7.33 precede several hymns of miscellaneous nature (SavitR in 7.45, Rudra in 7.46, a handful to water and the Rbhus, 2 to the AAditya's, one to Heaven and Earth, and two to VastoSpati), apparently resuming the proper ordering according to number of hymns addressed to each deity. However, another series of incorrectly ordered hymns are next, beginning with 4 hymns to the Maruts, 7 to Mitra and VaruNa (again breaking "normal" order), eight to the Ashvins, and 7 to USas. After this disruption, the normal criteria again emerges with four hymns addressed to Indra and VaruNa, and 4 to Vaayu and Indra-Vaayu. These "combination hymns"-- or, as Gonda terms them, dual-deity hymns (1974)--also include two to Indra and Vaayu as distinct entities rather than "dual deities." The remainder of RV 7 is quite diverse: 2 hymns to Sarasvatii, a handful to Indra, BRhaspati, and ViSNu, 2 to Parjanya, and one each for the frogs and the charm against evil in 7.104.

    There arises a clear distinction in the chosen emphasis upon various deities from family to family that is not limited to the squabbling of the VasiSTha and Vishvaamitra clans. It appears, then, that careful observation of the structure of each Family Book yields patterns reflected in the usage, by each clan, of the lexicon under examination in this study. Tmán shows a prevalence of occurrences in MaNDala's that emphasize Agni, the Maruts, and the Vishve Devaas. The frequency of its use with Agni attests to the clearly-defined characteristics attributed to the deity. Tanuú is most predominant in the smaller collections of MaNDala's 2-4, and decreases in frequency in the larger collections, with significantly fewer occasions by relative frequency in the largest of the Family Books, RV 7. Indra hymns have the greatest frequency of the uses of tanuú (excepting RV 7). This correlates with the predominance of occurrences of bráhman with Indra which serve to call forth a strength and power from him that is not considered intrinsic to his nature (i.e., tmán is not used to designate it).

Composition and Arrangement of the Later Rg Veda


    Following MaNDala's 2-7, the homogeneity of each MaNDala is much less clear. There are multiple subdivisions representing a multitude of poets, none of whom are credited with more than 20 hymns. There are four identifiable periods in the later RV. The first identifiable segment is 1.51-191 and 8.1-66--with the exception of the Vaalakhilya hymns 8.49-59 (Gonda, 1975: 11; Witzel, 1995b: 309). For convenience, this period is abbreviated as RV Late-a in the following chapters. The variety of R'Sis in

cluded and the relatively small number of hymns attributed to each makes the distinctions of deity and so forth too isolated to be of great use, though it is still maintained (e.g., 1.58-1.64, attributed to Gautama Nodhas, follows the order of 3 to Agni, 3 to Indra, and 1 to the Maruts). RV Late-a contains eight occasions of aatmán and a wide range of uses for tanuú. The ratio of hymns where tanuú refers to human frailty as opposed to the manifestations of divine presence is still consistent with the Family Books. As aatmán appears more frequently, tanuú takes on decidedly more physical characteristics. In 1.162.20 the tanuú is vulnerable to the axe in the horse sacrifice.

    In the uses of tmán we see Indra having more characteristics of his own nature that are consistent with his exploits in the mythology of the Family Books. Indra himself gives vigor (1.63.8), and he himself strikes down the demon Shambara (1.54.4). The use of tmán that was primarily seen with Agni, the Vishve Devaas, and Maruts in the earlier MaNDala's disintegrates into a largely unsystematic usage. Agastya--to whom is credited 1.170-184, eight of which are to Indra--shows a substantial predilection for the word, using it twice with the Ashvins (1.183, 184), with Heaven and Earth (1.185.1c), in an AAprii hymn (propitiation or invitation hymns to various deities whose presence is needed at the sacrifice, here in 1.188.10a), and once to Indra (1.178.3). Words for the self are comparatively scarce in this early portion of the 8th MaNDala (click here to review the 8th MaNDala, in a separate browser window-- scrolling through the book ismost recommmended), but 8.3 does include both aatmán and tmán. There is also a certain predilection for shatákratu/hundred willed in the Indra hymn 8.36 (six occasions). Indra receives the predominant uses of tmán, though in this portion there are only four occasions to be considered (8.3.21 and 8.6.8 are to Indra, 8.4.3 is to Agni, and 8.46.27 is to Vaayu).

    The next sequence, RV Late-b, includes the first MaNDala, with RV 1-50 and 8.67-100 (Witzel: 1995b, 310). There is no marked increase or decrease in uses of words for the self. Tanuú remains consistent in relative proportions to both aatmán and tmán. There is a reasonable scattering of bráhman, but with no particular patterns of association with other words related to the self and, in 1.18, many occasions of BrahmaNaspati (cf. Chapter 4 below). There are also many combinations of various terms with krátu. No real pattern emerges considering there are only two occasions of tmán, 1.30.14a to Indra and 1.41.6a to the AAdityaas. It is significant, however, that where the Family Books had isolated occasions of púruSa--it appeared in several times in RV 7, also in 3.33.8 to Indra, 4.12.4 to Agni,

and 5.48.5 to the Vishve Devaas--both RV Late-a and RV Late-b contain not one occasion of púruSa with the exception of 8.71.2 in RV Late-b to refer to Agni. In addition, there is one occasion of aatmán in RV Late-b, in 1.34.7 to the Ashvins. There seems to be a very limited concern with individual existence in RV Late-b, as there are also only eight occasions of tanuú and five occasions of tmán. No instances of púruSa are found in RV Late-b.

    As to the arrangement of the hymns in RV Late-b, we do not find a systematic order based upon deity, though frequently Agni or Indra are presented first in a R'Si 's collection. However, variations and interspersed exceptions throughout their "collection" obviate this as a reliable criteria of arrangement. Immediately following RV Late-b are the Vaalakhilya hymns, 8.49-59, in which there is one occasion of tmán to Agni (8.49.4) and one of tanuú (8.56.6). There are no occasions of aatmán or púruSa in the Vaalakhilya's and just two of bráhman (8.52.9, 8.53.8). Winternitz suggests they are of comparable date with the other portions--at least of RV 8--but that there seems no clear reason for their insertion in the course of redaction (1959: 60). Not all these 11 hymns are recognized--the BRhaddevataa 6.84, for instance, recognizes only 8.49-56 (not by those numbers). Gonda suggests they were considered "inferior and half-apocryphal" (1975: 37). Haug draws attention to the complexity of their recitation that was deemed a high level of achievement for the HotR (1977: 416).

    The last two segments of the Early Vedic period are, respectively, the 9th and 10th MaNDala's. Especially in the case of the 9th, however, as it is a collection of the hymns for the Soma ritual drawn from the other books, much of the material is quite old. Similarly, not all hymns of 10 are, by default, considered younger, whether due to their style or content (Witzel, 1995b: 310; Gonda, 1975: 11). RV 9's antiquity, by way of its relation to the Saama Veda (1/3 of the SV hymns are found in RV 9), has been suggested by Weber (1892: 8, 32) to stem from the absence of hymns from RV 10 in the SV (as compared with the large number of RV 9 hymns). Keith affirms the SV as prior to the TS (1967: 164). These are fairly broad inferences and suggest, if nothing else, that further study of the SV is warranted--and, by connection, study of RV 9 appears to be wanting as the familiar quote (Weber, Winternitz, Gonda) of "75 [or so]" verses additional to RV 9 are found in the SV is almost all that is said--Weber suggests 78 (1892: 65); Winternitz counts 75 (1972: 164); Gonda suggests 76 (1975: 314). Winternitz notes that the 75 not found in the RV (the other 1549

hymns of the SV, not counting repetitions come mostly from RV 8 and 9) can be variously found in disconnected segments of the other SaMhitaas. There are only three occasions of tmán, and there is not enough significant correlation between them to place them in the wider patterns noted in this section of the study. AAtmán is found several times in reference to Soma as the essence or vital center of the sacrifice (9.2.10, 9.6.8), and to describe Soma as Indra's primary food (9.85.3). Butter and milk are drawn from it in 9.74.4. There are no occasions of púruSa in RV 9.

    Witzel refers to RV 10 as "the great appendix" to the collection (1995b: 310). Stylistic issues, such the equivalence of 191 hymns in both RV 1 and 10, the use of agnim iiLe ("Agni I call upon") to commence 10.20-26 as found in RV 1.1.1, and so forth suggest that its authors were already conversant with the older segments of the text. Like RV 9, the 10th MaNDala contains hymns of an antiquity similar to those of the Family Books, especially that of the VasiSTha and Vishvaamitra collections (Gonda, 1975: 12-13). The book has three broad divisions, 1-60 with thirteen family groups, 61-84 is a series of pairs devoted to the same deity, and then hymns 85-191 representing the most recent and eclectic part of the collection (Gonda, 1975: 12). Another feature of this MaNDala is its close relation with the AV, excepting the final of its 20 books, but with variations (Gonda, 1975: 13; Arnold, 1967: 22f). When considering the one-seventh of the AV material coming from the RV, Winternitz notes that one half of that portion is derived from RV 10 (1972: 121; also Weber, 1892: 146).

    It is not surprising, then, that this late collection in RV 10, taken as a whole, yields frequent occasions of the terms of later predominance with regard to the self, aatmán and púruSa. In particular, it is only in RV 10 that púruSa signifies something abstract--as in RV 10.90, the PuruSa Suukta--in contrast to the uses in the Family Books where it refers to a sin-prone, vulnerable mortal (or mortals). There are hymns such as 10.97 to the herbs that present a study in and of themselves in the terminology's evolution as we find aatmán, púruSa, krátu, and tanuú in a hymn cited (Brown, 1931: 108ff.), among other things, as an origin of the later associations of Agni and púruSa which became vital to the mysteries of the Agnicaayana such as it is discussed in ShBM 10. There are, in general, frequent occasions--more than in any other MaNDala or, proportionally, temporal strata--where tanuú is used in close conjunction with aatmán (e.g., 10.16), tmán (10.110), and púruSa (10.51, 10.90) affording a closer point of discrimination in the developing signification of all four terms just at the

point where the former begins to decrease in frequency as the Middle Vedic period approaches. Regarding tmán, the apparent indication of later usage shown in RV 1 and 8--earlier and later portions--signified by association with Indra is consistent. Of 11 occasions, four are associated with Indra, two with Agni, and the remaining five scattered among various deities including BRhaspati (10.68.7b), Suurya (10.170.1b), an AAprii hymn (10.110.1a), and two hymns similar to the Family Strata pattern: 10.64.6b to the Vishve Devaas and 10.77.3a to the Maruts.

    The later portions of the RV are marked by several changes in the terminology that occur largely without specific association to one poet's hymns over those of another. The most significant change is the more corporeal meaning which begins to attach to tanuú once aatmán appears to signify the central or vital essence of an individual's presence/tanuú. In addition, the predominant association of tmán with Agni as the most defined deity in terms of its self-generated characteristics (as opposed to those called forth by prayers) breaks down and Indra comes to have his own characteristics. Most significant is the appearance of púruSa--after a complete absence in RV Late-a and only one occurrence in RV Late-b--as a being whose sacrifice creates the entire cosmos.

    Returning to Witzel's work with the geographical, archeological and linguistic evidence of the Vedic period, a general period during which the RV Family Books were composed would be close to 1250 B.C.E. (1989: 249). According to Witzel, the collection of these hymns occurred closer to 1180 B.C.E. when the "appendices" or RV Khila's were composed and collected. This latter group of materials is more properly classified as part of the earliest segment of Middle Vedic, the Mantra Language. Many of the Khila's are a few lines attached to the beginning or end of later RV mantra's (with a few exceptions, RV Kh. 2.5 is attached to RV 5.85; RVKh. 2.11 is attached to RV 6.1; RV Kh 2.13 is attached to RV 7.35; and RVKh 2.12.1 is attached to RV 6.45). Most of the Khila's, including those found in RV 8.49-59, 'stand alone' without attachment to any specific hymn (1946: 915f.). In this study I am including the survey of their contents in Chapter 5 with the later portions of the RV instead of in Chapter 6 that addresses the Mantra Language of Middle Vedic to which the KhilaaNi properly belong.

    The terminology for the self is widely scattered through the Khila's. Terms for the body such as déha and sháriira are as scarce as elsewhere in the RV with only four occasions of sháriira (RVKh 2.12.1, 4.8.5, and 4.11.7, 4.11.8), and one of déha (2.12.1). It is also interesting that púruSa

is significantly more common than aatmán, outnumbering it by 13 to 6. PúruSa continues to be a general term for mortals in the Khila's, (1.4.6, 2.6.2, 2.6.15, 2.9.1), but we do not see the connotations of the púruSa as sin-prone and wicked as with the occurrences in the RV other than, for instance, RV 10.90. Tanuú occurs 6 times displaying a range of meanings such as the many presences of the gods in RVKh 3.11.2 and 5.6.6. There are also occasions where tanuú is something that can be united with the gods as seen in RV 7.7.1 between VaruNa and tanuú, such as RVKh. 1.8.1. Thus the RV Khilas will provide a useful point of analysis for the transition from the RV to the Middle Vedic literature which has not been considered in any other study of the terminology related to the self.

    The importance of the results in this study for the earliest origins of the notion of the self in Vedic India, which are diagrammed in detail in the following chapters, cannot be adequately assessed without careful consideration of the findings in the literature that immediately follows Early Vedic. In this way the results of the research presented in Chapter 4, the Family Books, and Chapter 5, the later RV, can be placed in context with the changes and continuities in the next major period of Vedic literature that includes the Black and White Yajur Vedas, the Atharva Veda, and the early BraahmaNa's. This period is collectively called Middle Vedic (Mylius, 1979: 423; Witzel, 1995a: 97). The portions of Middle Vedic to be examined in Chapter 6 are divided into the Mantra or metric Language of the early Black Yajur Veda and Atharva Veda, the SaMhitaa Prose of these same texts, and the BraahmaNa Prose of the Shatapatha BraahmaNa, the Jaiminiiya BraahmaNa, and the Aitareya BraahmaNa (cf. Narten, 1968: 115; Witzel, 1995a: 96-97).

Middle Vedic


    In this period of the Vedic literature the development, specialization, and even abandonment of various terms becomes clearly apparent. The analysis of terminology is confined to an assessment of the findings in Chapter 4 and 5 with the terminology for the self and how the uses of each change in this period. In addition, there is a primary emphasis upon two main groups of Middle Vedic texts. The material of the Black Yajur Veda (BYV), particularly the MaitraayaNi SaMhitaa (MS), the KaaThaka SaMhitaa (KS), and the Taittiriiya SaMhitaa (TS), receives greatest attention as it was completely overlooked throughout the previous studies of the self in Vedic literature (cf. Chapter 1). In fact, there are few studies of these texts--es-

pecially the untranslated MS and KS--so this dissertation will be charting important new territory by exploring how these SaMhitaa's treated such key terminology. The second primary emphasis is upon the Jaiminiiya BraahmaNa (JB) and the Shatapatha BraahmaNa (ShB) for exactly the opposite reason. Both texts have received substantially more attention by scholars than the BYV texts (e.g., Bodewitz, 1973, 1976, 1990; Gonda, 1986, 1988; Smith, 1989; Thite, 1975; Tull, 1989; etc.). Another reason for attending to these two BraahmaNas, apart from their prominent place among recent publications, is the availability of electronic editions of each.

    By examining these two groups of texts I will also be addressing the sub-categories of Middle Vedic identified as Mantra Language (metric and prose mantras from the BYV), SaMhitaa Prose (non-metric expository passages that explain the rituals that are also included with Mantra Language sections of these texts, but represent a later period of composition), and BraahmaNa Prose (the material of the ShB and JB). I will also examine the Atharva Veda, PañcavMsha BraahmaNa, Taittiriiya BraahmaNa, Aitareya BraahmaNa, the AAraNyaka's, and UpaniSads whenever they quote--or are quoted in--these texts of primary emphasis. Accordingly, the following sections will summarize the existing research on the relative chronology of each period of texts in Middle Vedic to provide a timeline for comparing the changing uses of the terminology for the self.

Mantra Language


    Defined as the mantras of the RV Khila, the SV, ShYV and KYV, as well as the mantras in verse and prose of the AV (Witzel, 1995a: 96; 1989: 124), the majority of non-electronic texts studied here fall within this category. Terminological analysis will be carried out in these texts with the assistance of the VaidkapadaanukramakoSaH.66

    As Witzel suggests, Mantra Language includes the oldest Indian prose along with the metric portions, but it requires further study (1995a: 96). In this examination of the developing terminology for the self, changes in use and frequency of various elements of the early-Middle Vedic discussions that employ the words for the self will add to the understanding of stratification within this chronology. Witzel suggests that the period of phonological revision, or orthoepic diaskeuasis, during which a "free-floating" mantra collection underwent gradual linguistic change (1989: 124-5), formed the palette from which these texts drew their ritual invocations.

    Addressing the same issue, Keith posits an ur-YV [my term] from which the MS, KS, and TS derived their mantras. In this case, the TS would

reflect a slightly later or, at least, different ("derivate") collection from the ur-YV regarding its more consistent harmony with the VS and the RV (1914: 90). By contrast, the KaaThaka and MaitraayaNi SaMhitaas "mutilate" RV passages. Keith's discussion includes tentative agreement with Oldenberg's suggestion (1888: 304, n. 1) that the TS and MS together derive from an intermediate sub-collection of an ur-YV that was, in turn, a direct source for the VS as it was derived with stronger RV influence (1914:89).

    Unfortunately, accounts beyond these broad speculations, as Witzel observes, are largely missing from the pool of available scholarship on the subject. The paucity of research in this area is compounded by the view held by Keith that geographical data can be of little use (1914:92). When examined according to the relative chronology, changes in semantic field terms can reveal changing nuances of the key words which are traceable to geographic locations and corresponding historical peridos of settlement.

    To provide a working paradigm of the chronological stratification in the Mantra Language, I offer the following hypothetical framework of scholarship: if we were to accept Keith's rejection of L. L. Von Schroeder's suggestion that PaaNini's silence on certain forms of the MS therefore implies the older date of the KS, then the three BYV SaMhitaa's are of relatively equal date in their derivations of mantras from the ur-YV (1914: 96-7). Next, relying on the ratio of imperfects to perfects, a criteria first identified by Whitney (MS: 22.37 to 35, TS: 19 to 27, and KS "similar"), Keith feels that comparative equivalence in date is a fair conclusion (1914:97; also Gonda, 1975: 327).

    This quantitative similarity alone is not convincing, for two reasons: in the first case, Keith rejects geographical variants that may or may not have bearing on relative chronology--cf. regional variants tracked by Witzel--such that, in the second case, a combination of regional/dialect evidence when combined with philological evidence adduced in the current study allows for drawing finer distinctions. I will provide an example here of the level of detail involved in drawing such chronological stratifications, based on linguistic evidence. Much of this material, as it pertains to the study of the self, is of a more limited applicability. Where it is possible to regularly apply it, I have outlined the linguistic criteria that will be used in

the previous chapter.

    On the point of TS/KS/MS sequence, for instance, Witzel identifies the priority of the MS to the KS. Next he adds that there is an additional contrast of the MS and KS with the TS as can be seen with genitive feminine singular ending in -ai. This feminine singular ending is a later Taittiriiya development that differs from the earlier form with -yaaH as, for instance, in TS prásityai compared with KS 1.12 prashityaaH (1989: 135). Keith's examples, as Witzel notes, do not come from the Mantra portions of the SaMhitaa, a distinction of genre recognized by Keith, who even divides his sub-headings according to these categories (1914: 72f, 98f, 159ff.), but does not consistently observe them when drawing his conclusions (Witzel, 1989: 135, n. 80).

    Distinctions such as ai/-yaaH are not fixed by rule, as the occurrence of ai is often sprinkled in contradistinction to what would be expected with accepted chronology (e.g., the accepted posteriority of AB 6-8 does not reflect what would be expected and -ai is actually found in AB 1.27 and 4.27). Suffice it to say that in gaps and contradictions of Mantra Language and SaMhitaa Prose chronology, deference to Witzel's study will be the rule, with emendations as needed from the collation of data from the current study.

    Based on the evidence above, the general sequence of Mantra portions in the YV schools is accepted here as MS, KS, TS, and VS respectively. With regard to the VS, it is true the VSM uses ai, but this has been noted as more a case of redaction, while VSK aligns with the Western dialects and is later (1989: 137; also Gonda, 1975: 331), vs. Weber (1892: 90-1). In the chronological period immediately prior to the YV collection, the Mantra Language of the Atharva Veda Paippalaada recension precedes the Sama Veda, and will be treated in that order (Witzel, 1989: 250; Gonda, 1975: 274; Keith, 1914: 163-4). The RV Khilas are situated, as noted above, just after the RV and the AVP, but prior to the MS and KS. These texts fall within and shortly after 1180 B.C.E., the date mentioned for the RVKh above (Witzel, 1989: 250).

    The next consideration which must be distinguished is the relation of the Shaunaka and the Paippalaada recensions of the Atharva Veda, which Keith does not treat. In addition, following Weber (1892: 65), the antecedence of the SV to RV 10 is open to debate in light of, at least, the words currently under study since, for instance, aatmán is relatively more frequent in RV 9 than 10, while tanuú, the "earlier" term of choice is more

plentiful by proportion in the latter. The reliance of the SV upon RV 8 and 9 for much of its content draws Weber's conclusion into question.

SaMhitaa Prose


    Although the full study of SaMhitaa Prose is still in its infancy, the category SaMhitaa Prosea YV phenomenonwas nonetheless recognized as a distinct category from Mantra Language in earlier studies such as that of Keith (1914: 159f., albeit his categories were "Mantra portions" and "BraahmaNa portions"). Not to be confused with the language of the BraahmaNas to be discussed later, Witzel identifies this narrow strata as consisting of BraahamaNa-type expositions of the mantras in the previous category (1995a: 97).

    Keith suggests that the older AB 1-5 predates the prose portions of the TS, while the later 3 Pañcika's are not recognized (1914: 99). In addition, these portions are reckoned to be earlier than the ShB and JB. TS 3 and 7 are suggested as altogether later, with the PB (22.11, 14-16; 21.1, 9.2, 10.5-10) as the probable source for the latter with respect to the Ahiinas (1914:69, 72, 100-1). The TS is held by Weber (1892: 108) as equal in time of composition with VS 1-18 and some of 22-25 without distinctions of mantra/prose. All the TS prose, as Keith suggests, was known to Yaaska and PaaNini (Keith 1914:167-9). Witzel ascribes the YV prose of the MS and KS as prior to (c. 900 bce.) to the TS prose (1989: 250).

BraahmaNa Prose


    Witzel notes that this literary period should be divided into an earlier and later portion (1995a: 97), which I will abbreviate for convenience here as BPa and BPb. Again, a great deal of research could further these delineations, but I will attend primarily to the AB, ShBM/K, JB, TB, and PB that are the primary texts of interest for this portion of the Middle Vedic study. The AAraNyaka's, as part of the BraahmaNas, are treated as components of this period, with sequential distinctions as necessary or identifiable from available research.

    The chronology of the BraahmaNas presents a substantial challenge to even the most diligent scholar (Gonda abandons the task, after an uncharacteristically brief four-page foray, as laden with difficulties that "prove to be enormous" 1975: 360). Part of the problem arises from the fact that these texts are composed in several different segments and their chronological order is not reflected in the order of their arrangement. The safest course is to begin with the most widely agreed upon facts--both of them--and move from there instantly into the melee of Jaiminiiya jousting and

Shatapatha shuffling and the other minutiae of subdivisions in BraahmaNa prose chronology. There is general assent that AB 1-5 represents the earliest of the BraahmaNa portions (Keith, 1914: 99, 100; by implication 1981: 42, 46-7; Gonda--allowing possible precedence to the PB--1975: 357; Witzel, 1989: 250), though, for instance, Wackernagel (1896: xxx) dissents, preferring by linguistic argument the PB, JB, and TB in that order. Immediately the more fiendish of the features of this dating arises as there are a multitude of temporal layers in these texts, enough to make the deliberations above on Mantra Language appear not unlike a frolic.

    Witzel (1989) sorts these matters with sufficient clarity to serve the needs of the current study for a relative framework upon which to build the detailed analysis in the following chapters. Following AB 1-5, the TB is fairly well accepted as the next oldest and, where dissent over precedence between the TB and AB 1-5 arises, the consensus of scholars nonetheless favors it as next in sequence (Gonda, 1975: 357; Keith, 1914: 97; Witzel, 1989: 250-51). Next there is the tangle over the sequence between the JB and PB. Bodewitz (1990: 15f.) has taken the principle of subdividing the text to an altogether higher level of detail by comparing the JB with PB 6-9 as follows: JB 1.66-191, 206-228 (232), and 3.42-364 in that sequence, with JB 1.1-65, 192-205, and 233-341 as later additions. Caland, defers to linguistic arguments to place the entire PB as younger than the JB (1931: xix-xx). Gonda notes that Wackernagel holds the PB as older, and concludes himself that the JB and ShB are both later (1975: 357-8), where Keith says both the AB and KB predate the PB (1981: 42).

    Keith is closest to the mark, placing both the AB and KB prior to the PB, with the KB younger as a whole ((1981: 24), and AB 7-8 following the KB (1981: 47). Caland allows priority of the JB to the early portions of ShBK and ShBM. On the ShBM/K issue, Gonda punts entirely--as he does also with JB/PB--declining to identify one or the other redaction as older (1975: 354, 348). Most authors agree, however, that three distinct strata may be identified in both texts, most easily described in ShBM as 1-5, 6-9, 10-14. Weber broadly identifies ShBM 1-9 as older, with 10-14 successively younger (1892: 108, 119-20). Keith specifies ShBM 6-9, the ShaaNDilya books, as oldest, followed by 1-5, the Yaajñavalkya books, and finally the last five, 10-14 (1967: 102-3). Witzel's sequence is the same. In addition, he offers the following broad resolution, to the conundrum: AB 1-5; TB; then the KB. This group will be called BraahmaNa Prose-a (BPa). More divisible by regional variation than clearly defined temporal progression, and fitting the

designation of BPb, are the JB, ShBM 6-10, ShBM 1-5, ShBK 1-7, and somewhat subsequently, AB 7-8 follow as a largely overlapping group at roughly ~500 bce (1989: 250-51). The last strata under consideration here is composed of the Chandogya and Taittiriiya UpaniSads, and the Pañcavimsha BraahmaNa.


    This chapter has established the temporal framework on which analysis of the lexicon of words related to the self will be based. An overview presents the following broad sequences: for Early Vedic, RV 2, 4, 5, and 6 are the oldest, with MaNDalas 7 and 3 slightly younger, respectively. Following the results for BraahmaNa prose, it is also likely that RV MaNDalas 5 and 6 could simply represent regional variations rather than temporal ones. Consequently they are included as the oldest of Early Vedic, though under scrutiny. RV 3 and 7 reflect a unique set of contrasts and similarities that suggest a possible later date for one or the other MaNDala. This portion of the RV is followed both in time and in sequence--as Chapter 5--by RV 8.1-66 (excepting 8.49-59) and 1.51-191 as one unit, and 8.67-103 and 1.1-50 as a second, subsequent unit. Just prior to RV 9 are the Vaalakhilya hymns 8.49-59, then RV 9 and RV 10. It is not possible at this time to establish dates for the subsections of the RV.

    For Middle Vedic, the Mantra language of the Atharva Veda will be considered the oldest, with the recension of Paippalaada as prior to the Shaunika. This is followed immediately by the SV and the RV Khilas. These latter two mantra texts are likely earlier than the Shaunika which, in turn, will be scrutinized as later than the MS, KS, and TS Mantra language, while remaining earlier than that of the VSM and VSK. The SaMhitaa prose of the MS, followed by that of the KS and TS will be the accepted order, with careful attention to the relation to the later portions of the RV as noted with 10.121. BraahmaNa prose has three broad divisions of a, b, and c. BPa follows a clear sequence of AB 1-5, TB and KB. Following these RV and BYV BraahmaNas is BPb that consists of internal divisions possibly attributable in some cases to region rather than to sequence. In this category-BPb--I am placing, in chronological order, the JB, ShBM 6-10, ShBM 1-5, ShBK 1-7, AB 7-8, and ShBM 11-14. These texts are briefly touched upon with regard to the earliest appearances of the relation between bráhman and speech as well as aatmán and bráhman. Finally in BPc where temporal distinctions remain in need of further structure there is the Chandogya UpaniSad, the Taittiriiya UpaniSad, and the Pañcavimsha BraahmaNa. In a broader sense, however, the sequence of BPa, b, and c as temporal categories will be used for comparing the results of each synchronic analysis of terminology for the self.